2QUE ESSENTIALS ADD "COMPLETED CHAPTER X" WHEN DONE A CHAPTER. OTHERWISE NOT COMPLETED. ******************************************************************** CHAPTER 20. STANDARDS AND TERMINOLOGY ********************************************************************* DEFINITION OF A NETWORK A NETWORK IS A GROUP OF CONNECTED COMPUTERS THAT SHARE INFO+RESOURCES. CONNECTED COMPUTERS SHARING RESOURCES IS "NETWORKING" THE TRANSMISSION MEDIUM IS THE PATHWAY. WHEN A COMPUTER IS NOT PART OF A NETWORK, IT IS A "STANDALONE" NETWORK TOPOLOGIES 1.BUS 2.STAR 3.RING 1.BUS TOPOLOGY BUS TOPOLOGY IS PASSIVE. THE SIGNAL IS NEVER REGENERATED. EACH DEVICE ON THE BUS ONLY MONITORS EVERY PACKET. BUS IS SIMPLEST NETWORK and LEAST EXPENSIVE TO INSTALL. EACH END OF THE BUS NETWORK MUST BE PROPERLY TERMINATED. THERE CAN BE NO BREAKS IN THE CABLE, OTHERWISE AN UNTERMINATED CABLE EXISTS. 10BASE-2 IS THE MOST COMMON BUS TOPOLOGY. 10BASE-2 IS AKA THINNET. 10BASE-2 USES RG-58, 50 OHM IMPEDANCE. MAXIMUM DISTANCE OF A SINGLE 10BASE-2 CABLE IS 185 METERS (607 FEET) 10BASE-5 IS THICKNET 10BASE-5 USES RG6 IS THICK AND A MORE DIFFICULT CABLE TO INSTALL 2.STAR TOPOLOGY COMMON STAR NETWORKS ARE CAT 3, 10MBITS. NEW STAR NETWORKS ARE SHOULD AND USUALLY ARE CAT 5, 100MBITS OR BETTER STAR-BUSES ARE WHEN STAR NETWORKS ARE CONNECTED TOGETHER IN A BUS. EACH STAR NETWORK BECOMES A NODE ON THE STAR BUS. 3.RING TOPOLOGY TOKERING IS ACTIVE. THE SIGNAL IS REGENERATED BY EACH NODE IN THE RING NETWORK ARBITRATION TO GET ACCESS TO ANY NETWORK, THERE MUST BE SOME METHOD FOR THE COMPUTER S TO ACCESS THE NETWORK. THIS IS CALLED NETWORK ARBITRATION NETWORK ARBITRATION, THREE COMMON NETWORK ARBITRATION METHODS ARE 1.CSMA/CD (COLLISION DETECTION) 2.CSMA/CA (COLLISION AVOIDANCE) 3.TOKEN PASSING 1.CSMA/CD (ETHERNET) ETHERNET USES CSMA/CD. WHEN AN ETHERNET DEVICE WANTS TO TRANSMIT. 1.DEVICE SENSES IF THE CABLE IS FREE. IF FREE USE IT USES THE NETWORK. IF NOT, ETHERNET DEVICE MUST WAIT UNTIL NETWORK IS FREE. 2.WHEN NETWORK IS FREE, DEVICE IS ABLE TO TRANSMIT. 3.IF 2 OR MORE DEVICES TRANSMIT AT THE SAME TIME, A "COLLISION" OCCURS. EACH DEVICE WAITS A RANDOM AMOUNT OF TIME BEFORE TRYING AGAIN. AS MORE DEVICES ARE ADDED TO THE SEGMENT, MORE COLLISIONS OCCUR. BREAK LARGE SEGMENTS INTO SMALLER ONES TO REDUCE COLLISIONS. THAT IS, MAKE BROADCAST+COLLISION DOMAINS SMALLER. 2.CSMA/CA (LOCALTALK) CSMA/CA IS USED BY APPLE LOCALTALK. CSMA/CA IT IS NOT OFTEN USED. 3.TOKEN PASSING (IBM TOKENRING) THERE IS NO COLLISIONS IN TOKENRING. EVERY TOKENRING DEVICE HAS EQUAL ACCESS TO THE RING. WANS ARE USUALLY CONNECTED VIA 1.PSTN/POTS 2.ISDN 3.T1 4.MODEMS THE TCPIP SUITE OF PROTOCOLS INCLUDES 1.IP (MAIN ONE) 2.TCP (RELIABLE DATA TRANSPORT) 3.FTP (FILE TRANSFERS) 4.SNMP(NETWORK MANAGEMENT) 5.NFS (OPEN NETWORK FILE SYSTEM) 6.ARP (ADDRESS RESOLUTION MAC TO IP) 7.TELNET (REMOTE CONNECTIONS) 8.UDP (USER DATAGRAM TRANSPORT) THE NOVELL SUITE OF PROTOCOLS IS IPX/SPX (INTERNETWORK PACKET EXCHANGE/SEQUENCE PACKET EXCHANGE) NETBEUI= IS AN EXTENSION OF NETBIOS. NOT ROUTEABLE. DESIGNED BY IBM IN 1980. PRACTICE PROBLEMS STANDARDS AND TERMINOLOGY 1.CSMA STANDS FOR? A:COLLISION SENSE MULTIPLE ACCESS 2.WHICH NETWORK ARBITRATION METHOD ONLY TRIES TO DETECT COLLISIONS INSTEAD OF TRYING TO AVOID THEM? A:CSMA/CD (ETHERNET) DOESN'T MIND THE CRASHES! 3.WHAT CAN BE CONSIDERED A MEDIA? A:TWISTED PAIR,TELEVISION,RADIO WAVES,MICROWAVE SIGNALS,DISK DRIVES, TAPE BACKUPS, VHS TAPES, ALMOST ANYTHING CAN BE CONSIDERED A MEDIA. 4.WHO INVENTED ARCNET? A:DATAPOINT 5.TOKENRING USES CSMA/CA? A: NO. TOKENRING DOES NOT USE CSMA/CA OR CSMA/CD. IT DOESN'T NEED TO! NO COLLISIONS OCCUR ON TOKENRING. TOKENRINGS PASSES THE TOKEN ROUND+ROUND THE RING. 6.10BASE5 IS ALSO KNOWN AS? A:THICKNET 7.10BASE2 USES WHAT GRADE OF COAX CABLE? A:10BASE2 USES RG58 WITH 50 OHM TERMINATORS 8.WHEN USING 10BASE2, ALWAYS REMEMBER TO.... A:TERMINATE EACH END OF THE BUS WITH 50 OHM TERMINATORS. IF THERE IS A BREAK OR DISCONNECTION IN THE BUS WILL RESULT IN A UNTERMINATED BUS, THUS AN UNUSEABLE NETWORK. 9.10BASET IS ALSO KNOW AS? A:TWISTED PAIR/ UTP / STP. 10.ARE THESE PROTOCOLS? A:TCPIP YES NETBEUI YES APPLETALK YES ARP YES NFS YES ETHERNET NO! ETHERNET IS A NETWORK MEDIA ACCESS METHOD 11.THE FACT THAT STAR TOPOLOGIES ALWAYS REQUIRE A HUB MEANS? A:STAR NETWORKS MUCH MORE WIRING THAN BUS TOPOLOGIES STAR NETORKS ARE EASIER TO TROUBLESHOOT THAN BUS TOPOLOGIES IF THE HUB FAILS, THAT STAR NETWORK HAS COMPLETE FAILURE IT IS VERY EASY TO ADD MORE COMPUTERS TO A STAR NETWORK 12.SMP MEANS? A:SYMMETRICAL MULTI-PROCESSING. A FEATURE THAN NT COMPUTERS HAVE. SMP MEANS ALL THE CPUS IN THE COMPUTER SHARE THE LOAD. 13.WHAT TOPOLOGY DOES FDDI USE? A:FDDI ONLY USES A RING TOPOLOGY. DUAL COUNTER-ROTATING FIBRE OPTIC RINGS.. 14.A GIGABTYE IS 10^9 = 1 THOUSAND MEGABYTES (1024MB) 15.WEB BROWSERS USE WHAT PROTOCOL? A:HTTP. HTTP IS THE PROTOCOL! ALWAYS REMEBER THAT WEB BROWSERS USE THE HTTP PROTOCOL. HTML IS ONLY A MARKUP LANGUAGE. HTML IS NOT A PROTOCOL! 16.YOUR SETTING UP A 10 STATION WINDOWS PEER2PEER NETWORK. WHAT NETWORK PROTOCOL SHOULD YOU USE? A:NETBEUI. TCPIP REQUIRES SOME ADMINISTRATION AS DOES IPX/SPX. NETBEUI IS SIMPLE AND NO CONFIGURATION IS REQUIRED. USE NETBEUI ON SIMPLE NETWORKS. USE TCPIP AND IPX/SPX ONLY ON REQUIRED. 17.THE OSI HAS HOW MANY LAYERS? A: 7 18."RAID 0" IS ALSO CALLED A:"DISK STRIPING". ANYTIME YOU SEE JUST "DISK STRIPING" AUTOMATICALLY ASSUME "DISK STRIPING WITHOUT PARITY". IF THEY MEAN "DISK STRIPING WITH PARITY" THEY'LL SAY "DISK STRIPING WITH PARITY" 19. TO INSTALL 100MBITS NETWORK. WHAT MINIMUM GRADE OF CABLE DOES IT NEED TO BE? A:TO RUN 100MBITS, REQUIRES CAT 5 GRADE CABLING. 20.SQL IS USED TO? A:ACCESS DATABASES. SQL IS A "NETWORK APPLICATION" 21.WHAT SPEED IS T1? A:1.544MBITS /SEC. 22.UTP IS? A:UNSHIELDED TWISTED PAIR. THE MOST COMMON MEDIA IN TODAYS NETWORKS. 23.WHY DOES A BRIDGE HAVE BETTER NETWORK PERFORMANCE THAN A REPEATER? A:BRIDGES ONLY NEED TO ANALYZE EACH PACKET AND FORWARD TO THE OTHER SIDE IF NECESSARY. REPEATERS MUST AMPLIFY AND RESEND EACH PACKET (EVEN MALFORMED BAD PACKETS) 24.ON ISDN, WHICH CHANNEL IS DATA CARRIED? A:ACTUAL DATA IS MOVED ON THE 2 B-CHANNELS.(64KB EACHx2=128KB) CONTROL IS ONE THE SLOWER 16KB D-CHANNEL. 25.OLD LOCALTALK, WHICH YOU'LL NEVER SEE AGAIN USES? A:CSMA/CA. IF YOU SEE LOCALTALK, USE COLLISION AVOIDANCE! :) 26.ARCNET USES WHAT NETWORK ACCESS METHOD? A:TOKEN PASSING. ARCNET WAS A CRAZY TOKEN PASSING NETWORK THAT PASSED RIGHT OUT OF EXISTENCE. 27.OUR EVERYDAY ETHERNET USES? A:CSMA/CD. COLLISIONS OCCUR ON ALL ETHERNET NETWORKS. THE MORE COMPUTERS AND THE BUSIER THOSE ETHERNET NETWORKS ARE, THE MORE COLLISIONS WILL OCCUR. 28.CSMA/CA IS USUALLY SLOWER THAN CSMA/CD. WHY? A:CSMA/CA MUST WAIT A BIT BEFORE TRANSMITTING. CSMA/CD DOES NOT WAIT. IF A COLLISION OCCURS, CSMA/CD TRANSMITS AGAIN. 29.NETWARE USES IPX/SPX 30.WHICH IS NOT ROUTEABLE? A: TCP/IP/SPX/IPX/APPLETALK ARE ALL ROUTEABLE. NETBEUI/DLC ARE NOT ROUTEABLE. 31.ATM CAN SWITCH AND ROUTE PACKETS EASILY BECAUSE? A:ALL ATM PACKETS ARE 53 BYTES LONG. BECAUSE ATM PACKETS ARE 1 SIZE, MAKES ATM SWITCHING AND ROUTING VERY EFFICIENT. 32.HOW DOES TOKENRING DETERMINE WHO GETS THE TOKEN? A:NIC IDS. EACH NIC IN A TOKENRING HAS A UNIQUE ID. THIS DETERMINES WHO GETS THE TOKEN. 33.WHY ARE WANS SLOWER THAN LANS? A:TO SAVE MONEY, WAN TECHNOLOGIES OFTEN HAVE USE LOWER BANDWIDTH PUBLIC SWITCHING NETWORKS TO COVER THE LARGE DISTANCES. 34.WHAT TOPOLOGIES IS EASIEST TO TROUBLESHOOT? A:STAR IS THE EASIEST TO TROUBLESHOOT. IF SOMEONE OR ONE COMPUTER COMPLAINS ABOUT NETWORK PROBLEMS, YOU ONLY HAVE TO WORRY ABOUT THEM. NO ONE ELSE ON THE STAR NETWORK IS AFFECTED.(UNLESS THAT ONE COMPUTER IS A SERVER) 35.NETBEUI WAS DESIGNED BY IBM FOR SMALL WORKGROUP NETWORKS. 36.A STAR NETWORK IS MORE EXPENSIVE THAN BUS MAINLY BECAUSE OF THE? A:EXTRA CABLE THAT STAR NETWORKS USE. 37.WHEN STAR NETWORKS ARE CONNECTED TOGETHER IT FORMS A? A:STAR BUS NETWORK 38.WHICH TOPOLOGY IS PASSIVE? A:BUS IS PASSIVE. RING IS ACTIVE. STAR CAN EITHER BE PASSIVE OR ACTIVE. (DEPENDING OF THE HUB USED). (STAR IS USUALLY ACTIVE) 39.THE LONGER ALL SIGNALS GO, THE WEAKER THEY GET. THIS IS CALLED? A:ATTENTUATION. 40.A REPEATER REGENERATES THE SIGNAL, EXTENDS THE MAXIMUM DISTANCE AND NEVER ANALYZES PACKETS. 41.LOGICALLY THE STAR NETWORK SENDS DATA LIKE? A:BUS TOPOLOGY 42.IF YOU DISCONNECTED COMPUTER FROM A 10BASE2 CONNECTOR AND MEASURE THE SHIELD TO THE CORE, HOW MANY OHMS DO YOU EXPECT? A:YOU MIGHT EXPECT 50 OHMS, BUT THE TWO 50 OHM TERMINATORS ARE IN PARALLEL SO YOU'LL MEASURE ONLY 25. 43.IN A 10BASE2 NETWORK, IF ALL OTHER COMPUTERS ARE OKAY, BUT ONLY ONE COMPUTER DOES NOT CONNECT TO THE NETWORK, WHAT CAN BE WRONG? A:EITHER THAT COMPUTER IS DISCONNECTED FROM THE 10BASE2 NETWORK, OR THE NIC HARDWARE OR SOFTWARE IS NOT PROPERLY CONFIGURED. 44.ARCNET QUESTIONS ARE NOT RELEVANT. 45.YOUR THE ADMIN FOR A SMALL, BUT FAST GROWING COMPANY, NOW 150. YOU WISELY SETUP A STAR NETWORK A YEAR AGO. HOWEVER THE NETWORK IS SLOW AMD USERS ARE COMPLAINING ABOUT IT. WHAT DO YOU DO? A:ALL ETHERNET NETWORKS ARE CSMA/CD. 150 COMPUTERS IN A SINGLE SEGMENT IS AWFUL BUSY. BUST THIS BITCH UP INTO A FEW SEGMENTS OF ABOUT 50 COMPUTERS EACH. COMPARING FILE+PRINT SERVERS WITH APPLICATION SERVERS THIN CLIENTS= THE THINNER THE CLIENT, USUALLY THE BETTER. THE THIN CLIENT IS MOSTLY BROWSER, SOFTWARE AND TERMINAL. THE SERVER DOES ALL THE WORK AND GIVE THE RESULTS BACK TO THE THIN CLIENT. PROGRAMS ARE NOT EXECUTED ON THE THIN CLIENT, THEY EXECUTE ON THE SERVER THIN CLIENT LOCAL DISK STORAGE IS LOW OR NON-EXISTENET. THIN CLIENTS, WILL BE THE "NET PC" THICK CLIENTS= WINDOWS NT AND 95 AND OTHER MODERN OS'S ARE ALL THICK CLIENTS. THICK CLIENTS EXECUTE PROGRAMS LOCALLY IN THEIR OWN RAM THICK CLIENTS HAVE LOCAL DISK SPACE FOR STORAGE. FILE SERVERS RAID LEVEL 0 STRIPING RAID LEVEL 1 DISK MIRRORING RAID LEVEL 2 DISK STRIPING WITH ECC RAID LEVEL 3 ECC STORED AS PARITY RAID LEVEL 4 DISK STRIPING WITH LARGE BLOCKS RAID LEVEL 5 DISK STRIPING WITH PARITY ONLY RAID LEVELS 015 ARE COMMON TODAY. (THE SAME ONES ONLY NT SUPPORTS) RAID LEVEL 0 DISK STRIPING WITHOUT PARITY= FASTEST DISK ACCESS AND NO FAULT TOLERANCE RAID LEVEL 1 DISK MIRRORING= USE WHEN DOWNTIME IS TO BE KEPT TO A MINIMUM WRITES DATA TO 2 DISKS, KEEPING THEM EXACTLY THE SAME. ONLY FAULT TOLERANT METHOD THAT CAN BE USED ON SYSTEM+BOOT PARTITIONS RAID LEVEL 5 DISK STRIPING WITH PARITY= FAST DISK ACCESS AND BETTER UTILIZATION AS NUMBER OF DISKS INCREASE. 1/N DISKS IS USED AS PARITY. ALL PARTITIONS MUST BE EXACTLY THE SAME SIZE AND BE ON DIFFERENT PHYSICAL DISKS. PRINT SERVERS PRINT SERVERS MANAGE A SINGLE OR MANY PRINTERS FOR A NETWORK. APPLICATION SERVERS AN APPLICATION SERVER RUNS AN APPLICATION ON THE SERVER FOR THE CLIENT, AND GIVES RESULTS BACK TO THE CLIENT. SQL IS A GOOD EXAMPLE. CLIENT ASKS FOR AN SQL RECORD, SQL SERVER DOES ALL THE WORK AND EVENTUALLY GIVES ONLY THE RESULTS BACK TO THE REQUESTING CLIENT. APPLICATIONS SERVERS REQUIRE THE MOST CPU POWER. MORE PRACTICE PROBLEMS!! YEEEEEEEE HAAAAAAAAAAAAAAAA! ***************************************************** 1.YOUR INSTALLING AN SQL SERVER FOR 10,000,000 RECORDS DATABASE. WHAT KIND OF COMPUTER SQL SERVER SHOULD YOU SELECT? A:PICK THE COMPUTER WITH THE MOST CPU HORSEPOWER. LIKE A DELL XPS-10000 WITH 4 PENTIUM III-800MHZ CPUS'S SOUNDS LIKE A GOOD PICK. 2.NEXT YOU NEED TO SELECT THE FILE SERVER FOR 200 USERS. WHAT KIND SHOULD YOU SELECT? A:DISK ACCESS SPEED AND FAULT TOLERANCE ARE THE 2 MOST IMPORTANT FACTORS FOR A FILE SERVER. RAM AND CPU HORSEPOWER IS NOT AS IMPORTANT FOR A FILESERVER. PENTIUM 100 64MB GOOD HARDWARE RAID 5 WITH 4 SEAGATE 12GB ULTRASCSI LVD DRIVES @ 7200RPM. USE 3 DRIVES TO MAKE THE RAID 5, 1 AS THE HOT SPARE. 3.LASTLY, YOU NEED TO SETUP THE PRINT SERVER. WHAT IS A GOOD CONFIGURATION FOR A PRINT SERVER FOR THE 200 USERS? A:PRINT SERVERS REQUIRE VERY LITTLE CPU POWER. AFTER ALL MOST OF THE PRINT PROCESS IS DONE BY THE PRINT DEVICE. A FAIR BIT OF DISK STORAGE MAY BE NEEDED FOR THE SPOOLERS, ESPECIALLY IT IT IS ART WORK. BUY THE LARGEST DRIVE YOU'LL NEED. SAY 50 CMYK JOBS ?(TWO 25GB IDES AS A STRIPE) PENTIUM 100 64MB, ONE 25GB IDE DRIVE. 4.YOU HAVE 10 MARKET ANALYSTS THAT USE BROWSERS TO MONITOR STOCK INFO. USE THICK OR THIN CLIENTS? A:THIN. THESE GORDON GECKO WANNABES DON'T NEED ANY REAL COMPUTER. JUST GIVE THESE WEENIES A THIN CLIENT WITH A BROWSER. 5.A PRINT SERVER WILL? A:FREE UP CLIENT COMPUTERS FASTER SO THEY CAN GO BACK TO WORK. PRINT SERVERS ARE ALWAYS READY TO ACCEPT ANOTHER PRINT JOB. PRINT SERVERS WILL NOT MAKE YOUR PRINTERS FASTER, JUST MAKE THEM MORE EFFICENT. PRINT SERVERS DO NOT REQUIRE HARDLY ANY CPU HORSEPOWER. I MEAN HOLY COW, ALL PRINT SERVERS HAVE TO DO IS ACCEPT PRINT JOBS AND SEND THEM OFF TO THE PRINTER. HOW CPU-INTENSIVE IS THAT? HARDLY NADA. 6.A FILE SERVER REQUIRES? A:FILE SERVERS REQUIRE THE FASTEST DISK ACCESS SYSTEMS YOU CAN AFFORD. DISK I/O IS USUALLY YOUR FILE SERVERS MAIN ENEMY. SOMETHING LIKE A DYNATEK HARDWARE RAID 5 WITH SEAGATE 12GB ULTRASCSI-3 LVD DRIVES WILL PROVIDE AWESOME LOCAL DISK ACCESS. 7.WHY DO SYADMINS USE RAID? A:BECAUSE THEY WANT TO KEEP THEIR JOBS. BECAUSE ALL RAIDS PROVIDE FAULT TOLERANCE FOR SINGLE DISK FAILURES, ONLY. IF MORE THAN A SINGLE DISK FAILS, THE RAID IS USELESS. 8.THE SIMPLEST AND EASIEST RAID TO IMPLEMENT IS? A:RAID LEVEL 1. DISK MIRRORING. REQUIRES 2 PHYSICAL DISKS AND PARTITIONS ON EACH DRIVE OF EQUAL SIZE (PHYSICAL DRIVES DO NOT HAVE TO BE IDENTICAL MODELS) 9.YOUR AT A NEW MEDIA COMPANY WHERE THEY ARE STREAMING INTRANET VIDEO FILES. ALL STREAMING FILES ARE EASILY REPLACED VIA MASTER CDROMS. WHAT RAID SHOULD YOU SELECT TO STORE AND STREAM THESE VIDEO FILES ? A:RAID LEVEL 0 DISK STRIPING NO PARITY. IT IS FASTER THAN RAID LEVEL 5 SINCE NO PARITY CPU WORK NEEDS TO BE DONE. PLUS IF A DRIVE GOES DOWN, REPLACE IT AND BE BACK UP 15 MINUTES. 10.YOUR NETWORK HAS 5 HP4000TN NETWORK LASERPRINTERS AT A SINGLE LOCATION NEXT TO THE PRINT SERVER. MAXIMUM SPEED IS REQUIRED. HOW DO YOU CONNECT THEM? A: INSTALL THE PRINT SERVER AND THE 5 PRINTERS INTO A SEPERATE NETWORK. INSTALL ANOTHER NIC INTO THE PRINT SERVER FOR A TOTAL OF 2 NICS. WHY YOU ASK? BECAUSE IF YOUR PRINTERS EXIST ON THE SAME NETWORK AS THE SERVERS AND YOUR CLIENTS, EACH PRINT JOB MUST BE SENT TWICE OVER THE NETWORK!! ONCE FROM THE CLIENTS TO THE PRINT SERVER. ONCE FROM THE PRINT SERVER TO THE PRINT DEVICE. THIS ADDS UP TO MORE NETWORK TRAFFIC. IF YOU HAVE A REAL BUSY PRINT ENVIRONMENT, THEN PUT THE PRINTERS IN THEIR OWN "PRINT NETWORK" ONE NIC IS TO CONNECT TO YOUR ENTIRE NETWORK, THE OTHER NIC IS TO CONNECT ONLY TO THE "PRINTER NETWORK" COOL! WHO THINKS UP SHIT LIKE THIS? 11.WHAT IS ARE EXAMPLES OF CLIENT/SERVER COMPUTING? A:WORKSTATION NT ACCESSING SQL DATABASE ON SQL SERVER IS CLIENT/SERVER. OUTLOOK ACCESSING EXCHANGE TO FIND AN EMAIL ADDRESS IS CLIENT/SERVER. A TERMINAL ACCESSING A MAINFRAME DB IS NOT CLIENT/SERVER. 12.AN SQL SERVER IS AN EXAMPLE OF? A:AN SQL SERVER IS AN APPLICATION SERVER IN CLIENT/SERVER ARCHITECTURE. 13.WHAT IS TRUE ABOUT CLIENT/SERVER COMPUTING? A: 14.CLIENT/SERVER IS COMMONLY SEEN? A:DATABASE SERVERS LIKE SQL. 15.ON A SERVER-BASED NETWORK, WHAT HANDLES THE FILE AND PRINT REQUESTS ? A:THE REDIRECTOR 16.WHAT ARE THE BENEFITS OF STORING THE APPLICATIONS ON AN APPLICATION SERVER, INSTEAD OF LOCALLY? A:MUCH LOWER LICENSING COSTS.SINCE ONLY ONE COPY OF THE SOFTWARE IS RUNNING!! MUCH EASIER TO UPGRADE ALL THE SOFTWARE. ONLY ONE VERSION TO UPGRADE!! REQUIRES MUCH LESS DISKSPACE. ONLY ONE COPY NEEDS TO BE INSTALLED INSTEAD OF 10MB ON EVERY LOCAL MACHINE. IF THE APPLICATION SERVER EVER GOES DOWN..EVERYONE USING THAT SERVER GOES DOWN. 17.SPOOLING A PRINT JOB BASICALLY MEANS? A:SPOOLING MEANS TO STORE THE PRINT JOB IN MEMORY OR IN A QUEUE UNTIL THE PRINT DEVICE IS READY FOR IT. 18.YOUR CLIENTS PRINTOUT IS GARBAGE. WHAT IS LIKELY TO BE THE PROBLEM? A:THE SELECTED PRINT DRIVER DOES NOT MATCH THE PRINTER. 