Compiled by A.K.Asokan. (E-mail:email@example.com) Page Nos.1 of 16 WIDE AREA NETWORKLocal Area Network can be easily understood since typically every office has a LAN.One could personally see the switches connecting various computer nodes and may be arouter to which the switches are connected. Also we can see the Network Interface Card(NIC) at the back of the computer where we plug in the cable with the RJ45 connector,the blinking light etc. on the card. (Sometimes when it is not working, we pull out thecable from the card and plug it again to try our luck!!) All that we know. But how willyou establish a connectivity between your Head office located in Chennai and the Branchoffice in Singapore? What are all the possibilities and whom to approach? We know thatthere is no way to lay a cable all the way to Singapore! That will be extremely difficultand not a good idea in terms of cost and feasibility also. So, let’s first understand whatmakes the difference between a LAN and a WAN.The general and immediate answer comes to our mind is that LAN is ‘local’ and WAN isa “wide” area network. The Distance!. But now-a-days we have the wireless LANtechnology which can really cover a ‘wide’ area. Then is it the Bandwidth? Again we seethe competition at our door step that we can get gigabit broadbands from various serviceproviders at much, much cheaper rates. This is not either. What then?Generally, to set up a LAN, we buy computers, switches, cables, connectors, routers etc.But do you buy any equipment or wire to connect your Branch office located atSingapore or Delhi or Bombay? The answer is No. So we own the LAN infrastructure butwe generally lease WAN infrastructure from any third party Service Providers or from aTelephone company. WANs generally carry a variety of traffic types such as voice, dataand video.Imagine in your office there are 25 telephone extensions. So you have 25 telephoneinstruments placed on each officer’s desk wired and you own the infrastructure. If youwant to call your three Branch offices at Singapore, USA and Canada do you own theinfrastructure? i.e. the wires, telephone instruments connected till your branch offices?No. you ‘lease’ them. To make it more clear, you take a telephone line from thetelephone company with ISD facility. Now that using your telephone company’sinfrastructure, you can call to Singapore or any other branch offices anywhere in theworld!. Similarly we own the LAN infrastructure and we ‘lease’ the infrastructure of athird party provider or a Telco for Wide Area Network (WAN) communication.If you carefully read the previous notes on LAN communication, we came across lot ofterminologies. Like that here also we have to learn some terminologies. Let us firstunderstand the WAN connection types.
Compiled by A.K.Asokan. (E-mail:firstname.lastname@example.org) Page Nos.2 of 16WAN TechnolologiesImagine you have a branch office in Delhi and another one in Canada. Your requirementis that every two minutes, you will communicate with each branch office. You musttherefore have an always up connectivity for this at the same time cost effective. You cango for a dedicated line (a synchronous serial connectivity) so that it will be always on.But to establish a dedicated line, you will have to spend lot of money. If in case you arenot using the full capability of the dedicated line and it happens that sometimes you willcommunicate with your branch offices only for 3-4 hours a day, then spending so muchon a dedicated line is not a feasible solution.The alternative is a dial up line. Then every time you communicate, you have to dial andbring the line up and then send the data. It is tedious and though cost effective whencompared to dedicated line, the bandwidth is also not guaranteed!. What we need is aconnectivity, which is always ON at the same time cost effective also. One of suchservices is called Fame Relay. Frame relay is a Technology and not a protocol. We willbe discussing about frame relay in a few minutes. Before that let us quickly see someterminologies.Leased Line: Typically, these are referred to as a point-to-point connection or dedicatedconnection. A leased line is a pre-established WAN communication from the Customerpremises through the ISP’s infrastructure (DCE switches) to the customer of the remotesite. If cost is no object, it is really the best choice.Circuit switching: When you hear the term circuit switching, think of phone call. Costis the big advantage in circuit switching. You only pay for the time you actually use. Nodata can transfer before an “end-to-end” connection is established. Circuit switching usesdial up modems or ISDN and is used for low bandwidth data transfers. Every time youcommunicate, you have to bring the line up and establish connectivity and only then youwill be able to send the data. This is not always up connectivity.Packet switching: This is WAN switching method that allows you to share bandwidthwith other companies to save money. Packet switching can be thought of as a networkthat is designed to look like a leased line, yet charges you (and costs) more like circuitswitching. Frame relay and x.25 are packet switching technologies, Speeds can rangefrom 56kbps to T3 (45 mbps).Frame Relay: Frame Relay is a packet switched technology. Frame Relay is a data linkand physical layer specification. This provides high performance. Frame Relay is asuccessor to X.25. Frame Relay provides features for dynamic bandwidth allocation andcongestion control.