19.ON A CLIENT/SERVER NETWORK THE SERVER SHOULD BE? A:THE MOST POWERFUL WHOOP-ASS COMPUTER YOU GOT. 20.YOUR BOSS ASKS YOU WHAT CLUSTERING IS. A:CLUSTERING IS MAKING A GROUP OF COMPUTERS APPEAR ON THE NETWORK AS ONE SUPERFAST COMPUTER. CLUSTERING USES FAIL OVER. IF ONE COMPUTER IN THE CLUSTER FAILS, OTHER COMPUTERS IN THE CLUSTER TAKE OVER ITS WORK. 21.SMP WOULD BE VERY USEFUL FOR ? A:AN APPLICATION SERVER RUNNING SQL. SMP WOULD ALLOW THIS APP SQL SERVER TO HANDLE REQUESTS IN PARALLEL. SMP FOR A FILESERVER IS NOT BENEFICIAL. FILE SERVERS ARE DISK I/O INTENSIVE, NOT CPU INTENSIVE. SMP FOR A PRINT SERVER IS STUPID. PRINT SERVERS NEED VERY LITTLE CPU. THE ONLY TIME SMP OR HEAVY DUTY CPU POWER IS NEEDED IS FOR A RIP. IF YOU RUN A RIP SOFTWARE PRODUCT, THEN USE THE FASTEST CPU SYSTEM YOU GOT. A RIP RIPS PRINT JOBS. A PRINT SERVER SENDS PRINT JOBS OUT TO THE PRINT DEVICE. THESE ARE DIFFERENT TASKS. 22.ALMOST ALL NETWORK-AWARE APPLICATIONS LOCK THEIR FILES...WHY?? A. FILE-LOCKING PREVENTS FILES FROM BEING OPENED IN WRITE-MODE BY SEVERAL USERS AT ONCE. THIS WILL CORRUPT A FILE! FILES CAN BE OPENED IN READ-ONLY MODE BY SEVERAL COMPUTERS. THIS DOES NOT CORRUPT A FILE. 23.POSTSCRIPT IS A PRINTER LANGUAGE DEVELOPED BY ADOBE SYSTEMS. IT IS USED ON MOST PRINTERS. AND ALMOST EXCLUSIVELY ON ALL HIGH QUALITY OUTPUT DEVICES. 24.WINDOWS NT SUPPORTS RAIDS? A:015 25.WHAT DOES HOT-SWAPPABLE MEAN? A:A DISK THAT CAN BE REPLACED WITHOUT POWERING OFF THE SYSTEM. HOT SWAPPABLE WILL ONLY BE SEEN IN HARDWARE RAID SYSTEMS. COMPARING USER-LEVEL SECURITY WITH FILE-LEVEL SECURITY THERE ARE 2 KINDS OF SECURITY MODELS TODAY 1.SHARE LEVEL (PROTECTION USING PASSWORDS) 2.USER LEVEL (PROTECTION USING PERMISSIONS ASSIGNED TO THE RESOURCE) 1.SHARE LEVEL (PASSWORD PROTECTED) SECURITY EACH SHARED RESOURCE IS OPTIONALLY ASSIGNED A PASSWORD. THE USER PROVIDES THE CORRECT PASSWORD AND IS GRANTED ACCESS. SHARE LEVEL SECURITY DOES NOT USE USER AUTHENTICATION USING SHARE-LEVEL MEANS ONLY ONE LEVEL OF ACCESS EXISTS. WINDOWS 95/98/WFW USES SHARE-LEVEL 2. USER-LEVEL (ACCESS PERMISSIONS) USER-LEVEL ACCESS PERMISSIONS ARE MORE ADVANCED THAN SHARE LEVEL. PERMISSIONS ARE ASSIGNED USER BY USER BASIS. USER AUTHENTICATION IS USED IN USER-LEVEL SECURITY USERS=PASSWORDS, RESOURCES=PERMISSIONS NT USES USER-LEVEL SECURITY BOTH IN DOMAIN AND WORKGROUP MODELS WORKGROUP MODEL SECURITY IN THE WORKGROUP MODEL EVERY WORKSTATION AND SERVER MAINTAINS ITS OWN LOCAL SAM DATABASE. WHEN A USER LOGONS A TOKEN IS MADE FOR THAT USER. WHEN A USER ACCESSES A RESOURCE, THE USER EITHER LOGONS OR ACCEPTS THE TOKEN OF HIS CURRENT PERMISSIONS. IF YOU MAKE A PERMISSION CHANGE TO THE RESOURCE, THE USER DOES NOT NEED TO LOGOUT, TO USE THE RESOURCE. BUT IF YOU MAKE A CHANGE VIA THE USER MANAGER, THE USER MUST LOGOUT AND BACK AGAIN. EVERY RESOURCE IN NT HAS AN ACL. ACCESS CONTROL LIST.THE ACL LISTS ALL THE GROUPS OR USERS WHO HAVE ACCESS TO THIS RESOURCE. EVERY ENTRY IN THE ACL, IS AN ACE. ACCESS CONTROL ENTRY. THE ACE DETERMINES WHAT PERMISSIONS A USER OR GROUP HAS TO THAT RESOURCE. AS THE NUMBER OF WORKSTATIONS AND SERVERS GROWS, THIS WORKGROUP MODEL GETS TO BE EXTREMELY ADMINISTRATIVE. DOMAIN MODEL SECURITY THE DOMAIN MODEL USES ONE SAM ON THE PDC FOR THE ENTIRE DOMAIN. THE LSA, LOCAL SECURITY AUTHORITY USES THE DOMAIN SAM AS ITS PRIMARY AUTHENTICATION TOOL. WINDOWS NT GROUPS AND PERMISSIONS THE SECURITY OF A RESOURCE DEPENDS ON WHAT TYPE OF RESOURCE IT IS. A PRINTER IN NT HAS 4 LEVELS OF SECURITY 1.FULL CONTROL= USER OR GROUP ALLOWED FULL CONTROL OVER ALL ASPECTS OF THAT PRINTER 2.MANAGE DOCUMENTS= USER OR GROUP IS ALLOWED TO MANAGE ALL PRINT JOBS SENT TO THAT PRINTER. (MANAGE HIS OWN AND PRINT JOBS FROM OTHER USERS) 3.PRINT= USER OR GROUP IS ALLOWED TO MANAGE ONLY HIS PRINT JOBS SENT TO THAT PRINTER. 4.NO ACCESS= USER/GROUP IS DENIED ACCESS TO THE PRINTER. NTFS PERMISSIONS 1.FULL CONTROL= USER/GROUP HAS FULL CONTROL OF THAT FILES/FOLDERS 2.READ= USER/GROUP HAS RX (READ+EXECUTE) PRIVILEDGES OF THOSE FILES/FOLDERS 3.LIST= USER/GROUP IS ALLOWED TO VIEW DIRECTORY LISTING OF THAT FOLDER 4.ADD= USER/GROUP IS ALLOWED TO ADD FILES/FOLDERS TO THIS FOLDER 5.NO ACCESS= USER/GROUP HAS NO ACCESS 6.CHANGE= USER/GROUP IS ALLOWED TO READ/WRITE/DELETE/CREATE FILES/FOLDERS, BUT NOT ADD. 7.ADD+READ= ADD FILES/FOLDERS.CREATE FILES/FOLDERS SHARE PERMISSIONS (ONLY SECURITY AVAILABLE ON FAT FILESYSTEMS) 1.FULL CONTROL= USER/GROUP HAS FULL CONTROL OF THE SHARE 2.READ= USER/GROUP ALLOWED TO READ/EXECUTE FROM THAT SHARE 3.CHANGE= USER/GROUP ALLOWED TO READ/WRITE/DELETE/ADD FILES/FOLDERS 4.NO ACCESS= USER/GROUP HAS BE EXPLICITY DENIED NO ACCESS. THE EASIEST WAY IS TO REMOVE EVERYONE FROM NTFS PERMISSIONS THEN ASSIGN A SPECIFIC USE/GROUP TO THAT RESOURCE. PRACTICE PROBLEMS 1.WINDOWS FOR WORKGROUPS USES WHAT KIND OF SECURITY? A:WFW WIN95/98 ALL USE SHARE-LEVEL (PASSWORD PROTECTED) 2.USER-LEVEL SECURITY IS GOOD FOR? A:LARGE NETWORKS. WITH LARGE NETWORKS IT BECOMES MANDATORY TO USE USE-LEVEL SECURITY. RUNNING A LARGE NETWORK BASED ON SHARE-LEVEL SECURITY IS INSANE. SHARE-LEVEL=SMALL WORKGROUPS, USER-LEVEL=LARGE DOMAINS 3.TO USE USER-LEVEL SECURITY, WHAT FILEYSTEM IS REQUIRED? A:NTFS. NTFS FILESYSTEMS CAN USE BOTH USER AND SHARE LEVEL SECURITY. FAT FILESYSTEMS CAN ONLY USE SHARE LEVEL SECURITY. 4.IF YOU WANT TO SET PERMISSIONS ON SHARES, WHAT FILESYSTEM DO YOU NEED? A:ANY. PERMISSIONS CAN BE ASSIGNED TO SHARES ON ANY FILESYSTEM. DOESN'T MATTER IF THE SHARE IS ON HPFS, NTFS OR FAT. 5.DISCRETIONARY ACCESS CONTROL MEANS? A:THE OWNER OF THE RESOURCE DECIDES WHO AND THE TYPE OF ACCESS TO THAT RESOURCE. 6.WHICH OPERATING SYSTEMS SUPPORT DISCENTIONARY CONTROL? A:WINDOWS NT ONLY 7.WHAT 2 TYPES OF NETWORKING MODELS DOES NT SUPPORT? A:DOMAIN MODEL, WORKGROUP MODEL 8.IN YOUR NT WORKGROUP, THERE ARE 100 WORKSTATIONS AND 5 NT SERVERS, HOW MANY SAMS DO YOU HAVE TO MAINTAIN? A:YOU HAVE TO KEEP ALL 105 SAMS IN PEFECT SYNC FOR YOUR NETWORK TO BE RELIABLE. HAVE FUN. 9.IN YOUR NT DOMAIN THERE ARE 1400 NT WORKSTATIONS AND 72 NT SERVERS. HOW MANY SAMS DO YOU HAVE TO MAINTAIN? A:ONE. (ON THE PDC). 10.NT WORKGROUP IS EASIER WITH 9 CLIENTS AND 1 SERVER. ANY BIGGER, USE THE DOMAIN. 11.WHAT DOES THE SAM DATABASE STORE? A:USERNAMES AND PASSWORDS AND USER ACCOUNT INFORMATION. 12.IN PEER2PEER, SHARE-LEVEL SECURITY IS USED. 13.ACE MEANS ACCESS CONTROL ENTRY. 14.ACL IS THE ENTIRE LIST OF ACE'S. 15.SHARE-LEVEL USES PASSWORDS FOR THE RESOURCE, NOT PERMISSIONS. USER-LEVEL USES PERMISSIONS FOR THE RESOURCE, NOT PASSWORDS. 16.WHEN A USER LOGONS, WHO DOES THE AUTHENTICATING? A:LSA, LOCAL SECURITY AUTHENTICATION 17.WHEN YOU ASSIGN SOMEONE "PRINT" PERMISSIONS TO A PRINTER, WHAT RIGHTS DOES HE HAVE? A: TO CREATE/DELETE HIS OWN PRINT JOBS. NO ONE ELSES. 18.WHAT ARE THE 4 SHARE PERMISSIONS? A:FULL CONTROL, READ, CHANGE, NO ACCESS 19.WHAT ARE THE 7 NTFS (USER-LEVEL) PERMISSIONS? A: FULL CONTROL, READ, LIST, ADD, NO ACCESS, CHANGE, ADD+READ 20.WHAT ARE THE 4 PRINT PERMISSIONS? A:FULL CONTROL, MANAGE DOCUMENTS, PRINT, NO ACCESS 22.IN THE DOMAIN MODEL, THE LOCAL LSA, LOCAL SECURITY AUTHORITY MUST VERIFY LOGINS WITH A DOMAIN CONTROLLER. 23.DOMAINS ARE EASIER TO MANAGE THAN WORKGROUPS BECAUSE? A:ONLY 1 SAM NEEDS TO BE ADMINISTERED 24.WHEN A RESOURCE IS USED, THE USERS ACCESS IS VERIFIED BY? A:CHECKING THE USERS SID TOKEN AGAINST THE ACL+ACE ENTRY. 25.IF YOU DELETE A USER ACCOUNT AND RECREATE IT EXACTLY THE SAME AGAIN WITH THE SAME PERMISSIONS AND GROUPS, WHAT HAPPENS WHEN YOU TRY AND USE IT? A:ALL SECURITY INFORMATION IS LOST. YOU MUST NOW MANUALLY RECREATE ALL PREVIOUS SECURITY PERMISSIONS ON ALL RESOURCES ALSO. THIS IS BECAUSE WHEN YOU DELETED THE ACCOUNT, YOU ALSO DELETED IT'S SID. RECREATING THE ACCOUNT, WILL GENERATE A NEW SID NUMBER. SO TO YOU THE ACCOUNT MAY HAVE THE SAME NAME AND SETTINGS, BUT TO NT, IT IS NOT THE SAME ACCOUNT IS WAS ORIGINALLY. 26.A USER OF YOUR COMPLAINS OF "NO ACCESS" WHEN ACCESSING AN NTFS FOLDER. HE HAS "CHANGE" ACCESS TO THE SHARED FOLDER. WHAT IS HIS PROBLEM? A:HE HAS "CHANGE" ACCESS FOR THE SHARED FOLDER. THAT IS OKAY. BUT HIS NTFS PERMISSIONS FOR THAT FOLDER ARE ARE WRONG. YOU FORGOT TO GIVE HIM NTFS "CHANGE" PERMISSION TO THAT FOLDER. 27.TO CREATE FOLDERS, WHAT IS THE MINIMUM PERMISSION NEEDED? A:ADD IS THE LOWEST PERMISSION TO THAT CAN CREATE A FOLDER. THE USER CAN CREATE FOLDERS, BUT CAN'T ACCESS THE FOLDERS HE HAS CREATED. KOOKY. ALMOST ALL NETWORKS HAVE SOME OF THE FOLLOW COMMON COMPONENTS 1.SERVERS POWERFUL COMPUTERS THAT SHARE (USUALLY LOCAL DISK) RESOURCES 2.RESOURCES DISKS, PRINTERS, FAXES, AND ALL OTHER RESOURCES 3.CLIENTS MACHINES THAT USE THE RESOURCES OF THE SERVER AND THE NETWORK 4.MEDIA SOME FORM OF CABLE OR WIRELESS MEDIA TO NETWORK COMPUTERS TOGETHER. SERVER-BASED NETWORK CLIENTS REQUEST AND ACCESS RESOURCES FROM SERVERS PEER2PEER A NETWORK OF EQUALS. ALL CLIENTS SHARING EACH OTHERS RESOURCES. WHEN DECIDING WHICH ONE TO INSTALL, CONSIDER THESE 4 FACTORS 1.NUMBER OF COMPUTERS OVER 10 USER SERVER BASED. UNDER 10, USE PEER2PEER. 2.COST SERVER-BASED IS MUCH MORE EXPENSIVE. PEER2PEER IS CHEAP. 3.SECURITY SERVER-BASED CAN BE MADE SECURE. PEER2PEER IS INHERENTLY INSECURE. 4.ADMINISTRATIVE SERVER-BASED IS FAR EASIER TO ADMINISTER PEER2PEER IS VERY BAD TO ADMINISTER (BELIEVE ME I KNOW) PEER2PEER ADVANTAGES EASY TO SETUP AND INSTALL LOWER COST NO NETWORK ADMINISTRATOR IS REQUIRED (IF ALL USERS ARE NT MCSE'S) CLIENTS HAVE FULL CONTROL OVER THEIR OWN RESOURCES/SHARES/SECURITY A CHEAP NETWORK INSTALLATION, LIKE A SMALL STAR OR BUS FITS THE BILL DISADVANTAGES OF PEER2PEER PEER2PER IS NOT SCALABLE (SERVER-BASED IS SCALABLE) PEER2PEER NETWORKS ARE LIMITED TO ABOUT 10 CLIENTS AT MOST! NO CENTRAL ADMINISTRATION OVERALL WEAK SECURITY COMPUTERS MUST SHARE THE LOAD OF BEING BOTH SERVERS AND CLIENTS. ADVANTAGES OF SERVER-BASED CENTRALIZED ADMINISTRATION MEANS THOUSANDS OF CLIENTS CAN BE HANDLED STRONG SECURITY AND CAN BE MADE STRONGER IF NEEDED SERVER-BASED IS SCALABLE (PEER2PEER IS NOT) DISADVANTAGES OF SERVER-BASED MORE EXPENSIVE DUE TO EXPENSIVE SERVER HARDWARE AND PERSONNEL LONGER AND MORE DIFFICULT TO INITIALLY SETUP A DOMAIN THAN A PEER2PEER REQUIRES A ADMINISTRATOR TO MAINTAIN. PRACTICE PROBLEMS 1.WHAT TYPE OF NETWORKING DO WFW AND WINDOWS 95/98 USE? A:PEER2PEER, (SHARE-LEVEL) SECURITY 2.PEER2PEER SHOULD ONLY BE CHOSEN WHEN THE NETWORK WILL BE 10 OR LESS CLIENTS FOR THE LONG TERM. 3.SECURITY IN PEER2PEER IS A:THE RESPONSIBILITY OF EACH CLIENT. MAY REQUIRE AN ADMINISTRATORS ASSISTANCE. 4.SECURITY IN SERVER-BASED IS A:STRONG AND CENTRALLY MAINTAINED 5.A WINDOWS WORKGROUP IS PEER2PEER. 6.A WINDOWS DOMAIN IS SERVER-BASED. 7.IF COST IS MOST IMORTANT WHEN DECIDING ON A NETWORK, WHAT IS TRUE? A:A PEER2PEER WILL BE THE SHORT-TERM CHEAPEST CHOICE. IF THE NETWORK IS MORE THAN 10 CLIENTS, AND THEY WANT IT DONE AS CHEAP AS POSSIBLE BY INSTALLING PEER2PEER, THE EXTRA SUPPORT CALLS WILL QUICKLY EAT AWAY AT THE MONEY THEY SAVED INITIALLY. AND KEEP EATING AWAY UNTIL THE NETWORK IS CONVERTED OVER TO DOMAIN, AND HIRE A PT NETWORK ADMIN. 8.A NEW NETWORK HAS 25 CLIENTS, ONE SERVER AND NO ADMIN. WHAT SHOULD YOU SET THEM UP AS? A:STANDALONE NT SERVER IN A WORKGROUP. MAKE ALL CLIENTS, WINDOWS 95 AND JOIN TO THAT WORKGROUP. THIS MAKES GREAT SENSE, SINCE THE NT SERVER CAN BE SETUP FOR SECURITY AND HANDLE ALL THE FILE SERVER AND PRINT SERVER DUTIES. CLIENTS ARE OFTEN MORE USED TO WIN95 THAN NT, AND CAN USE THEIR CLIENT MACHINES EFFECTIVELY, BUT STILL ENJOY THE BENEFITS OF A SERVER IN A PEER2PEER NETWORK. 9.WHAT IS THE MAX SIZE OF A FILE IN NT? A:16EB 10.A UNIX FRIEND ASKS YOU HOW MANY CHARACTERS DOES NT FILENAMES SUPPORT?, YOU SAY? A:255 CHARACTERS. JUST LIKE UNIX. 11.HOW BIG IS AN EXABYTE? A:A MILLION GIGABYTES. 12.YOUR BOSS HAS READ THAT NT CAN SMP. HOW MANY HE ASKS? A:OUT OF THE BOX, WINDOWS NT SERVERS CAN SUPPORT SYSTEMS WITH 4 PROCESSORS. (WORKSTATION ONLY 2) WITH A NEW HAL FROM THE OEM, SUPPORT FOR 32 IS POSSIBLE. 13.TO SAVE MONEY, YOUR BOSS IS THINKING OF PURCHASING 50 DELL PENTIUM75 DEMO SYSTEMS. THESE DEMO SYSTEMS HAVE ONLY 8MB OF RAM. IF WE BUY THESE P75 DELLS, WHAT OPERATING SYSTEM SHOULD WE USE, HE ASKS? A:NT WORKSTATION IS OUT THE WINDOW. NTWS NEEDS 12 AT MINIMUM! NT SERVER IS ALSO OUT THE WINDOW. SERVER NEEDS 16 AT MINIMUM! WINDOWS 3.1 IS NOT A GOOD CHOICE. NO NETWORKING! WFW OR WINDOWS 3.11 IS A POSSIBLE OPTION, SINCE IT DOES NETWORKING AND WORKS WITH 8MB JUST FINE. WINDOWS 95 IS THE BEST CHOICE. W95 WILL WORK WITH 8MB FINE AND DO ALL THE NETWORKING THAT IS REQUIRED. YOU LATER DO SOME INVESTIGATION AND FIND THAT ALL THE DELL P75'S HAVE 1GB IDE DRIVES, 3COM NICS 10BASE-T NICS, AND USE 72-PINS SIMMS. YOUR COMPUTER DEALER IS SELLING 16MB 72-PINS SIMMS FOR $10 EACH! YOU TELL YOUR BOSS TO BUY THE P75'S AND GET 50 16MB SIMMS. THEN INSTALL WINDOWS NT ON EACH AND INSTALL AS A DOMAIN. A P200 WILL BE YOUR SERVER. 14.IN WHAT NETWORK, WILL CLIENTS HAVE TO BE BOTH A SERVER AND CLIENT? A:IN PEER2PEER WILL A CLIENT HAVE TO BE BOTH SERVER AND CLIENT. 15.USER-LEVEL ACCESS MEANS? A:RESOURCES HAVE PERMISSIONS, NOT PASSWORDS. 16.SHARE-LEVEL MEANS? A:RESOURCES HAS PASSWORDS, NOT PERMISSIONS. 16.YOUR BOSS HAS BROUGH IN A MIPS RX4000 CPU FROM AN AUCTION TO USE AS A TEST COMPUTER. CAN WE PUT WINDOWS ON IT HE ASKS? A:ONLY WINDOWS NT WORKSTATION OR SERVER VERSION 4. 17.YOU WORK AT GE. GE HAS DECIDED TO START AN INTERNAL ISP FOR 2000 GE EMPLOYEES. YOU HAVE 6 NT SERVERS TO SETUP TO HANDLE THE REMOTE CALLS. HOW MANY CALLS IN TOTAL WILL THESE 6 NT SERVERS HANDLE? A:EACH NT SERVER CAN HANDLE 256 RAS CONNECTIONS. X6 =1536 18.YOUR AT THE COMPUTER STORE. THERE IS THIS DELL PC 486/33 WITH 12MB OF RAM. YOUR BOSS WANTS ANOTHER BDC. WILL THIS BE ENOUGH FOR A BDC? A:NO, NT SERVER REQUIRES 16MB. PERHAPS IF THE DEALER WERE TO INSTALL SOME MORE RAM? THE 486/33 IS OKAY FOR NT SERVER. RUNNING A BDC FOR A REAL BUSINESS ON A 486/33 WILL WORK.(BUT IS NOT COOL) USING CONNECTION-ORIENTED VERSUS CONNECTIONLESS NETWORK PROTOCOLS, AND COMMUNICATIONS CAN EITHER BE 1.CONNECTION-ORIENTATED TCP 2.CONNECTION-LESS UDP 1.CONNECTION-ORIENTATED PROTOCOLS LIKE TCP WORK BEST IN WANS. BECAUSE MORE TRANSMISSION ERRORS OCCUR IN WANS. 2.CONNECTION-LESS PROTOCOLS LIKE UDP DO NOT WORK WELL IN WANS. WHERE MORE TRANSMISSION ERRORS OCCUR. 4.CONNECTION-LESS PROTOCOLS LIKE UDP WORK BEST IN LANS. BECAUSE LESS TRANSMISSION ERRORS OCCUR IN LANS. 3.CONNECTION-ORIENTATED PROTOCOLS LIKE TCP DO NOT WORK WELL IN LANS. TRANSMISSIONS ERRORS DO NOT OCCUR AS OFTEN IN LANS. IT WASTES OVERHEAD. CONNECTION-ORIENTATED A LOGICAL CONNECTION EXISTS FROM THE SOURCE TO THE TARGET. CONNECTION-ORIENTATED IS LIKE FEDERAL EXPRESS. THE PACKET IS USUALLY DELIVERED, BUT IS NOT GUARANTEED TO BE DELIVERED. IF IT IS NOT DELIVERED, YOU'LL KNOW WHY. CONNECTION-LESS MODE CONNECTION-LESS IS LIKE THE REGULAR MAIL. THE PACKET OFTEN GETS DELIVERED, BUT NOT GUARANTEED, JUST LIKE FEDEX. HOWEVER WITH FEDEX, YOU'LL KNOW WHY THE PACKET WAS NOT DELIVERED. WITH REGULAR MAIL (CONNECTIONLESS) YOU MAY EVER KNOW WHY. THE PACKET(MAIL) MAY EVEN GET LOST. POST OFFICE DOESN'T CARE. THEIR ATTITUTE IS TOO BAD, TRY AGAIN. IS AN ENVIRONMENT WHERE MANY TRANSMISSION ERRORS OCCUR (WAN) (LOTS OF MAIL GETS LOST), CONNECTION-LESS IS A BAD CHOICE. PRACTICE PROBLEMS 1.WHAT ISO LAYER DETERMINES THE ROUTE A PACKET WILL TAKE THROUGH A NETWORK? A:3.NETWORK LAYER IS WHERE ALL ROUTING DECISIONS ARE MADE. 2.WHEN THERE ARE FEW TRANSMISSION ERRORS, WHAT CONNECTION MODE IS FASTER? A:CONNECTION-LESS IS ALWAYS FASTER THAN CONNECTION-ORIENTATED. BUT WHEN LOTS OF TRANSMISSION ERRORS OCCUR, CONNECTION-ORIENTATED IS FASTER. 3.IN A LAN, WHAT CONNECTION-MODE SHOULD YOU USE? A:USE A CONNECTION-LESS PROTOCOL LIKE UDP, SINCE FEW ERRORS OCCUR. 4/WHAT INTERNET PROTOCOL USES CONNECTION-LESS? A:UDP. 5.WHAT INTERNET PROTCOL IS CONNECTION-ORIENTATED? A:TCP 6.WHAT OSI LAYER HANDLES THE TRANSMISSION OF DATA BEWEEN NODES? A:DATALINK LAYER. 7.WHICH OF THE 2 CONNECTION MODES, HAS NO ERROR CHECKING? A:CONNECTION-LESS HAS NO ERROR CHECKING. IF ERROR CHECKING IS TO BE DONE, IT MUST BE DONE AT HIGHER OSI LEVELS. 8.THE INTERNET USES WHAT CONNECTION MODE? A:THE INTERNET USES BOTH CONNECTION-ORIENTATED AND CONNECTION-LESS. SOME OF THE MANY INTERNET PROTOCOLS ARE BETTER SUITED FOR CONNECTION-LESS, SOME BETTER SUITED FOR CONNECTION-ORIENTATED. 9.WHAT MODE GUARANTEES DELIVERY OF ALL PACKETS WILL BE DELIVERED IN SEQUENCE? A:CONNECTION-ORIENTATED MAKES SURE ALL PACKETS ARE IN PROPER SEQUENCE. THE PACKETS MAY ARRIVE AT THE TARGET COMPUTER OUT OF ORDER, BUT THE NETWORK LAYER, ASSEMBLES THEM IN PROPER SEQUENTIAL ORDER. CONNECTION-ORIENTATED ALSO DOES ERROR RECOVERY 10.A CONNECTION IS? A:ONLY AN TEMPORARY ESTABLISHED SESSION BETWEEN SOURCE AND TARGET. 11.WHAT MODE SUPPORTS CONCURRENT CONNECTIONS BETWEEN NODES ON A LAN? A:CONNECTION-ORIENTATED. BECAUSE AN ESTABLISH SESSION IS CREATED. 12.WHICH POORLY CHOSEN PHRASE IS OFTEN USED TO DESCRIBE CONNECTION-ORIENTATED PROTOCOLS? A:"GUARANTEED DELIVERY" THIS IS A BAD PHRASE, BECAUSE DELIVERY CAN NEVER BE GUARANTEED. THE BEST YOU CAN DO IS BE LIKE FEDERAL EXPRESS. FEDEX CANNOT GUARANTEE DELIVERY OF A PACKAGE. IF IT IS NOT DELIVERED, AT LEAST THEY'LL TELL YOU WHY. LIKE A CONNECTION-ORIENTATED PROTOCOL, FEDEX GETS IT DELIVERED TO THE RECEIPIENT 99% OF THE TIME. 13.WHAT OTHER POORLY CHOSEN PHASE IS OFTEN USED TO DESCRIBE CONNECTION-LESS ORIENTATED PROTOCOLS? A:"UNRELIABLE". IN FACT A CONNECTIONLESS PROTOCOL LIKE UDP IS JUST AS RELIABLE AS A CONNECTION-ORIENTATED PROTOCOL LIKE TCP. WHEN THERE IS NO TRANSMISSION ERRORS. WHEN TO CHOOSE SLIP, AND WHEN TO CHOOSE PPP SLIP AND PPP ARE YOUR 2 MAIN DIALUP ACCESS PROTOCOLS. THERE ARE 2 TYPES OF SLIP. 1.SLIP REGULAR SLIP 2.CSLIP USES COMPRESSION. 1.PHYSICAL LAYER=SLIP 2.DATALINK LAYER =PPP PPP IS FASTER AND MORE RELIABLE THAN SLIP/CSLIP. SLIP BEFORE DIALING THE SLIP SERVER, ALL TCPIP SETTINGS MUST BE CONFIGURED CORRECTLY. SLIP DOES NOT DO ANY DHCP. SLIP DISADVANTAGES 1.SLIP HAS NO DHCP (DYNAMIC IP SUPPORT) 2.SLIP SUPPORTS ONLY ONE PROTOCOL, TCPIP. 3.SLIP HAS NO ERROR CHECKING 4.SLIP HAS NO COMPRESSION (CLISP DOES) 5.SLIP HAS NO SECURITY! YIKES! IS IT ALL CLEARTEXT? WINDOWS RAS NT USERS CAN DIALOUT USING SLIP, BUT WINDOWS NT WILL NOT ACT AS A SLIP SERVER. REPEAT! WINDOWS NT WILL NOT ACT AS A SLIP SERVER OR ACCEPT ANY INCOMING RAS SLIP CONNECTIONS. PPP POINT TO POINT PROTOCOL (PPP IS RFC 1171) PPP HAS 3 MAJOR COMPONENTS 1.ENCAPSULATING DATAGRAMS OVER A SERIAL LINK 2.EXTENDABLE LINK CONTROL PROTOCOL (LCP) 3.A SUITE OF NETWORK CONTROL PROTOCOLS FOR ESTABLISHING AND SETTING UP DIFFERENT NETWORK LAYER PROTOCOLS. PPP FIXES THE PROBLEMS THAT SLIP HAS 1.PPP SUPPORTS DHCP (DYNAMIC IP ADDRESSING) 2.PPP HAS ERROR CHECKING (ON EACH PPP FRAME) 3.PPP SUPPORTS MULTIPLE PROTOCOLS (NOT JUST TCPIP) 4.PPP PACKETS CAN BE AS LARGE AS 1508 BYTES. (THE SAME MAXIMUM SIZE AS AN ETHERNET FRAME) PRACTICE PROBLEMS 1.YOUR DIALLING INTO A RAS SERVER, BUT IT IS NOT WORKING. YOUR TCPIP SETTINGS ARE NOT BEING INITIALIZED BY THE RAS SERVER. WHAT IS POSSIBLY WRONG? A:IT IS POSSIBLE THE RAS SERVER IS A SLIP SERVER. ANYTIME A RAS SERVER HAS NO DYNAMIC IP SUPPORT, IT PROBABLY IS A SLIP SERVER. 2.YOU WANT TO CONNECT TO NOVELL AND NT SERVERS FROM HOME VIA DIALUP. NT SERVERS AT WORK USE TCPIP. NOVELL SEVERS ALL USE IPX/SPX. WHAT DIALUP PROTOCOL DO YOU USE? A:YOU MUST USE PPP, TO SUPPORT MULTIPLE PROTOCOLS. SLIP ONLY DOES TCPIP. 3.WHAT DIALUP PROTOCOL IS RELATED TO HDLC? A:PPP 4.IS PPP A BIT-ORIENTATED PROTOCOL? A:YES 5.PPP WHEN CONNECTING, PPP CHECKS THE LINE QUALITY TO SEE IF IT CAN HANDLE THE CONNECTION PROPERLY. WHAT COMPONENT OF PPP DOES THIS? A:LCP, LINE CONTROL PROTOCOL. PART OF PPP CHECKS LINE QUALITY. 6.WHAT DIALUP PROTOCOL HAS USES 2-BYTE CRC'S WITH EVERY FRAME? A:PPP. ONLY SLIP AND PPP ARE DIALUP PROTOCOLS, BUT ONLY PPP USES ERROR CHECKING. 7.WHAT IS PPTP? A:PPTP IS THE PROTOCOL THAT RUNS ON TOP OF PPP. PPTP ESTABLISHES A SECURE VIRTUAL LINK THROUGH THE INTERNET TO A REMOTE LAN. THIS IS A VPN. 8.WHICH DIALUP PROTOCOL USES THE 1.PHYSICAL LAYER? A:SLIP IS A 1.PHYSICAL LAYER PROTOCOL. 9.WHICH DIALUP PROTOCOL USES THE 2.DATALINK LAYER? A:PPP 10.ON OLDER UNIX DIALUP SERVERS, WHAT ARE YOU LIKELY TO FIND? A:SLIP SERVER 11.TODAY, WHAT IS THE MOST POPULAR DIALUP PROTOCOL? A:PPP BY A LANDSLIDE. 12.YOU WANT TO ACCESS THE OFFICE'S NT SERVER FROM HOME. YOUR HOME LAPTOP ONLY HAS NETBEUI PROTOCOL. EVERYONE AT THE COMPANY HAS ALWAYS DIALED INTO THE RELIABLE UNIX SCO BOX THAT DOES SLIP. WHAT DO YOU DO? A:PROBABLY BEST TO ADD TCPIP TO THE NT SERVER, SINCE SLIP CAN ONLY SUPPORT TCPIP. YOU'LL NEVER BE ABLE TO ACCESS THE NT SERVER IF ITS ONLY USING NETBEUI. OR YOU YOU REMOVE THE SLIP SERVICE ON THE SCO BOX AND REPLACE IT WITH PPP. 13.YOU CEO WANTS TO ACCESS THE CORPORATE EMAIL ON THE NOVELL SERVER VIA THE 1-800 NUMBER WHILE ON THE ROAD. WHAT DIALUP SERVER SHOULD YOU USE? A:SET UP A PPP RAS SERVER, THE CLIENT AS A PPP DIALUP CLIENT. 14.WHY USE DHCP? A:TO SET ALL THE TCPIP SETTINGS UP CORRECTLY EVERYTIME. DHCP "TIME-SHARES" A SMALL NUMBER OF IP'S FOR A LARGE NUMBER OF USERS. 15.YOUR UNIX MACHINE IS DIALING INTO THE CORPORATE SLIP SERVER. IT CONNECTS OKAY, BUT THE TCPIP IS STILL NOT WORKING. YOU HAVE DHCP SETTINGS SET AND INTERNET ACCESS STILL IS NOT WORKING. WHATS WRONG? A:DUMMY! SLIP SERVERS DON'T SUPPORT DHCP! YOU HAVE TO SET YOUR TCPIP SETTINGS UP MANUALLY FOR SLIP, THEN DIAL THE SLIP SERVER. 16.CAN A MULTILINK SESSION BE ESTABLISHED USING SLIP? A:NO. ONLY PPP SUPPORTS MULTLINK SESSIONS. 17.A SOLARIS ULTRASPARC AT WORK, YOU DISCOVER IS A CSLIP SERVER. WHAT IS THIS? A:CSLIP IS SLIP WITH COMPRESSION. IE CSLIP IS FASTER THAN SLIP. 18.YOUR LOOKING THOUGH SOME DIAGNOSTIC LOGS. YOU SEE DOZENS OF LOGS ABOUT A RAS SESSION USING LLC ERRORS. WHAT DIALUP SESSION IS THIS COMING FROM? A:ONLY PPP USES LLC. 19.LLC BELONGS TO WHAT OSI LAYER? A:LLC IS A SUBLAYER OF THE 2.DATALINK LAYER 20. WHICH PROTOCOL WORKS AT THE MEDIA ACCESS CONTROL (MAC) LAYER? A:PPP 21.(MAC) MEDIA ACCESS CONTROL IS A SUBLAYER OF WHAT LAYER? A:2.DATALINK LAYER 22.MOST ISDNS USE WHAT SERIAL LINK PROTOCOL? A:PPP. IF YOUR ISDNS ONLY USED SLIP/CSLIP, THEN YOU WOULD BE LIMITED TO ONLY TCPIP PROTCOL AND THE DUAL CHANNELS OF ISDN WOULD NOT WORK! SLIP/CSLIP DOES NOT SUPPORT MULTILINK (PPP DOES SUPPORT MULTILINK 23.YOUR INSTALLING RAS CLIENT SUPPORT FOR 25 MACINTOSH G4 POWERBOOK USERS. THEY'LL ALL BE USING APPLETALK. WHAT DIAL-UP PROTOCOL WILL YOU USE? A: PPP. YOUR ONLY CHOICE. 24.HOW MUCH DATA CAN A PPP PACKET HOLD? A:1500 BYTES, PLUS 8 FOR HEADER INFO FOR A TOTAL OF 1508. SAME SIZE AS AN ETHERNET FRAME. 25.WHEN DID SLIP APPEAR? A:1984. GEORGE ORWELL'S 1984. DEVICES THAT COMMUNICATE AT EACH OSI LEVEL 1.PHYSICAL LAYER NETWORK MEDIA COMMUNICATIONS 2.DATALINK LAYER TRANSLATES BETWEEN BITS FROM 1.PHYSICAL LAYER AND FRAMES FOR 3.NETWORK LAYER 3.NETWORK LAYER DETERMINES THE ROUTE PACKETS TAKE TRANSLATION BETWEEN PHYSICAL MAC AND LOGICAL IP ADDRESSES 4.TRANSPORT LAYER ENSURES PACKETS ARE DELIVERED ERROR-FREE AND IN PROPER SEQUENCE LARGE SESSION LAYER MESSAGES BROKEN DOWN INTO SMALLER PACKETS. 5.SESSION LAYER CREATES/MAINTAINS AND DELETES A SESSION BETWEEN REMOTE COMPUTERS. 6.PRESENTATION LAYER DEFINES A COMMON FORMAT FOR APPLICATIONS TO USE DATA ENCRYPTION AND COMPRESSION OCCURS IN 6.PRESENTATION LAYER 7.APPLICATION LAYER ACTS AS THE INTERFACE BETWEEN THE APPLICATION AND THE NETWORK 1.PHYSICAL LAYER INCLUDES HUBS, TRANSMITTERS, RECIEVERS, CABLES, CONNECTORS AND REPEATERS 2.DATALINK LAYER INCLUDES INTELLIGENT HUBS, BRIDGES, NICS 3.NETWORK LAYER DETERMINES THE ROUTE PACKETS TAKE TO THEIR DESTINATION ROUTERS/BROUTERS AND GATEWAYS DO THIS FUNCTION CIRCUIT/PACKET/MESSAGE SWITCHING ALSO DETERMINE THE ROUTE PACKETS TAKE GATEWAYS USUALLY RESIDE AT THE 7.APPLICATION LAYER, BUT CAN RESIDE AT THE 3.NETWORK LAYER A BROUTER IS A BRIDGE AND ROUTER THAT ACT AS A BRIDGE FOR NON-ROUTEABLE PROTOCOLS 20.71 EXERCISE YOU HAVE 7 24 PORT HUBS. YOU JUST ADDED AN 8TH 24 PORT HUB. EVERYONE SAYS THE NETWORK IS SLOW. YOUR BOSS WANTS TO UPGRADE TO 100MBIT ETHERNET TO FIX THE PROBLEM? WHAT IS THE RIGHT SOLUTION? SPEED IS NOT THE PROBLEM. IN FACT CONVERTING EVERYONE AND ALL DEVICES TO 100MBIT AND ALL CABLING TO CAT5 SPECIFICATION MAY IN FACT MAKE THE PROBLEM WORSE. THE PROBLEM IS YOU HAVE 190 DEVICES ON ONE SEGMENT ALL COLLIDING TOGETHER. THE RIGHT SOLUTION IS TO GET A BRIDGE AND SLICE THE NETWORK IN HALF. 20.7 PRACTICE PROBLEMS 1.THE PRESENTATION LAYER IS CALLED THAT BECAUSE A:IT PRESENTS A UNIFORM DATA FORMAT TO THE 7.APPLICATION LAYER 2.WHAT OSI LAYER USES CALBE AND CONNECTORS? A:1.PHYSICAL LINK LAYER 3.WHAT OSI LAYER DEFINES THE NETWORK TOPOLOGY? A:THE 1.PHYSICAL LAYER ALSO DEFINES THE NETWORK TOPOLOGY USED 4.WHAT LAYER USES BITS? (NOT FRAMES) A:1.PHYSICAL LAYER USES BITS (NOT FRAMES) 5.WHAT LAYER TAKES THAT BITSTREAM AND FORMS FRAMES? A:THE 2.DATALINK LAYER ACCEPTS THE STREAM OF BITS AND FORMS FRAMES 6.THE IEEE 802 PROJECT PROVIDED ENHANCEMENTS TO WHAT OSI LAYER? A:THE 802 PROJECT DIVIDED THE 2.DATALINK LAYER INTO MAC MEDIA ACCESS CONTROL AND LLC LOGICAL LINK CONTROL 7.THE MAC MEDIA ACCESS CONTROL IS PART OF WHAT LAYER? A:THE MAC IS PART OF THE 2.DATALINK LAYER 8.WHAT LAYER IN THE OSI DETERMINES THE ROUTE THE PACKETS WILL TAKE TO THEIR DESTINATION? A:3.NETWORK LAYER DETERMINES THE ROUTE PACKETS TAKE TO THEIR DESTINATION 9.NICS AND NIC SOFTWARE RESIDE AT THAT LAYER IN OSI? A:NICS AND NIC DRIVERS RESIDE AT 2.DATALINK LAYER 10. WHAT DEVICE CAN BRIDGE NON-ROUTEABLE PROTOCOLS? A:A BRIDGE AND A BROUTER CAN BRIDGE NON ROUTEABLE PROTOCOLS 11.WHAT DOES A GATEWAY DO? A:A GATEWAY DOES PROTOCOL CONVERSION. THIS IS MOST OFTEN DONE BY DISASSEMBLING THE PACKETS AND REFORMING THEM INTO ANOTHER FORMAT. 12.WHAT DEVICE USES 3.NETWORK LAYER? A:ROUTER MOST OFTEN IS AT 3.NETWORK LAYER 13.WHAT DEVICE USES 2.DATALINK LAYER A:BRIDGE USES 2.DATALINK LAYER 14.WHAT IS A DEVICE THAT DOES NOT USE THE 1.PHYSICAL LAYER? A:BRIDGE OPERATES AT 2.DATALINK, NOT 1.PHYSICAL 15.WHAT DEVICE SIMPLY REGENERATES THE INCOMING SIGNAL? A:REPEATERS SIMPLY REGENERATES INCOMING PACKETS, GOOD OR BAD PACKETS 16.WHAT DEVICE DOES NOT USE ROUTING TABLES? A:REPEATERS OPERATE AT 1.PHYSICAL LAYER. THEREFORE DO NOT EVER USE ROUTING TABLES 17.WHAT HAPPENS WHEN A BRIDGE IS UNABLE TO DETERMINE WHAT PORT A PACKET IS ON? A:IF A BRIDGE CANNOT DETERMINE THE PORT A PACKET IS DESTINED, IT BROADCASTS THE PACKET OUT ALL PORTS (EXCEPT THE PORT THE PACKET CAME IN ON) 18.WHAT LAYER TRANSLATES NETWORK LOGICAL ADDRESSES INTO PHYSICAL ADDRESSES? A:THE 3.NETWORK LAYER DOES THE TRANSLATION BETWEEN LOGICAL AND PHYSICAL ADDRESSES 19.WHAT LAYER DOES THE REDIRECTOR OPERATE? A:THE REDIRECTOR OPERATES AT 6.PRESENTATION LAYER 20.CIRCUIT SWITCHING PROVIDES A DEDICATED PATH AND BANDWIDTH. 21.MESSAGE SWITCHING (AKA STORE+FORWARD) TREATS EACH MESSAGE AS A SEPARATE ENTITY 22.PACKET SWITCHING BREAKS UP A LARGE MESSAGE INTO SMALLER PARTS. EACH PART GETS TO THE DESTINATION INDEPENDANTLY. 23.THE SESSION LAYER CREATES MAINTAINS AND THEN DELETES A SESSION BETWEEN 2 COMPUTERS. 24.THE PRESENTATION LAYER DOES COMPRESSION AND ENCRYPTION. 25.THE SESSION LAYER INSERTS CHECKPOINTS IN THE DATASTREAM BETWEEN APPLICATIONS, 26.THE APPLICATION LAYER PROVIDES API SYSTEMS CALLS TO THE APPLICATION. 27.THE 1.PHYSICAL LAYER IS TH ONLY OSI LAYER THAT CAN DIRECTLY TALK TO ITS PEERS ON THE NETWORK. ALL OTHER LAYERS MUST GET PROCESSED IN THE STACK. 28.THE 2.DATALINK LAYER ADDS THE CRC CHECKSUM TO EACH FRAME. 29.THE MAC (MEDIA ACCESS CONTROL) IS LOWER THAN LLC, AND CLOER TO THE NIC HARDWARE. 30.GATEWAYS ARE OFTEN USED TO INTERFACE BETWEEN PC NETWORKS AND MAINFRAMES. 31.GATEWAYS ARE NOT FAST. GATEWAYS ARE TASK SPECIFIC, EXPENSIVE AND LOSW. IT TAKES A LOT OF CPU HORSEPOWER AND RAM TO DO PROTOCOL CONVERSIONS QUICKLY. 32.GATEWAYS RESIDE USUALLY AT THE 7.APPLICATION LAYER 33.GATEWAYS CAN SPAN ALL 7 OSI LAYERS 34.BRIDGES ARE OFTEN EMPLOYED TO STOP BROADCAST STORMS. 35.NETBEUI/DLC WILL NOT WORK WITH ROUTERS SINCE THESE ARE NOT ROUTEABLE. 36.OSPF AND RIP AND NLSP ARE ROUTING PROTOCOLS. 37.THERE ARE 2 KINDS OF ROUTERS 1.STATIC 2.DYNAMIC 38.USE OF THE OSI MODEL IS ALSO CALLED THE PROTOCOL STACK. 39.THE TRANSPORT LAYER WORKS WITH IP DATAGRAMS. DEFINING THE CHARACTERISTICS AND PURPOSE OF MEDIA IN IEEE 802.2 AND 802.5 AS NETWORKING BECAME MORE POPULAR, THE IEEE STARTED THE 802 PROJECT. FEBRUARY 1980 19-802 802.1 INTERNETWORKING 802.2 LLC LOGICAL LINK CONTROL 802.3 CSMA/CD ETHERNET 802.4 TOKEN BUS 802.5 TOKENRING 802.6 MANS DEMAND PRIORITY ACCESS LAN 100BASEVG-ANYLAN 802.7 BROADBAND ADVISRORY GROUP 802.8 FIBRE OPTICS ADVISORY GROUP 802.9 VOIP INTEGRATED VOICE AND DATA NETWORKS 802.10 NETWORK SECURITY 802.11 WIRELESS NETWORKING 802.3 CSMA/CD ETHERNET Xerox designed Ethernet in 1970. In 1980, DIX (Digital Intel Xerox) began to publish Ethernet. Ethernet is a baseband system that uses CSMA/CD as the media access control. In baseband, the entire bandwidth can be used by one signal. In 1985, version II of Ethernet was released. The IEEE Project 802 chose version II as the basiss for its 802.3 standard. The only major differences between original Ethernet and Ethernet II is the packet headers. When using CSMA/CD 1.the Ethernet devices listens to the media, and waits until no traffic 2.the Ethernet device transmits is data on the media. While the data is being transmitted, the media is in use, no other Ethernet device will try and transmit. A collision can occur when 2 or more Ethernet devices transmit at the same time when the media is free. This is a collision and both Ethernet devices must transmit again. Using CSMA/CD means several limitations. 1.a maximum distance between any two nodes 2.a maximum number of repeaters between nodes 3.a maximum segment between nodes Different cables have different characteristics. Two cable items are important to Ethernet. 1.IMPEDANCE 2.PROPAGATION DELAY Propagation delay is the amount of time a signal takes to travel through a media. Propagation delay through RG58 is about 66% speed of light. If the time for a signal from end to end is longer than to transmit a frame, then collisions may not be detected!!!! Late collisions are collisions that are not detected, because the propagation delay is too long. Items like switches hubs and repeaters all contribute to propagation delay. Ethernet cabling has many forms, some of the forms are 10base2 segments upto 185 meters.(sometimes upto 200m) coften called thinnet Cabling is RG58U 50ohms 1/4" thick 10base5 segments upto 500 meters often called thicknet Cabling is RG6 10base-F segments upto 2km, using fibre optics 10Broad36 segments upto 3600 m 3.6km using analog broadband signals in a dual cable config 10base-T segments upto 100m (sometimes upto 150meters) signal loss cannot exceed 11.15 db's The 543 Rule There can be no more than 5 repeated segments No more than 4 repeaters Only 3 of the 5 segments can be populated On 10base-T networks, cascading hubs each count as a repeater. 802.25 TokenRing A tokenring network looks just like a star network, but it really is a ring. A TokenRing network is physically a star, logically a ring. 20.81 Exercise You need to network 50 computers. Each office has RG59 Coax (NOT Thinnet!) and 6 pair UTP cabling. The company wants an easy to manage network. What do you implement? A:You may be tempted to use the Thinnet cabling. However RG58 is 75 ohms, and Thinnet uses RG58U 50 ohm! The better choice is to use the installed UTP cabling. A TokenRing network would be much more expensive that an Ethernet star network. 