Compiled by A.K.Asokan. (E-mail:email@example.com) Page Nos.3 of 16ISDN: Integrated Services Digital Network (ISDN) is a set of digital services thattransmit voice and data over existing phone lines. No need to change the telephone cable.It defines high speed digital data transfer over standard phone lines. In dial-up, themodem is converting the digital data from computer and sends as analog data through thephone lines. ISDN can offer a cost effective solution for remote users who need a higherspeed connection than analog dial-up links offer. ISDN is also a good choice as a back-up link for other types of links such as Frame Relay or a T1 connection. (The speed of theT1 connection is 1.544mbps full duplex. Full duplex means both ways at the same timedata can be transmitted).It is important to understand what is a back-up line and how and why ISDN is used as abackup line for leased line. Please refer the following diagram.In the above diagram, there is a Head office at Chennai and the branch office in Delhi andboth the offices are connected via a leased line through RA and RB routers. These officesare also connected through an ISDN line as a back-up line. Back-up line means,originally the communication is being sent through the leased line which is always up.Suppose if the leased line is down for some reasons, then the ISDN line willautomatically come up and establish the connectivity provided both the routers RA andRB are configured in such a way. When the leased line comes up again, the ISDN back-up line will automatically disconnected. Thereby there is no downtime in theconnectivity. We may even use the ISDN line for all communication. But it is verycostly.Another use of this set up is to send ‘interesting traffic’. What is meant by interestingtraffic? Suppose if you have a mail server at the Head office and you want tocommunicate with the mail server at the branch office, then you can direct these SMTP
Compiled by A.K.Asokan. (E-mail:firstname.lastname@example.org) Page Nos.4 of 16packets alone through the ISDN line and all other packets like http and ftp traffic throughthe leased line. This has to be configured in the router’s serial interface.HDLC: High level Data link control (HDLC) is a protocol at the data link layer. HDLCis a point-to-point protocol used on leased lines. (Remember the following diagram? wediscussed point-to-point protocol and multi access networks in the previous OSI Layernotes.)High Level Data link Control (HDLC) is not intented to encapsulate multiple NetworkLayer protocols across the same link. The HDLC header carries no identification of thetype of protocol being carried inside the HDLC encapsulation. Because of this, eachvendor that uses HDLC has their own way of identifying the Network layer protocol.This means that each vendor’s HDLC is proprietary for their own equipment. To make itclear, Cisco’s HDLC is proprietary to Cisco and it is the default encapsulation used byCisco routers. It will not communicate with any other vendor’s HDLC implementation.So if you have only Cisco routers, how will you connect your routers and network toothers? If you have Cisco router and a non-cisco router with a serial connection, you mustconfigure PPP or another encapsulation method such as Frame Relay because the HDLCdefault won’t work.Point to point protocol (PPP) is an industry standard protocol. Because all multi-protocol versions of HDLC are proprietary, PPP can be used to create point to point linksbetween different vendors’ equipment. It uses a Network Control Protocol field in theData link header to identify the Network Layer protocol. It allows authentication andmulti-link connections and can be run over asynchronous and synchronous links.To summarize the above two paragraphs, protocol means ‘rules’. If Head office routerand Branch office router want to exchange communication, there must be some‘protocols’ which both the routers must accept. HDLC (High Level Data Link Control)and PPP (Point-to-point protocol) are two such protocols. However, if both the routersare of the same vendor (i.e. Cisco) then HDLC is default. If both the routers are not fromthe same manufacturer, then HDLC cannot be used. Here PPP can be used. However, ifboth the routers are from the same vendor and HDLC is the default, still if you prefer,PPP can be used instead of HDLC. But one side HDLC and other side PPP will not work.