20.8 Practice Problems 1.Xerox invented Ethernet in 1970. 2.IBM invented 4Mbit TokenRing in the mid 80's 3.IEEE 802.3 is Ethernet. 4 IEEE 802.5 is TokenRing media access control 5.Maximum of 5 segments between any 2 nodes on 802.3 Ethernet networks 6.Maximum of 4 repeaters between any 2 nodes on an 802.3 Ethernet network 7.A MAC address on an Ethernet network is a burned in unique physical 48-bit number. 8.Tokens are claimed in TokenRing by converting them to frames. 9.CSMA/CD have length restrictions because of propgation delays. 10.Broadband carry several signals on the media. Baseband carry one signal. 11.There are minor differences in the header between Ethernet and Ethernet 802.3 frames. 12.An 802.3 MAC address is 48-bits (6 bytes) 13.Three bytes (24 bits) is the OEM, 3 bytes (24-bits) is the unique MAC ID. 14.CRC is used to detect errors in transmission. 15.TokenRing is physicallly a star network, logically a ring. 16.Tokenring effiecny (not reliability) scales better as the number of computers increases. 17.185 meters is the maximum distance between 10base2 nodes. 18.500 meters is the maximum distance between 10base5 nodes. 19.10base-F is rated at 10mbits/sec, but can go much faster. 20.The hub in a TokenRing is called MAU. Multistation access unit. 21.To increase the distance of 10base-2, you should use a bridge. 22.10base2 Thinnet uses RG58 50-ohm cable. 23.Thicknet uses RG6 coax. 24.Only 1 token can exist on a IBM tokenring network. (fddi supports multiple tokens) 25.There is only 1 active monitor on a TokenRing network. Usually the device with the lowest MAC address is the active monitor. 26.Electronic LAN signals travel through the media at slower than speed of light. 27.Propogation delay is the amount of time a signal takes to travel though a media. Explaination of NDIS and ODI Network Standards In the 1980's as Networking PCs together increased, so did many different standards. Everyone had their own standard. It was a nightmare to network dissimilar equipment. NDIS and ODI provided a standard common interface. Microsoft and 3com made NDIS in 1989. NDIS made the NIC work with any NDIS compliant driver. NDIS enabled multiple network protocols to work on one NIC. NDIS enabled binding one protocol to multiple NICs. ODI Open Datalink Interface was designed by Apple and Novell in 1989. ODI has the same goals as NDIS. The goals of ODI and NDIS are to provide a seamless integration between 2.datalink and 3.network layer. Odi has 3 major parts 1.protocol stacks 2.mlids (multiple link interface drivers) 3.LSL link support layer 20.91 Exercise Your working at a company that is migrating from Novell to NT. A VP wants to get some files off the old Novell server. However the VP cannot "see" the Novell Server on the network. Whats wrong? A:Either there is no NWLink installed on the VP's NT computer, or the NWLink is not binded to the NIC properly. It could also be an improper frame type. 20.9 Practice Problems 1.Microsoft and 3com designed NDIS. 2.Appl and Novell designed ODI 3.Novell replaced dedicated IPX with ODI starting with Netware 4.0 4.The Protocol Manager handles binding in NDIS 5.Nt 3.51 does not use the Protocol Manager, it store binding info in the Registry. 6.2.Datalink Layer is where NDIS resides 7.Dedicated IPX can only support 1 network protocol. 8.MLIDs are a component of ODI. 9.PROTMAN, Protocol Manager is used by NDIS 10.NDIS routes packets round-robin. This is why when using NDIS, it is very important to have the most often used protocols binded first. 11.Both NDIS and ODI support 802.3 Ethernet frames. 12.NDIS 3.0 dos not support quality of service (whats quality of service?) 13.In NT, the binding information is held in the Registry 14.Both NDIS and ODI define a set of low level protocols that Novell and Microsoft both support. Practice Exam Standards and Terminology 1.CSMA/CD is Collision Sense Multiple Access/Collision Detection 2.CSMA/CD tries to detect collisions (not avoid collisions) 3.10base% is called thicknet 4.When using 10base2 Thinnet, you must terminate each end of the bus with 50 ohms 5.The OSI model has 7 layers 6.CAT5 cabling is rated for 100mbits 7.Bridges have better network performance than repeaters. Because bridges forward packets only to the destination port. Not out all ports. This reduces network traffic and collisions, and increases network efficiency. 8.Ethernet uses CSMA/CD 9.Novell Netware uses IPX/SPX. (Microsoft uses it's compatible NWLink) 10.There a 6 levels of RAID 0 to 5 makes 6 11.A terminal accessing a mainframe database is NOT client /server computing 12.User-level security is goof for large networks 13.A Peer2Peer network uses share-level security 14.User-level access means resources have permissions, not passwords! 15.When there is no errors, connectionless protocols are faster (UDP) When there is errors, connection-oreintated protocols are better (TCP) 16.If you want to support multiple network protocols across your remote access link, you must use PPP. SLIP only supports TCPIP 17.SLIP is commonly seen in older dialup systems 18.DHCP is uses to timeshare a few Ips with a large pool of users. 19.The OSI presenetation layer presents a uniform data format to the 7.Application layer 20.The 1.Physical Layer is the only OSI layer that can talk to its peers. All the other layers must go through the OSI stack. 21.Gateways commonly interface PC networks to mainframes 22.IEEE 802.3 descibes Ethernet 23.IEEE 802.5 descibes TokenRing 24.a MAC address is a 48-bit (6 byte) number assigned by the OEM. 3 bytes for the vendor ID, 3 bytes for the unique ID. 25.CSMA/CD networks have length restrictions because of propagation delays. For collisions to be detected, the propagation delay time MUST be less than the time it takes to trasnmit packet on the media. Chapter 21 PLANNING The most common LAN media is copper. Twisted Pair or coax. Fibre Optics are becoming more comon for high end installations. Each type of media has some important characteristics 1.COST 2.Installation requirements 3.Bandwidth 4.Baseband or Broadband 5.Attentuation 6.EMI 7.Security and Snoopability Bandwidth Bandwidth refers to the amount of data a media can carry. Data Transmission rates refer to how many bits of data can be transmitted per second. Ethernet LANS can theoretically support 10 million bits/second and a bandwidth of 10 megabits /second. Bands can be either Baseband (Digital) or Broadband (Analog) Baseband (digital) use the entire bandwidth for one signal Broadband (analog) can use multiple signals in the media. Duplexing is easier in Broadband Broadband therefore uses multiplexing to support it's many inband signals Multiplexing makes good sense in a few situations 1.When media bandwidth is expensive. Sharing a high-speed line with others makes sense. 2.When bandwidth is not being used. This means the bandwidth can be used by other tasks. 3.Trasnmitting large amounts of dat though a many low capacity channels. The data can be demuxed at the other end and combined. Attenuation Attentuation is how much a signal gets weakened. As a rule, the higher the frequency, the greater the attentuation that occurs. If attentuation gets too high, the signal becomes too weak, and indistinguishable from background noise. EMI EMI is background noise that distorts a signal in a media. The greater EMI gets, the harder it is to distinguish the signal from the noise. Fibre optic has zero EMI problems. UTP has a large amount of EMI problems. Crosstalk is a type of EMI. Crosstalk is when nearby signals mix with each other, resulting in no good signals. Cable MEDIA Coaxial Cable UTP Cabling GRADES CAT1 Voice grade and 4Mbits or less CAT2 Voice grade and 4Mbits or less CAT3 10mbits (3 twists per foot) posible to get 100mbits CAT4 4 twisted pairs upto 16mbits CAT5 4 twisted pairs rated for 100mbits IBM has their own grading of cables TYPE 1 STP 2 pairs of 22AWG TYPE 2 Voice+Data 2 pairs of 22AWG for data, 2 pairs of 26AWG for voice TYPE 3 Voice 4 solid UTP pairs 22/24 AWG TYPE 4 undefined TYPE 5 Fibre Optics 1 pair of 62.5/125 micron multimode TYPE 6 Patch cable 2 pairs of 26 AWG TYPE 7 undefined TYPE 8 Carpet Grade 2 pairs of 26AWG with shield TYPE 9 Plenum grade 2 pairs of 26AWG with shield Thinnet ThickNet STP UTP Fibre Network Topology BUS BUS STAR STAR STAR COST low HIGH moderate moderate HIGH Ease of Installation easy DIFFICULT moderate moderate DIFFICULT Speed 10mbits 10mbits 4-100mbits 4-100mbits 100mbits+ Distance 185m 500m 100m 100m 100KM EMI low low EMI HIGH EMI NO EMI Security moderate moderate moderate low HIGH Wireless Media You can devide wireless media into 3 types 1. Wireless LANS A LAN that is based entirely on wireless media. 2. Extended Local Networks A wireless connection that connects 2 LANS together. 3. Mobile computing A mobile laptop that connects to a LAN using wireless. Wireless is best used when 1.Regular cable runs are impossible or very difficult to implement. 2.People or equipment is very mobile Network administrators/doctors/nurses etc often move constantly 3.A temporary LAN must be installed 4.Business travellers Wireless with LANS Infrared Broadband Optical Uses analog broadband technology Infrared Line of Sight Transmissions must occur over a line of sight Infrared Reflective All transmissions are sent to a common hub and back to each wireless node Infrared Scattered Infrared signals are reflected off most surfaces. Due to refractions, data rates are very low Most infrared systems are limited to about 100 feet. Although once in this 100 foot llimit, bandwidth can get up to nearly 10 mbits!! Infrared is not bothered by EMI or radio sources, but other forms of light, including bright white ligths can hamper infrared. The darker the room, better, for infrared. Laser Transmissions via Infrared High power lasers can transmit data for several thousand yards in a line of sight. Lasers can be used in the same capacity as microwave links, without FCC or CRTC authorizations. For indorr use, lasers are not often used. Single Frequecy Radio (narrow band radio) Single frequency radio is better than infrared in many cases. Single Frequency Radio does not require line of sight. The radio signal can bounce off many surfaces. However concrete and buildings can block the signal. Spread Spectrum Radio Spread Spectrum radio is a technique designed by the military. Spread spectrum improves reliability, reducesinterference and jamming and less prone to snooping than a single radio frequency. Spread Spectrum uses multiple signals to transmit its data. Two Spread Spectrum techniques often used are Frquency hopping and Direct Sequence Modulation. Frequency hopping is changing diffeerent frequencies periodically. Frequency hopping is good for 2 miles outdoors, 400 feet indoors. Frequency hopping can work from 250kb to 2Mb. Direct Sequence Modulation breaks the message into blocks. Each block is transmitted on a different frequency. It is easy to add false data and use different frequencies. Wireless Bridging A wireless bridge connects to LANS together. Wireless bridges often use spread-spectrum. And can work upto 3 or 4 miles. Long range wireless bridges can work upto 25 miles! Mobile Computing There are 3 types of mobile computing 1.Packet Radio networking the mobile unit communicates network data via satelitte. 2.Cellular networking CDPD Cellular Digital Packet Data. This method uses the cellular phone network. This can provide a very decent solution. 3.Satelitte Station networking Used with microwave dishes, communicate directly with satelittes. Microwave Technology Microwave technology cn be used in all 3 wireless areas. 1.LAN (using a wireless solution on a LAN) 2.Extended LAN (connecting 2 LANS together using a wireless solution) 3.Mobile networking (connecting mobile users to a LAN) Microwave Technology has 2 types 1.Terrestial Microwaves 2/Satelitte Microwaves Terrestial Microwaves Terrestial microwaves uses transmitters and recievers that are on earth. Terrestial microwaves are line of sight. When line of sight is possible, a terrestial microwave link is a one time expense. A leased line is a cost that never goes away! Very bad weather can affact a microwave link, depending on the distance and power of the microwave. Microwave links often need to get encrypted since they are in the public domain. However it would take some very sophisticated hackers to hack onto a microwave link. Satelitte Microwaves Use satelittes at 22,300 miles above the earth in geo-synchrnous orbit. Because the range is global, this permits the widest range. Because it takes 22,300 miles up and 22,300 miles down the distance from you to next door is the same as if you were transmitting on the other side of the world. The propogation delay in satelitte microwaves can be .5 to 5 seconds long. 21.1 Practice Problems 1.The primary limitation on long cable runs is attenuation. The loss of signal power is in db's 2.Baseband is a method of trasnmitting only one signal through a media 3.Multiplexing is used on 56k, T1 and T3 lines. A T1 is 24 multiplexed DS0's. A T3 is 24 multiplexed T1's. 4.When using TDM, Time Division Multiplexing, the bandwidth is determined by the MUX's configuration. 5.Snchronous TDM is using fixed time intervals for each channel. 6.Stat TDM is more efficent than TDM for line utilization. 7.Most LAN cabling is made from copper. 8. 9.All LANS Ethernet, TokenRing FDDI, 10base-T, 10base-% 10base-2 all use BASEBAND! 10.Crosstalk occurs from nearby wires. EMI can occur from crosstalk or nearby EMI sources. 11.BNC connectors are used on thinnet networks 12.Thicknet cabling is larger because the inner copper core is bigger. 13.Thicknet has a much longer distance (500m) than does thinnet (185m) 14.A 50 ohm terminator is required at each end of a Thinnet bus. 15.Thicknet cable segments are connected with N-conectors 16.Vampire taps are used to tap into Thicknet cable trunks 17.The outer shield on coax serves as a ground and shield. To balance the signal. 18.Thinnet cable lengths are a maximum of 185m 19.Thicknet cable lengths are a maximum of 500m 20.Thicknet and Thinnet both have the same bandwidth and speed. Thicknet can go for a longer distance. 21.CAT5 cabling requires 8 wires (4 pairs) 22.Florecent lights can generate enough EMI to damage signals in UTP cabling. So avoid florsecent lights as much as possible when installing UTP/STP or any copper cabling. 23.The twisting of the pairs is important to keep crosstalk at a minimum. 24.STP uses the shielding for grounding and shielding. The shielding must be properly grounded in order for STP to be effective. 25.UTP is rated upto 100mbits (CAT5) 26.Twisted pair is most common for LANS because it is cheap, easy to install and very compatible with networking components and devices. 27.IBM's Tokenring 4MB networks use TYPE 1 STP cabling. This is the most common STP network. 28.STP is a thicker cable and is more expensive than UTP and more difficult to work with. 29.UTP cable lengths must be 100m or less (you can get away with more, but don't bet on it) 30.UTP is more suspectible to EMI because it has no shielding. 31.The 2 strands in fibre optics are for receiving and sending. 32.Signals loss in fibre optics occurs mostly at the connectors. 33.Fibre optics theoretically offer the highest transmission rates of any media. 34.In order for data to be send down fibre optics, it must be converted into light and back again. 35.Fibre optics, while much more expensive, offers top security an no EMI or spark hazard. 36.A multimode fibre optics cable at 10mbit/sec ehternet would be at most 2km. 37.The 2 major drawbacks to fibre topics are its cost and the expertise in its installation. The connectors must be spliced near perfectly. 38.There are 2 types of fibre optics. Single mode, and multi mode. Single mode allows only 1 light frequency. Multimode has many frequencies. 39.When using fibre optics, the fastest data rates are made with lasers. 40.Fibre optics are by far the most expensive. And not very durable. And require expertise installers and very expensive networking hardware. 41.A LAN that includes wireless is called a hybrid. 42.Wireless computers connect to the LAN via a special HUB access point. 43.Wireless networks employ LASERS. Infrareds and RADIO. 44.Scatter Infrared has the shortest distance and slowest speeds of any infrared system. 45.RADIO LASER and Infrared are the most common wireless LAN solutions. Satelitte and microwaves are used less often due to cost. 46.LASER wireless systems does not require CRTC or FCC. Radio or Microwave may require a special license. 47.Using satelittes means an average of 22,300x2=45,000 miles. This long delay makes using satelittes a poor choice when timing is critical. 48.Satelitte communications are microwave transmissions. 49.Of all the wireless ystems, Infrared, LASER, RADIO, Spread Spectrum. Spread Spectrum is the most secure. The military had to make it like the USS Enterprises's command of "modulate shield frequencies" 50. When planning a wireless line of sight solution, remember weather resistance, data speeds, and the technology you will be using. 21.2 Selecting the Right Network Topology for TokenRing and Ethernet Networks. Media access methods (network arbitration) have 2 kinds 1.First come first serve (CSMA/CD) (contention and competition occur) 2.Take turns (token ring) no contention or competition) Contention On contention-based networks, devices all have equal priority. Any device can transmit anytime the media is free. When multiple devices transmit at the same time, a collision occurs. When a network gets very busy, most transmits result in collisions, resulting in very little work getting done. Contention Features 1.Most popular form of media access on LANS (Ethernet uses contention) 2.Contenetion is simple 3.Contenetion works well until traficc levels are 30%+ of the bandwidth 4.Good performance at a low cost 5.A devices chances of being able to transmit cannot be predicted 6.Collisions can occur at any time DEMAND PRIORITY Demand priority is a new media access method used with 100VG-AnyLAN. Demand Priority is different than CSMA/CD. Contention occurs at the hubs for 100VG-AnyLAN 100VGAnyLAN uses 4 pairs (8 wires) called quartet signalling Token Passing Token Passing uses a frame called a token. Token passing can uses station prioriities and other methods to prevent one device from hogging the network. Each device has equal channce to transmit whenever a free token makes the rounds. Systems that using Token Passing Access Control (not CSMA/CD or Demand Priority) Tokenring 802.5 The most common tokenpassing network, defined as 802.5 TokenBus 802.4 A rare Token implemented in a bus topology. Defined as 802.4 FDDI 100mbit dual counter-rotating rings using fibre-optics. Uses multiple tokens. Comparing Contention(Ethernet CSMA/CD) and TokenPassing (FDDI/TokenRing) Tokenpassing is better than Contention for 1.Timing critical data systems Tokenpassing has more predictable delivery, that does contention 2.Very busy networks Tokenpassing networks do not get bogged down as the network gets busier. Contenetion networks have more car crashes, to a point where the network is unuseable. 3.TokenPassing networks can assign priorities Networks that are not busy, Contention is a better system! Physical Topologies Describes the layout of the cabling Logical Topology Describe the pathway the signals flow in the network BUS TOPOLOGY A bus topology is where all the devices are connected to a central backbone. The bus topology is well suited to contention. (bus works well with CSMA/CD) RING TOPOLOGY A ring topology is wired in a big ring. Each device in the ring acts as a reciever and transmitter. Because the signal is regenerated at each device, signal loss is low. It is important to remember the signal gets regenerated at each device! Rings are well suited to tokenpassing. STAR TOPOLOGY Star networks form trees. Ethernet frames can be from 64 to 1518 bytes long. 18 bytes are used for the Ethernwet header. The 543 Rules applies to all Ethernet networks (Star BUS or STAR/BUS) 1.max of 5 cables segments in a series 2.max of 4 repeaters 3.max of 3 populated segments (only applicable to coax), since UTP+fibre are point2point 10base2 Requirements 1.minimum distance between 10base2 devices is .5m 1.5 feet 2.maximum distance of the segment be 185m or less 3.entire network with repeaters cannot exceed 925m (3035feet) 4.maximum devices per segment is only 30!!! 5.each end of the bus be terminated with 50 ohms 10base5 Requirements 1.miniumum distance between devices is 2.5m (8 feet) 2.trancievers to be installed at multiples of 2.5m (marks on the cable) 3.maximum of 500m (1640 feet) per segment 4.entire network maximum of 2500m (8200 feet) 5.trunk to device can be as short as you want, but no more than 50 feet 6.maximum of 100 devices per segment 10Base-T 10Base-T is always wired in a STAR topology The central hub is a multiport repeater Physically it is a STAR, but logically functions as a BUS. 10base-T STAR has many advantages 1.network is more reliable and easier to manage one bad node does not affect anyone else, makes diagnosis very easy You can add and remove entire branches of the tree. This is very flexible You can add sections to your LAN easily 10base-T is cheap 10base-T networks are very easy to repair and fix for administrators 10base-T requirements 1.maximum devices in a collision domain is 1024 (in realworld, is much less than 1024) 2.cabling must be UTP CAT 3/4/5 3.maximum distance is 100m from device to hub 10base-FL 10base-FL is 10mbit Ethernet over Fibre optics. 10base-FL can support up to 2000m 10base-FL has zero EMI or spark hazard 100VG-AnyLAN (IEEE 802.12) 100VGAnyLAN 802.12 is a standard for combining Ethernet and Tokenring packets. 100VGAnyLAN is sometimes called 100Base-VG. (VG means voice-grade) 100VGAnyLAN has demand priority access. 100VGAnyLAN uses a cascaded star topology. Parent and Child hubs. In normal star networks, all hubs are peer2peer. 100Base-X means on of 3 types 100Base-X is often called FAST ETHERNET. 100Base-TX 2 pairs (4 wires) of CAT5 100Base-T4 4 pairs (8 wires) of CAT5 100Base-FX Fibre Optics using 2 strand fibre optics TokenRing TokenRing uses a tokenpassing system. Toeknring is Ieee 802.5 (802.4 is TokenBus) 4Mbit TokenRing devices can only operate and run in a 4Mbit mode. 16Mbit Tokenring can operate either in 4 or 16mbit modes All devices in a TokenRing must be running at the same speed. It is not possible to have some devices at 4mbit speeds and some at 16mbits. Either all 4mbits or all 16mbits speed. 21.2 Practice Problems 1.In CSMA/CD networks, access to the media is probabilistic. Meaning access to the media depends of how busy the network is. 2.Contention-based networks like Ethernet efficently falls as the load increases. . Tokenpassing networks operate very efficently under high loads. 3.The most common access method is contention-based Ethernet. 4.Tokenpassing is a DETERMINISTIC access method. Each device is guaranteed access. Contenrtion Ethernet is probablitics. It probably the device will get access. Not guaranteed. 5.However, uner light loads, contention Ethernet is more efficient than tokenpassing! 6.As the network load increases, tokenpassing networks perform better than contenetion Ethernet. 7.TokenRing networks are physically a STAR topology, but logically a RING 8.Star Topologies can be used with LOGICAL BUS'S and LOGICAL RINGS. 9. a Central hub is required when using a PHYSICAL STAR topology. 10.Physical star topology lets you do LOGICAL BUS/LOGICAL RING/CSMA/CD/tokenpassing 11.An Ethernet network is called probablistic. It is very probable each device will get access to the network, but not guaranteed. 12.10base2 is the best network to install for very small networks under 185meters. 13.If you have 2 Ethernet repeaters 200 meters apart, you could proably get away with high quality thinnet coax, but you should try ThickNet to be sure. 14.For large computer networks, always use UTP. A Large bus network based on thinnet or Thicknet would be very hard to install and administer. 15.UTP star network is very flexible. If the office environment changes, or needs to be reorganized, a star network can easily be rerouted. A thinnet bus or thicknet would be very difficult to change once installed. 16.A new building a new office is moving into has CAT3 cabling. You should use the CAT3 to make a 10mbit Ethernet star 10base-T network. 17.If you need to upgrade a network from 10 to 100mbits, install 10base-T STAR network, using CAT5 cabling. 10 and 100mbit devices can both work together. 18. If you need to connect to LANS together over a 500m, you should try and use 10base-FL. If the LANS are large, or expected to grow in the future, you should install optical fibre. A Microwave terrestial link or other wireless solution is also possible. If it is not possible to run a cable, then a wireless solution maybe your only option. LASER/Microwave Terrestial/Infrared/RADIO 19.If you have CAT5 cabling, you can either go with 10mbit Ethernet or FAST ETHERNET 100base-X. 20.You diagnosing a network and find they have 4 repeaters. The 543 rule allows 4 repeaters ONLY is 3 of the 5 segments are populated. 2 segments must not be populated! 21.All tokenring networks a physically STARs, logically RINGS 22.The TokenRing MSAU lets TokenRing devices dynamically leave and join the TokenRing. 23.The standard IBM tokenring speed are 4mbit and 16mbits. 24.A beacon on a Tokenring network indicates a problem with a nearby device. 25.When connecting TokenRing MSAUs RO Ring out goes into RI Ring IN of the other MSAU. 26.The classic IBM TokenRing using 8228 MSAUs and TYPE 1 STP cabling allows 260 computers on that ring. 27.The IBM TokenRing using 8228 MSAUs and TYPE 3 UTP cabling allows 72 computer on that ring. 28.The ACTIVE MONITORs job in a TokenRing network is to create the token. 29.On TokenRings, Beacons occur when a device joins or leaves the ring. 30.TokenRing is deterministic. Each device is guaranteed to get access to the network (token) 21.3 Selecting the Right Network Protocols for TokenRing and Ethernet Networks The OSI model is rather new and not all netyworking protocols (the older ones) do not conform very well to the OSI model. Basically there are 3 major parts to a packet. 1.HEADER start of the packet, contains all the important information like source and destination addreses. Package number etc. 2.DATA the DATA section contains the actual data 3.TRAILER the trailer marks the end of the packet and contains error checking data like CRCs. Is is important to remember that every device has to look at every packet that arrives, to check to see if that packet is for that device. If so, it gets processed. If not, that packet is ignored. NDIS lets all the ntworking protocols works with the lower layers. NDIS wrapper/NDIS interface/NDIS drivers let TCPIP/NetBEUI/Appletalk/DLC all interact with the lower layers. NDIS makes this happen. The TDI Transport Driver Interface is an interface that keeps SERVER SERVICE/REDIRECTOR/and FILE SYSTEM DRIVERS independent of the network protocol being used. 7.APPLICATION REDIRECTORS SERVER 6.PRESENTATION TDI INTERFACE 5.SESSION TCPIP NWLINK NETBEUI APPLETALK DLC 4.TRANSPORT NDIS 3.NETWORK NDIS WRAPPER NDIS NIC-DRIVERS 2.DATALINK NDIS WRAPPER NDIS NIC-DRIVERS 1.PHYSICAL PHYSICAL INTERNET PROTOCOLS (TCPIP) 7.APPLICATION FTP TELNET SMTP NFS 6.PRESENTATION FTP TELNET SMTP NFS 5.SESSION FTP TELNET SMTP NFS 4.TRANSPORT TCP UDP DNS 3.NETWORK ICMP IP ARP 2.DATALINK 1.PHYSICAL Department of Defense DODs Model of TCPIP DOD maps to OSI layers 4. Process/Application 5.session 6.presentation 7.application 3. Host 2 Host 4.transport 2. Internet 3.network 1.Network 2.datalink 1.physical IP Internet Protocol A connectionless protocol at the 3.network layer IP is mostly addressing RIP Routing Information Protocol TCPIP RIP and Novell RIP are 2 separate things, but they both do the same job, mostly. IP RIP does it route discovery by DISTANCE VECTOR. OSPF Open Shortest Path First OSPF is a LINK STATE protocol that overcomes many limitations of IP RIP. OSPF learns about changes in the network much faster than RIP and uses less bandwidth. TCP Tranmission Control Protocol A connection-oreientated protocol. TCP works at 4.transport layer! TCP and UDP both operate at 4.transport layer!! TCP has error checking features and retranmissions. This overhead makes TCP slower than UDP. UDP User Datagram Protocol A connectionless protocol. UDP does not do error checking or retransmissions. UDP only moves datagrams. If they get there great, if not too bad. ARP Address Resolution Protocol Given an IP address, the ARP protocol can find its hardware MAC address, if that computer is LOCAL. DNS Domain Name Service Provides name resolution for TCPIP devices. Given a HOST name will find its IP address. FTP File Transfer Protocol Provides file transfers for any TCPIP computer platform. SMTP Simple Mail Transfer Protocol Used to move mail through the Internet. POP and IMAPI are only to download email. SMTP is the only protocol that can move mail throught the Internet, from system to system. TELNET A TCPIP application that lets you connect as a terminal to a remote TCPIP server. NFS Network File System File sharing system for TCPIP. NFS was designed by SUN. NETWARE IPX/SPX Novell Netware protocols IPX/SPX/RIP were designed to be very modular. NT computers uses NWLink to communicte with Netware resources. NWLink and IPX/SPX are smaller and faster than TCPIP. Like TCPIP, NWLINK nd IPX/SPX are routeable. THIS MAKES NWLINK IP/SPX THE BEST CHOICE FOR SMALL NETWORKS THAT DO NOT NEED THE INTERNET. How Netware protocols map to the OSI model 7.Application NCP 6.Presentation NCP 5.Session Named Pipes and NETBIOS 4.Transport SPX 3.Network IPX NIC drivers 2.Datalink ODI and NDIS 1.Physical Physical IPX Internetwork Packet Exchange Operates at 3.Network layer. IPX is Novells connectionless datagram protocol. IPX is much easier to manage than TCPIP and does most of the same job. IPX is like IP. IPX has no error retransmissions and no error correcting. SPX Sequenced Packet Exhange Operates at 4.Transport layer. SPX is like TCP. SPX is a connection-orientated with a reliably delivery system. SPX has retransmissions and error correcting. NCP Netware Core Protocol A high level network protocol, like SMB. NCP is used at Session Presenetation and Application layer. NETBEUI NetBEUI is a network protocol that is an extension of NETBIOS. NetBEUI has limited abilities in todays networks, because it is not routeable. However on the plus side, NetBEUI is still the smallest and fastest network protocol in small networks. Thanks to NDIS, NetBEUI can coexist with other network protocols. Since NETBEUI is not routeable, you must use a BRIDGE SWITCH or a ROUTER with BRIDGING ENABLED. APPLETALK Appletalk is the network architecture designed by Apple Computer for their Macintosh systems. 7.Application Appleshare ASIP 6.Presentation Appleshare ASIP AFP 5.Session ADSP ZIP PAP ASP 4.Transport ATP NBP RTMP 3.Network DDP 2.Datalink AARP Localtalk 1.Physical Ethertalk Tokentalk Localtalk Apple DDP Apples Datagram Delivery Protocol A connectionless network protocol between Appletalk computers, operates at 3.Network layer ATP A connectionless network protocol that provides error checking and retransmissions. Operates at 4.Transport layer AFP Remote and Local File Services. Provides filesystem security. Encrption of logon names and passwords during logons. Operates high at 6.Presentation and 7. Application Appleshare Apples equivalent to SMB. DLC Datalink Control In Window networks DLC is often used by HP JetDirect network printers. DLC was often used to connect IBM Mainframes via 802.2 LLC type 2. Basic Functions of Major Network Protocols DLC Not routeable Used for HP JetDirect and IBM MainFrame connections Should only be used when absolutely required. Use HP's JetAdmin or TCPIP for JetDirect today. Appletalk ROUTEABLE Used on Macintosh computers Should be used only when Macintosh's are present IPX ROUTEABLE Used for Novell Netware networks. Can be used from small to medium sized networks. IPX/SPX should be the main choice for a small to medium network when INTERNET or TCPIP is not required. TCPIP ROUTEABLE Use for medium to large networks. Use for WANS Use when Internet access is required NETBEUI NOT ROUTEABLE Use only for small networks Is the fastest of all the network protocols Cannot be used to access the Internet or WANS NFS ROUTEABLE Unix filesharing Use only when required on UNIX servers Is more complex than SMB SMB Routeable and Not routeable (depends on what network protocol SMB is running under) Use for Windows file and print sharing Use in a Windows PC network Exercise 21.31 Because NetBEUI is not routeable, is is suitable for only small networks. This means the entire NETBEUI network must be in the same broadcast domain. If all the NETBEUI clients are not in the same broadcast domain, that means their in different networks, are since Netbeui is not routeable, not able to "see" or communicate with each other. So your back to square one, if your using NetBeui, they all have to be in one big broadcast domain. The only problem with all NetBEUI in the same broadcast domain, is that EXCESSIVE NETBEUI BROADCASTS cause too much network traffic and collisions start to occur. The Network becomes unusable when too many NETBEUI users are in the same network. The is the long way of saying, NETBEUI does not scale well. IPX scales well, but TCPIP scales very well, best of all the network protocols. That is to say, as your network grows in size, TCPIP is your best choice, then IPX. 21.3 Practice Problems 1. If you have a large network with WANS, TCPIP should be your network protocol. If you have a medium network with WANS without Internet , look at IPX/SPX. If you have a small network (under 25), look at NETBEUI. 2. If your network needs access to the Internet, you should use TCPIP. 3. If you need to reconfigure the NT Server for Macintosh support, you must install SFM and Appletalk protocol. 4. TCP provides for error correction and UDP does not. TCP is faster in a error-prone environment. UDP is faster in an error-free environment. 5. NFS is a network protocol designed by SUN Microsystems for multiplatform filesharing. 6. NT Server uses SMB as its filesharing protocol. TCPIP NwLink DLC Appletalk are NETWORK protocols. 7. Today, the HP JetDirect is easiest to setup with HP JetAdmin using TCPIP. However DLC can still be used, but to use DLC you must get the MAC address of that printer. Since DLC is not routeable, this printer is reachable only by clients on the same segment. Therefore your much better off using JetAdmin and TCPIP (which of course is routeable) 8. The fact that your using UNIX servers, makes TCPIP the choice by default. SNMP can be used by other netqork protocols, by SNMP is used most often on TCPIP networks. 9. ARP (Address Resolution Protocol) finds the MAC address. 10. NFS and SMB both do filesharing. NFS does filesharing mostly for unix. SMB does filesharig mostly for Windows networks. 11. Novells IPX/SPX and Microsofts 100% compatible NWLink are both routeabe protocols. 12. When setting up a unix server to support filesharing, NFS will most likely be used. However if this unix server will be a fileserver for a PC network, SAMBA should be used. An NT server will make a better Windows server than a UNIX Samba server (until they work out the SAMBA bugs) 13. There is a 200 client network Window 95 users. Some TCPIP, some NetBEUI. Due to too many collisions, the network performance is bad. If you install a bridge, 14. There are 200 Window 95 clients all using TCPIP. You need to add 100 more users, but broadcast storms are a problem. You should install a router. Routers stop broadcast storms. 15. When designing a WAN, especially a large WAN, TCPIP should almost always be your choice. 16. A small company that uses Internet related products asks you to setup their network. You should use TCPIP, since they'll be using the Internet and internet-related equipment. 17. Windows networks use the SMB filesharing protocol. 18. When connecting NT computers to an IBM mainframe, use DLC. It maybe possible to use TCPIP or make a gateway if DLC is not supported. 19. A unix to NT printer sharing system needs to be setup. The NT server will have to get TCPIP installed on it, so the unix and NT servers can communicate. 20. If you must choose one network protocol, and you have all Netware servers, then choose NwLink IPX/SPX. 21. NDIS and ODI both do the same job. However NDIS and ODI are NOT compatible with each other. They allow multiple network protocols and NICS. 22. A socket is a application/port combination. TCPIP uses port numbers to determine the session of that application. 23. Header Data and Trailers are part of a network packet. PREAMBLE is part of an ETHERNET frame. 24. OSPF and RIP can both be used on routers to subnet networks. 25. Converge is how fast a network learns about network changes. RIP learns slowly about changes in the network. OSPF learns very quickly about changes in the network. 26. NCP os Novells filesharing protocol. SMB is Microsofts filesharing protocol. NCP and SMB do the same job, SMB on Windows, NCP on Novell networks. 27. If an application needs a reliable IP transport protocol with error recovery, then uses TCP. 28. If an application needs a transport protocol with no error recovery, then use UDP. 29. If your NT servers need to talk with Netware servers, make sure to install NWLink, and set the frame types up correctly! 30. SPX=TCP IPX=UDP NCP=SMB ODI and NDIS do the same job but not comatible 31. RIP is distance vector. (Think RIP DV) OPSF is Link State, Think (OPSF LS) OPSF LS converge faster than RIP-DV. However RIP-DV are easier to configure. RIP-DV converge slower waste more bandwidth and easier to configure. OPSF-LS converge faster use less bandwidth and more complex to configure. 32. OPSF-LS use less bandwidth, converge faster, and more complex to configure. 33. RIP-DV uses more bandwidth, converge slower, and easier to configure. 34. Using a web browser on the Internet uses ARP to find the default route out of your network. TCP for the actual internet network connection. DNS to resolve the DNS name to its IP. 35. If your routers are taking too long to converge (ie. Taking too long to learn about changes in the network) Changing from RIP-DV to OPSF-LS can help a lot. 36. FTP is uses to transfer files upload/download files over the internet. 37. If you have no need for DLC remove it. 38. To view the configuration or data or configure routers remotely, use TELNET. 39. Macintosh computers need SFM and Appletalk and a NTFS partition! 21.4 Selecting the Right Devices for TokenRing and Ethernet Networks Connectivity devices can often mean REPEATERS BRIDGES SWITCHES ROUTERS BROUTERS and GATEWAYS. 1.PHYSICAL REPEATERS amplifies the incoming signal operate at 1.PHYSICAL LAYER no filtering at all. Simply amplifiy the signal (good and bad packets) Repeaters amplify any and all signals that it recieves Repeaters are cheap (mostly) REPEATERS CANNOT CONNECT NETWORKS WITH DIFFERENT FRAMES. A REPEATER CANNOT CONNECT A TOKENRING AND ETHERNET NETWORK TOGETHER. REPEATERS CAN CONNECT NETWORKS WITH THE SAME FRAME TYPE BUT DIFFERENT CABLING!! AS EXAMPLE, A REPEATER CAN CONNECT 2 ETHERNET NETWORKS TOGETHER ONE FROM 10BASE-5 ONE ON 10BASE-T. All networks have limitations on size. The important reason of a limit on size is due to a PHYSICS problem called PROPOGATION DELAY. A signal must always bounce back faster, from the farthest end of the network and return than the time to transmit a signal. Otherwise network errors and collisions could not be detected. HUBS ARE MULTIPORT REPEATERS. 2.DATALINK BRIDGES bridges operate at 2.datalink layer MAC layer bridges can be used to extend a network, since it joins network segments together A REPEATER PASSES ALL SIGNALS A HUB IS A MULTIPORT REPEATER A BRIDGES ON PASSES PACKETS IF IT BELONGS ON THE OTHER SIDE OF THE BRIDGE. IF THE BRIDGE DOES NOT KNOW WHERE THE PACKET BELONGS, IT GETS BROADCASTED. A BRIDGE BUILDS UP A TABLE OF MAC ADDRESSES AND LOOKS AT EACH PACKET THAT ARRIVES EACH EACH SIDE OF THE BRIDGE. Todays bridges are all LEARNING bridges. That is they keep the MAC tables upto date on their own. Really old bridges, the poor sysadmin had to manually build the MAC tables by hand!! Learning bridges also know when devices join and leave the network. BRIDGES DO 2 MAJOR THINGS 1. DIVIDE BUSY NETWORKS INTO SMALLER SEGMENTS 2. EXTEND THE LENGTH OF AND SIZE OF A NETWORK BRIDGES DIVIDE BUSY NETWORKS INTO SMALLER SEGMENTS A bridge should be is placed where the most overall broadcast packets can be sliced in half. A bridge makes smaller segments that are more efficient. Bridges have limitations that are significant in complex situations. Bridges do not use redundant paths like routers do! Bridges cannot determine the best route to take BRIDGES DO NOT STOP BROADCAST STORMS. ROUTERS STOP BROADCAST STORMS. Bridges cannot join different network types together like Ethernet and Tokenring, since bridges rely on 2.Datalink MAC layer. 2.DATALINK SWITCHES A switch is a multiport hub. Switches do not stop broadcast storms! People use switches to reduce network contention, thus improving network performance. ROUTERS STOP STORMS! BRIDGES DO NOT STOP STORMS SWITCHES DO NOT STOP STORMS 3.Network Layer ROUTERS Each network is logically seperated from other networks by routers. Routers assign each logical network and address, plus each router interface has an ip address. Routers are smarter than bridges. Routers build a table of networks and use RIP-DV or OPSF-LS to determine the best route to take. Even if the destination is remote, the router can determine the best route to take. Routers can be used to divide networks into smaller segments like bridges can. However routers can connect networks of different types (something bridges cannot do) There are 2 kinds of routers. 1. STATIC ROUTERS Do not determine routes dynamically. Instead you must configure routes manually. If the network connections change, you must also change the static route tables. Obviously static routers cannot adjust automatically to any routing or network changes. Static routers do not select the best route to take. They are pre-programmed. 2.DYNAMIC ROUTERS These routers do have the ability to determine routes automatically. DYNAMIC routers also have the ability to select the best route to take. Routers on the network talk to each other and share routing information and thus are able to make the best routing decisions. Dynamic Routers use 2 methods to discover routes. DV-Distance Vectoring. LS-Link State. ROUTERS CONNECT DIFFERENT NETWORKS USING THE SAME NETWORK PROTOCOL!! Any subnetting requires a router or brouter. CHARACTERISTICS OF A ROUTER 1. Routers connect networks of different media together 2. Routers connect different networks together 3. Routers stop broadcast storms. (Storms do not pass a router) 4. Routers only work on routeable protocols (nonrouteable packets don't pass and are ignored) 5. Routers often connect LANS to WANS. 6. There are 2 types of routers, STATIC and DYNAMIC DV-Distance Vector (RIP) Routing table information is broadcasted repeated between routers All known routes are broadcasted All DV routers can take several minutes before getting converged DV routers are easy to setup DV routers waste network bandwidth LS-Link State (OPSF) Routing table information is sent only when a change occurs Only the routes for directly connected networks or subnets is sent All LS routers can usually get converged in a few seconds LS routers are more complex, not as easy to setup and maintain LS routers do not waste network bandwidth, even in a large network BROUTERS A brouter is a bridge and router 21.4 Practice Problems 1.To connect a TokenRing and Ethernet network, use a Router. 2. To connect 2 computers 180 meters apart using 10base-T (more than 10base-Ts limit of 100m) the cheapest solutionis a repeater. 3. A HUB is a multiport repeater. 4. The cheapest solution to connect or extend a network without packet filtering is a repeater. 5. Installing too many repeaters (543 rule) will make a lot of Ethernet errors because the network length or size has probably been exceeded. 6. If you cannot add another repeater to your Ethernet network, you should use a bridge. This creates a new network at each side! 7. A router is used to connect different networks or subnets together. 8. To connect TokenRing to Ethernet, you must use a router or brouter. 9. For maximum network reliability, you must use LS routers. Link State routers can dynamically take advantage of the best routes to take. A static route will not. 10. Both LS and DV routers decide the best route to take, an LS router just does the job better. 11. The cheapest solution to connect the same coaxial connector is a barrel connector , if within 185m limitation. If over 185m limits, use a repeater. 12. A repeaters retiming feature somehow helps to remove background electrical noise. 13. A repeater is faster than a Bridge Router or Gateway. 14. A 100+ NetBEUI network traffic problem can be resolved with a Bridge or Switch. 15. Routers are used to connect any subnet. Ethernet Tokenring ATM etc. 16. Bridges work at 2.Datslink layer and 1.Physical. 17. na 18. A switch is the good network performance since it is faster than a router and only sends packets to the destination port like a multiport bridge. 19. If you have NwLink IPX clients that need to connect to a LLC Type 2 IBM Mainframe, a Gateway will be required. A Gateway is required since there is protocol conversion involved. 20. Routers stop broadcast storms. Why? Because they segment networks. 21. Gateways are the only devices to do protocol conversions. 22. If you need to extend a network farther and repeaters are exhausted, then use bridges. A router cannot be used to extend a network, a router segments networks. 23. Routers use LAN/WAN bandwidth more efficiently than do other network devices. 24. Routers, especially LS-Link State routers use the best route currently available on a network. 25. Gateways must be used to connect different protocols together. 26. Repeaters clean up the electrical noise and otherwise don't modify the signal. 27. You have 200 NWLink IPX and NETBEUI clients. Broadcast storms are a problem. Installing a new 100baseX network and cabling is not going to change or fix the broadcasting problem. 28. You have 200 NWLink IPX and NETBEUI clients. Broadcast storms are a problem. You need to add 100 more clients. A good solution might be to remove NetBEUI and convert everyone to NWLINK IPX. Then use a Netware SERVER or NT SERVER as an IPX router that segments the network and cuts the broadcast storms in half. 29. A repeater can connect different Ethernet media together! However they must both be the same media access method. Ethernet to Ethernet Tokenring to Tokenring. 30. Your network has Ethernet TokenRing and FDDI. A router can join all these networks together. 31. Bridges pass packets based on the MAC hardware address in the packet. 32. Routers pass packets based on logical network addresess in the packet. 33. To reduce broadcast storms and traffic you decide to subnet. To subnet a network requires a router. 34. Only a router will connect you to the internet. Your ISP has the router at their location. 35. RIP and OSPF will converge a network. OSPF will converge faster. NT Servers are often setup with STATIC ROUTES. Therefore almost all NT servers will not respond to any change in the network routes. The routing tables on the NT server will have to be configure it manually. 36. A switch is a mlutiport bridge. 37. Firewalls should use static routes. Firewalls should not change their routing tables dynamically. 38. RIP-DV generates more traffic since the entire routing tables must be broadcasted periodically. 39. Gateways require different protocols. 40. Too much broadcast traffic can be fixed with a router. Since routers stop broadcast storms. Features and Choosing the right WAN Service PSTN The Bell Public Service Telephone network has 2 types of service 1.Dialup Service Pays on a per use basis. Client has to exclusive access to any path 2.LEASED DEDICATED SERVICE Client is given a dedicated service LEASED DEDICATED T1 Digital point2point line. 24 B channels over 2 pairs (4 wires) 1 pair for sending. 1 pair for receiving. 1.544Mbits in both directions. DS-1 is a full T1 line. LEASED DEDICATED T3 Digital point2point line. 45Mbits LEASED DEDICATED FT1/FT3 FT1 is a fractional part of a T1 line. FT3 is a fractional part of a T3 line. LEASED DEDICATED DDS Digital Data Service DDS is slow speed for SNA connections. 2400/4800/9600/56000 LEASED DEDICATED SWITCHED 56 Full duplex wide area digital line 56kbits on a dialup service PACKET SERVICES Many companies must be able to communicate between several locations. Leased dedicated systems line T-lines or DDS or Switched 56 can be too expensive for mlutiple locations. A leased line system is for permanent connections, that do not move around. Packet-Routing service are X.25 FrameRelay ISDN and ATM The service provider of these services maintains the routing of your packets. Many subscribers use packet-routing since it is shared among them all. With leased line, only you aand you alone get the entire bandwidth. With packet-routing, you must share the bandwidth with other subscribers. The bandwidth on packet-routing networks is not guaranteed like it is on a leased line. SVC, SWITCHED VIRTUAL CIRCUIT Switched virtual circuit creates a specific path for that session. A new session may create a different path. packet-routing uses a specific route for the packets. Virtual Circuit packet-switching therefore has better error-checking and flow control. PVC, PERMANENT VIRTUAL CIRCUIT Permanent virtual circuit has a permanent static route through the packet-switching network. X.25 X.25 is a packet switching network used by really old original WANS. X.25 was the first real WAN. X.25 can use PVC or SVC. X.25 must do all its own error checking and flow control, and this makes X.25 slow. Real slow. Maximum of about 64kbits/sec. X.25 is not suitable for todays 1Mbits/sec speeds and up. The computer/terminal is the DTE. The telephone equipment/modem is the DCE. X.25 should only be used when PSTN service is only availabel in remote locations or no other service is available. FrameRelay FrameRelay was made for ISDN-B support. FrameRelay was designed to fix the problems with X.25 FrameRelay is packet-switching network like X.25 is, BUT FrameRelay was designed with FibreOptics! Not crappy PSTN phone cable. X.25 does not assume the network itself is reliable, so X.25 must do all kinds of error checking, making X.25 slow as a dog. FrameRelay assumes its network IS reliable, and thus if so, much better performance. FrameRelay uses PVC. FrameRelay goes from 64kbits to 1.544Mbits (T1 speed) ISDN Integrated Digital Services Network The idea behind ISDN was to make the Bell PSTN carry digital signals instead of normal voice/phone analog signals. ISDN IS DIALUP, not a permanent 24-hour dedicated connection!!! ISDN is based on the Bell PSTN Dialup, so it is like using the phone. You may have the ISDN connected all the time, but this is like using your phone constantly. ISDN is just like a phone, but for digital wans. ISDN have the SPID phone numbers to call. ISDN BRI is Basic Rate. BRI uses 3 channels. 2 B channels 64kbits/sec and 1 D channel 16bits/sec. This is called 2B+1D. Using both 64kb B channels together with compression can result in 128kbits/sec or higher. ISDN PRI Primary rate 23 B channels @ 64kbits /sec each. 1 D channel at 64kbits. ISDN PRIMARY is the equivalent of a T1, but ISDN is a dialup service, T1 is a dedicated line. ISDN-B Broadband ISDN-B supports super speeds of 51mbits 155mbits and 622mbits. ISDN-B is powered by ATM and SONET technology. ISDN-BRI Basic rate should be used at locations where connection is not fulltime or high speeds. ATM ATM is the best choice for mixing VOICE VIDEO AND DIGITAL DATA. ATM is the best choice for multimedia. ATM is supported by lower layerrs of SONET or SDH. ATM frames are always fixed 53-byte cells. 5 bytes of ATM frame is header, 48 bytes of data. Other network frames can be of various sizes, but ATM cell frames are always 53 bytes long. This makes ATM routing very efficent and fast. ATM can truly cruise if it has to upto Gigabytes/sec, but currently ATM is 155 or 622 mbits/sec . ATM networks obviously must use expensive ATM routers, ATM switches, ATM NICS, ATM cabling , everything on the Network, or anything connecting to the ATM network must be ATM-compatible. ATM can operate as a LAN (expensive) or most common today, ATM can also operate as a WAN. (aslo not cheap) If you need superfast 100mbits/sec or more over a WAN, use ATM. DS0 =1B channel at 64 kbits/sec T1 = 24 DS0s each of a 64kbits/sec B Channel = 1.544mbits/sec T3 = 28 T1s = 672 DS0s = 45mbits/sec Ethernet networks usually operate in half-duplex T1 and T3 lines are always full duplex. Ethernet 10mbit networks are must faster than a T1 line. T3s are must faster (45mbits) than a 10mbit Ethernet network. A T3 line (45) with a Fast Ethernet network would be better (100). 21.5 Practice Problems 1. Analog signals can have many values since they're constanty changing. 2. Digital signals can only have one of 2 values, 1 or 0. 3. The cheapest WAN connection of 1mbits/sec or faster is T1. (1.544mbits) 4. If you have a bunch of T1s (1.544) and need faster yet...get a T3(45) . A T3 is 28 T1s. 5. There are 28 T1s in a T3. T3s are smoking. 6. DS0=1 B channel DS3=T3. 7. The cheapest GUARANTEED (ie. LEASED) connection of 1.544mbits WAN, is T1. 8. For an international WAN for terminal emulation use only should be based on X.25(cheapest) and most flexible since X.25 is based on PSTN. A FrameRelay will require FIBRE OPTIC support by the service provider. Not likely to be found in developing countries. 9. The cheapest ANY2ANY WAN of 1mbits/sec or faster, is FRAME-RELAY. 10. A SVC switched virtual circuit is only active when there is a connection/data to be sent. 11. X.25 uses PADs to synchronize its WAN data packets. 12. X.25 also has the slowest and most overhead of any WAN due to all its error checking. 13. A LEASED T1 line has the LEAST overhead since it is POINT2POINT and doesn't need the overhead of a PACKET SWITCHING NETWORK. Anytime your WAN runs POINT2POINT instead of going through some PACKET-SWITCHING NETWORK "WAN CLOUD", the faster it is... 14. ATM is the FASTEST WAN you can get. 100mbits/sec or faster through a WAN. Rocketing speeds! ATM leaves T1=1.544mbits X.25, even FrameRelay in the dust. 15. X.25 is the only WAN that does its own reliable error checking. All other WAN technologies rely on the WAN network being reliable, thus error checking and correcting occurs in higher layers. 16. ISDN is a DIALUP service and can be used as a backup WAN solution in case of primary WAN failure. 17. X.25 is limited to a maximum of 64kbits/sec (not bad altogether really). 18. ISDN BRI is called 2B+D. 19. FrameRelay has low overhead PACKET-SWITCHING network. Your FrameRelay packets must go though that Packet-switching network "WAN CLOUD" of your service provider. 20. ISDN BRI is perfect for occasional (not permanent) DIAL ON DEMAND INTERNET or WAN service. 21. ATM is a rocket WAN and LAN. ATM is a speedfreak. ARM is like superexpensive too. Everything on the network needs to be ATM-compliant. 22. FRAME-RELAY and X.25 are both good choices for an INTERNATIONAL WAN. Use Frame-Relay if possible due to better performance. 23. ATM is the only WAN that can also be a WAN. ATM is for the super rich. As a WAN solution, ATM is also a possibility. 24. ATM is your most expensive WAN. SONET and FDDI aren't cheap either. 25. ISDN BRI is your overall cheapest WAN. ISDN BRI is cheapest since it is a DIALUP ON DEMAND service. It does not require a fulltime dedicated LEASED connection. 26. ATM, not only being the FASTEST WAN, the only WAN capable of being a LAN, also is the best choice for any multimedia. ATM is the only technology that supports VOICE VIDEO and DIGITAL DATA. ATM WILL BE THE FUTURE. 27. FRAME-RELAY is a packet-switching network "WAN CLOUD" with speeds of 1mbits/sec and more. 28. ATM is the WAN/LAN solution that uses 53 bytes cells. 5 for header 48 for data. 29. LEASED T1/T3 and X.25 and FRAME-RELAY all can have variable data lengths. Only ATM has a set cell size of 53 bytes. 30. Routers can all used on LEASED T1/T3 X.25 and FRAME-RELAY and ATM. Obviously, the router selected must be compaible with that technology. 31. Frame-Relay has a cool feature that lets you send data faster that the guaranteed delivery speed. If it is recived great, if the data is lost, no big loss, it'll just be resent. This is a very cool feature since you can pump more data thorugh your Frame-Relay at the right time of day, if you know how. 32. FRAME-RELAY comes with a CIR. Committed Information rate. FRAME-RELAY connections all have a guaranteed delivery speed. 33. If you have many ISDN BRI users needing to connect to a main corporate office, use a ISDN PRI (handles 23 ISDN BRI callers) 34. Frame-Relay assumes there will be very errors on the FR network, so it doesn't do error checking. X.25 does not make the assumption of few errors, X.25 does lots of error checking. 35. ATM networks talk to each other by establishing virtual circuits. 36. The 53 byte cells of ATM means that ATM can safely multiplex timing-critical packets with other packets. 37. ATM can run at speeds high enough to demand FIBRE OPTIC, 38. To support multimedia across a very fast WAN, choose ATM. 39. To have full redundancy requires a WAN solution between each location. 40. Routing between a point means you have to rely on both WAN solutions of that point. Practice Exam Planning 1. To connect 5 busy LANS together on a backbone, use a 100base-FL FIBRE BACKBONE. 2. A LAN of 100mbits/sec should be CAT5 or better. Coaxial is 10mbits/sec, Fibre cannot be use a LAN cable unless you got cash to burn. 3. A vampire tap is used to suck the blood out of a 10base-5 ThickNet cable. 4. Fibre cable has the best transmission speeds and performance of any cabling. 5. UTP is 100m limit. Thinnet is 185m. Thicknet is 500m. Fibre is kilometers. 6. CAT3 supports upto 10mbits only. CAT5 cabling needs to be installed to support 100mbits. 7. A single cable failure in a BUS topolgy kills that network until it is found and fixed. 8. Segmenting the network is one easy way to increase network bandwidth. 9. A mesh requires a point2point connection between all devices. Miles of cable can be used. 10. Collisions in an Ethernet network is normal. Excessive collisions can make the network slow, even unuseable. 11. 4.Transport Layer makes sur the data is delivered reliability. 12. An Ethernet NIC accepts a packet for processing if DESTINATION DATALINK (MAC) address matches the NIC. 13. TokenRing and Ethernet networks can both support multiple network protocols. 14. Installing a switch is the easiest and most effective way to to reduce collisions without segmenting the network. 15. If you have too many network broadcasts, the best solution is to install a router and segment the network. 16. If your using NETBEUI and TCPIP and having poor network performance due to broadcasts, a BROUTER is required. A bridge portion is needed for NETBEUI, the router portion is for TCPIP. IMPLEMENTATION CHAPTER 22 Network Essentials test concentrates on Microsoft network. Appletalk and Novell not so much. Resources Resources are anything like Disk Drives, Printers. Gateways, Shared Modems, faxes, tape drives etc etc are all resources. Some are software resources, some are hardware resoruces. If it can be accessed via the network, it is a network resource. Anything sharing over the network can be referred as a "server". PERMISSIONS Permissions is the right/access/security a user or group has to a resource. Permissions can either be USER-LEVEL or SHARE-LEVEL. SHARE-LEVEL SHARE-LEVEL can only be READ-ONLY or FULL-ACCESS. SHARE-LEVEL can be protected with a password . The password given either grants READ-ONLY or FULL-ACCESS. SHARE-LEVEL is the only security option you have if NOT using NT. (for Windows PCs) SHARE-LEVEL is far more administrative since you only have 2 choices in SHARE-LEVEL. USER-LEVEL USER-LEVEL security is far more secure that SHARE-LEVEL. To use USER-LEVEL you must have a database of user accounts somewhere. (NT or Novell Server, usually) When a user attempts to use ANY network resource, the user account and password are verified every time, then assigned permission is granted. USERS A user is any entity (usually a personal user account) that uses network resources. GROUPS Any number of users can be included in a group, and this greatly simplifies management. Instead of administering each user account, you only have to admin the group account. RIGHTS Rights are abilities a user or group is granted. Rights are assigned to a user or group. Some common NT rights that can be granted are 1. Create user accounts 2. Logon to a particular computer 3. Logon as a service 4. Take ownership of files or objects 5. Backing up and restoring files 6. Printing and print management of documents NT User Accounts and NT Groups A very important issue in user mangement is user accounts and groups they belong to. NT has GLOBAL and LOCAL group acounts. GLOBAL GROUPS Groups that exist across the DOMAIN. GLOBAL groups can contain only user accounts. GLOBAL groups can exist in other DOMAINS by TRUST relationships between domains. LOCAL GROUPS Local Groups can contain users and GLOBAL GROUPS. GLOBAL GROUPS contain users accounts LOCAL GROUPS contain GLOBAL groups and user accounts LOCAL GROUPS are assigned to resources GG DOMAIN USERS GG Domain users is a group of ALL DOMAIN USERS. User rights and resources granted to this group are very limiting. GG DOMAIN USERS can do very little. GG DOMAIN ADMINS Administrators of the Domain. Most powerful gg. LG ADMINISTRATORS GG DOMAIN ADMINS is always added to this LG ADMINISTRATORS on each server in the domain. LG POWER USERS A group that has full control over their workstations. More than GG Domain USERS have. EVERYONE A special group that contains all users in local or trusted domains. Anonymous users also. It is not possible to add or delete users from this group. IMPLEMENTING SECURITY ON WINDOWS NT Put users into GLOBAL GROUPS Put GLOBAL GROUPS into LOCAL GROUPS Assign permissions to those LOCAL GROUPS NT Security has 3 forms 1. ASSIGN PERMISSIONS TO A SHARED NETWORK RESOURCE Full access and read-access passwords for shares 2. USER LEVEL security applied to a network resource Share a network resource and assign groups or users to it 3. FILE-LEVEL SECURITY Highest level of security. Requires NTFS. Files or directories are assigned to users or groups. 1. Put users into Global Groups 2. Put Global Groups into Local Groups 3. Grant permissions to those local groups Windows 95 can act as a weak server but only has SHARE LEVEL. Windows 95 has no USER LEVEL. SHARE LEVEL has 3 types of access 1. READ-ONLY User can read and access files+folders, but cannot delete or save files. 2. FULL ACCESS User has full read-write-delete privledges on files+folders. 3. DEPENDS ON PASSWORD 2 passwords can be created. One that provides READ-ONLY ACCESS, the other FULL-ACCESS. The user types in the password and is given that level of access. Advantages of using Shares 1. Simple to administer 2. Depends on password is somewhat flexible Disadvantages of using Shares 1. The larger the network, the more shares and passwords each user must remember. This can easily become unuseable since there will be dozens and dozens of shares+paswords to remember. 2. Shares and passwords can easily be forgotten. If users write down shares and passwords, this bypasses any security efforts. 2. Users can share sharenames and passwords freely. In the public domain. USER LEVEL Security on Windows 95 Windows 95 cannot do user-level security. Windows 95 requires an NT or Novell server to do its authentications. SHARES under USER-LEVEL security can be assigned 1. READ-ONLY 2. FULL ACCESS 3. CUSTOM readfiles/writefiles/createfiles/listfiles/deletefiles/changeattributes/changepermissions Printer Security Setup the printer first on a server. Share the printer from that server When using Windows 95/98, "File+Print Sharing" must be enabled. When using Windoed NT, "File+Print Sharing" does not exist since it is always enabled. Creating a user account in Windows NT 1. Administrative Tools>User Manager/for Domains 2. The guest account is disabled by default. 3. Usernames can be upto 20 characters long, not case-sensitive 4. Passwords can be upto 14 chars long and ARE case-sensitive Sharing a Folder using SHARE-LEVEL Security 1. Create a folder 2. Right click this new folder and select properties>Sharing TAB SHARES have 4 access types 1. NO ACCESS This share has been explicity denied to these users/groups 2. READ Users/Groups cannot change or delete 3. CHANGE Users/Groups cannot change permissions or take ownership 4. FULL CONTROL Users/Groups have full control 22.1 Practice Problems 1. Files/NICS/Documents are not part of an administrative plan. USERS/RESOURCES/SHARING/PERMISSIONS are part of an administrative plan. 2. Any item that can be shared via the network, is a network resource. 3. Permissions are used to restrict access to network resources. User rights are used to enable user abilities. 4. Network resources can be almost anything. If it can be shared via the network. It's a network resource. 5. A networked printer MUST BE SHARED!!! By a SERVER or WORKSTATION. 6. Shares is is using passwords for a group of people 7. A UNC sharename is used to access the resource and should have a standard naming convention across the network. 8. UNC syntaxis \\servername\sharename 9. Enabling auditing itself does nothing. You must enable auditing, then audit those items you want to track. Auditing can only be done on NTFS partitions. Auditing does not protect the data at all! Auditing only shows who how when data was accessed. Data files+folders can only be protected with permissions. 10. Changing from WFW to NT is a major upgrade, that will interfere with the entire business flow. Implementing USER-LEVEL or SHARE-LEVEL access does not track data access. USER-LEVEL protects the data with NTFS permissions, but to track data requires that AUDITING be enabled then auditing be checked on those objects. 14. NT Workstation and NT Server and Novell Netware servers have USER-LEVEL security. Windows 95 dos not support USER-LEVEL security on its own, but can use servers that do support USER-LEVEL security. 15. When accessing ANY NT network share, that username is checked by the PDC or BDC database. If the username exists, the passwordis checked. Then the user is granted access. Remember, EVERYTIME you access ANY NT network share, you username and password are verified by the PDC/BDC. 16.A User is placed into groups. A User is granted permissions to access resources. A user is an item that requests to use network resources. 17. Windows 95/98 has only SHARE-LEVEL security. Therefore only has passwords to ensure security. SHARE-LEVEL is easier to administer for a small network. 18. Global Groups can contain USER accounts. Local Groups can contain USER accounts and Global Groups. 20. Window NT user accounts are created with User Manager/User Manager for Domains. 21. SHARE-LEVEL security uses passwords to restrict access from FULL ACCESS or READ-ONLY. SHARE-LEVEL can work in combination with NTFS permissions. 22. NT Administrator account cannot be delete, but it can be renamed. A very wise policy to do this. 23. User accounts are usually put into local and global groups to make overall administration easier. 24. UNC SHARE names must always be unique. 25. Passwords should only be used for user accounts. Never write down passwords and make a good password policy for the enterprise. 26. A users profile can contain extra info like home directory and profile type and location and desktop settings. 27. Groups are not added to resource ACLs. USERs in the group are added to ACLs. After assigning permissions to a group, added users to that group, to give then instant access to all permissions assigned to that group. 28. A Special user account is a System account that logs in as a Service. A Special user account is a user account that logs in as a service. Administrator and GUEST accounts are special user accounts. 29. User Manager for Domains is the only tool for user account management. 30. The default user groups in NT can handle most networks requirements. 31. Put users in Globsl Groups. Put Global Groups into Local Groups. Give permissions to local groups. 32. A system adminsitrator must balance all of 1)ease of use 2)user convenience 3)security 33. Erfe 34. A group you called "BACKUP" must be given the user manager for domain rights to backup. 35. The EVERYONE group is all users. Known and unknown and anonymous. No users can be added or delete from EVERYONE group. It is like a container. 36. Most modern operating systems like NT manage their groups with a user management tool .In UNIX this is often manually done with /etc/GROUPS and vi text editor. 37. User accounts assigned to Global Groups. Global Groups are added to LOCAL groups. LOCAL groups get assigned permissions to network resoources. 38. You can assign permissions to user accounts, but do this on a one-time only basis. It is far betterr to create a group, add the users to that group and and assgin rights to that LOCAL group. 39. Eferf 40. SHARES uses passwords, one for full access, one for read-only 41. SHARES is not very secure, but it is the easist to administer for small networks. 42. USER-LEVEL security on a SHARE uses the ACL to authenticate users. 43. A Windows 95 share, using USER-LEVEL security, can authenticate users using NT server or a Novell server. 44. NTFS file-level security requires an NTFS partition. 45. Rights and permissions can be added to user accounts, but in the long run will be very hard to work with an complex. In the short term will look like a solution. 46. File-level security is NTFS, and therfore requires an NTFS partition. File -level security cannot be don on any FAT partition. 47. NTFS enales you to do auditing, and can be used as a security feature to lock out users, since there is no way to boot an NTFS server from a diskette. NTFS should not be used as a security feature, since other security features should be used, like physical security measures. 48. Always watch that EVERYONE is removed from any NTFS permissions. 49. A weak password policy is okay if security is not an issue. 50. HOME Directory/Login SCRIPT/ROAMING or LOCAL PROFILES are in a profile. PASSWORDS and usernames are in the user database. 51. Windows 95 SHARE-LEVEL has FULL-ACCESS, READ-ONLY, Depends on password. 52. If no password is used in Windows 95/98 the admin can still choose between full and read-only access. 53. Windows 95/98 and ANY SHARE-LEVEL is weak security. 54. Windows 95/98 and ANY SHARE-LEVEL is easy to administer for small networks. Disaster Recovery Plans A System Admin must find ways to ALWAYS protect the DATA on all the drives, and reducing downtime, especially during work hours. Natural disasters, hardware failures, power surges, employee and hacker vandalism, and employee and operator and system admin errors all must be protected against. Some ways to handle this are a good security policy, good network security, UPS's, BACKUPS. BACKUPS A single server with super backup hardware, dedicated to backing up all servers over the network, is a good solution for large environments. Especially if all the servers are spread all over the place. The backup of this method will require a server with multiple tape drives and a very experienced backup operator. Backups over the network should really be done over the network for client machines, since backing up huge amounts of data over the network is putting a truly fantastic amount of data on the network. It is possible to have tape drives installed on each server and back up each server individually. This is a much faster solution, and possibly better, since data does not have to get sent over the network. Large data should always get backed up on each server. Never over the network. Any good backup plan should include FULL and INCREMENTAL backups. Weekly FULL backups on Fridays. (end of the workweek) and dailly incrementals. It is very important to keep a CATALOG of the backup files on a separate backup. I mean if the server crashes with the file needed to restore the backup, the backup tapes are useless. Unless the backup file can be regenerate somehow. UPS A UPS provides power after a power failure of most kinds. A UPS can shut the server (UNIX NT MAC) server down normally, instead of killing the power and possibily really damaging the servers filesystems. A UPS can altet all its users that a power failure has occurred (a building-wide power failure, shuts down all the client workstations, so that point is mute) A UPS can send a special alert (computer or pager) about the UPS or server problem Most UPS's provide about 15 minutes of power before a shutdown must occur. You must use fault-tolerant sytems for mission criticals. 22.2 Practice Problems 1. A major system crash will occur. Plan for it. Then when it happens, it won't be so bad. 2. The core of Disaster Recovery is safeguarding data and reducing downtime. 3. A UPS is critical for a backup plan since it lets servers shutdown properly in power failures. 4. If you decide to use a centralized server for backups, if that server crashes, there is no way to backup the network. Have dual centralized servers. Plus you'll have to backup gigabytes over the network. This can prove to be too slow and inefficient, with more and more data. You'll also saturate the network during backups over the network. MOST IMPORTANTLY, BACKUPS OVER THE NETWORK RELY ON THE NETWORK! IF THE NETWORK IS DOWN OR FAULTY, THE BACKUPS CANNOT BE DONE OR RESTORED!! 5. Backing up directly (each backup is done locally) is WAY WAY faster and more efficient. The network does not get saturated. However, this method requires a tape drive on each server. This is probably a better solution really than a centralized backup server. 6. Redundant fail-over servers are always expensive. Not cheap. 7. Using a RAID will not protect you if more than 1 drive fails. A Raid will also not help if someone deletes the data or an operator or sysadmin deletes data. Only a tape backup can help in the event of data loss! 8. A network backup server will saturate the network. 9. Network backup servers always have the best, largest and most expensive tape backup hardware. 10. A network backup server centralizes backups of the network, but at the high price of network saturation, slow speeds, and reliance on the network. 11. A redundant backup strategy is to use a local tape drives on each server as the primary backup method, and a network backup server as the secondary backup method. 12. An incremental backup, backs up all files changed since the last (NOT FULL) backup. 13. Smaller amounts of data can and should be backed up with a FULL backup daily. Larger amounts of data should be done incrementally each day and a FULL backup once a week. 14. A differential backup backs up all files changed since the last FULL backup. 15. Extra tapes of of very important data should be kept off location. That way the entire system could be rebuilt with new hardware. 16. Tape backup logs should not be relied upon, but can save you hours in a tape backup emergency. 17. A backup strategy should be tested in parallel. That is don't kill your network. Try and restore the server on a test machine. Don't ever pull "surprises" on your clients. They might "surprise" you with a pink slip. 18. An extra NIC card in each server can be used to make a new network segment to be used on for backups. This network-backup segment will be far more complicated to setup than just installing fast and easy tape drives in your servers. Simply the game! Keep it simple! 19. Full backups should be done daily if possibly. If you have too much data to backup, it'll burn out your tape drive and media quick. 20. If it is important enough to put onto tapes, it is important enough to bring a copy home (or offsite) 21. An up to date tape log can be your best friend when recreating data from tapes A bad tape log can be your worst enemy. 22. Microsoft recommends backing up the tape log file and keeping it with the offsite backup tapes. 23. A properly configured UPS can shutdown the server when local power fails. Send a broadcast message to its clients, send a message by pager to the sysadmin. 24. A UPS's main job is to provide power so the server can shut down properly. It also conditions the line so no spikes or brownouts get to the server. 25. The number one failure in computers is the hardest working part. The hard drives. 26. Being able to function despite a HARDWARE FAILURE (not operator or user error) is called FAUL-TOLERANCE. Fault tolerance only protects you from hardware failures. Fault tolerance can't protect you from stupid user mistakes. That's what tape backups are for!!! 27. Fault-tolerance must factor in and balance downtime, price, performance. 28. Good fault-tolerance design, you must carefully weigh what a loss will cost, and the price to prevent that loss with fault-tolerance and statistics. 29. If the projected loss is high enough, then the fault-tolerant designs must increase proportionally. 30. Hardware RAID is much faster than software raids(built into NT) and groups drives into a redundant group. 31. RAIDS 1+5 are are fault tolerant. RAID 0 is NOT fault -tolerant. 32.