Compiled by A.K.Asokan. (E-mail:email@example.com) Page Nos.5 of 16For example, all of us can speak English. That is like HDLC. A default language. But Iknow French and Mr.Neil also knows French. Then we have the option either to speak inFrench or in the default language English. But if I speak German and Mr.Neil speaks inSpanish, then we will not be able to communicate anything!Setting up these protocols with the router’s serial interface using router commands iswhat is known as ‘encapsulation’. Another thing is that IP is not the only routed protocol,there are other networks like ATM, AppleTalk, IPX/SPX etc. apart from IP. (You mayread once again the above 3 paragraphs, if you are not sure of some concepts).Point to point protocol (PPP) is a Data link layer protocol that can be used over eitherasynchronous serial (dial-up) or synchronous serial (ISDN) media. It uses somethingcalled the LCP (Link control protocol) to build and maintain data link connections.Network control protocol (NCP) is used to allow multiple network layer protocol (routedprotocols – IP, ATM, AppleTalk etc.) to be used on a PPP connection. LCP is a methodof establishing, configuring, maintaining and terminating the point to point connectionand NCP is method of establishing and configuring different network layer protocols.Since HDLC is the default serial encapsulation on Cisco serial links and it works great,when would you use PPP?The basic purpose of PPP is to transport layer 3 packets across a Data link layer point topoint link. It is non-proprietary, which means that if you don’t have all Cisco routers,PPP would be needed to be encapsulated on your serial interfaces. The HDLCencapsulation would not work because it is proprietary to the specific vendor. In addition,we should understand that PPP can encapsulate several layer 3 routed protocols andprovide authentication, dynamic addressing and call back facility etc. This may be theencapsulation solution of choice for you over HDLC.PPP Authentication methodsBefore establishing connectivity between two routers, it has to be authenticated eachother. There are two methods of authentication that can be used with PPP links i.e. PAPand CHAP. PAP stands for Password Authentication Protocol and CHAP stands forChallenge Handshake Authentication Protocol. It is upto the Network administrator toselect PAP or CHAP. However CHAP is the preferred protocol as it is a 3 way handshakeprotocol.
Compiled by A.K.Asokan. (E-mail:firstname.lastname@example.org) Page Nos.6 of 16Password Authentication Protocol (PAP)The Password Authentication Protocol (PAP) is the less secure of the two methods.When the PPP link is first established the remote node sends back to the originatingrouter the username and password until authentication is acknowledged (Refer diagrambelow).In the above diagram, PAP authentication is described. RA wants to communicate withRB. As both are configured for PPP, RB knows the username and password. When RAsends the username and password, RB verifies it and then accept/reject it so thatcommunication is established / not established. If the username and/or password arewrong, then RB rejects the request to establish connectivity. PAP is not a strongauthentication protocol because the username and password are sent across the link as‘clear text’ and hence there is no protection.Challenge Handshake Authentication Protocol (CHAP)The Challenge Handshake Authentication Protocol (CHAP) is used at the initial start upof a link and at periodic checkups on the link, to make sure the router is stillcommunicating with the same host.After PPP finishes its initial link establishment phase, the local router sends a challengerequest to the remote device. The remote device sends a value calculated using a oneway hash function called MD5. The local router checks this has value to make sure itmatches. If the value doesn’t match, the link is immediately terminated. See diagrambelow.
Compiled by A.K.Asokan. (E-mail:email@example.com) Page Nos.7 of 16Initially a challenge packet is sent and a response is received for that with the usernameand password. Then the acceptance / rejection are established. (Both the routers send thepackets to each other to make sure their identity) (Those who work on the router canmake sure this by first encapsulating the serial interface with ppp chap and thendebugging the same while bringing the interface down and again bringing it up).Frame RelayFrame Relay is a bit more complex than the leased line networks we have discussedabove i.e. HDLC and PPP protocols. These leased line networks are easy toconceptualize. Not so with Frame Relay. It can be significantly more complex, which iswhy it is often represented as a cloud in networking graphics. I have made every attemptto make you understand the concepts though only certain topologies are explained indetail down the page. Please have the diagram referred often while you read theexplanations so that it will help you to understand it in a better way. At some places,there will be repetition of concepts which is deliberately made in order to give an impactfor understanding. Let us try to understand Frame Relay. Read on……Frame Relay has become one of the most popular WAN services deployed over the pastdecade. It is a packet switched technology. Frame Relay technology saves money overalternatives. By default it is classified as a non-broadcast multi access (NBMA) network,which means that it does not send any broadcasts, such as RIP updates across the networkby default.
Compiled by A.K.Asokan. (E-mail:firstname.lastname@example.org) Page Nos.8 of 16In the above diagram, there is a Head office located in Chennai and there is a Branchoffice located in Delhi. In order to communicate with the two offices, both of them takea frame relay connectivity. The infrastructure of the service provider will give you aVirtual Circuit (VC) which is always up and running. As long as you pay the charges, itis permanent and hence it is known as Permanent Virtual Circuit (PVC). There will behundreds of PVCs in a frame relay cloud.Frame relay comes in between a leased line and a dial-up line. Under frame relay, theService Provider commits that any given time, you are guaranteed with a certainbandwidth. This is known as Committed Information Rate (CIR). You can buy yourCIR. Frame relay is a technology and not a protocol.Imagine you get a connection that looked like a leased line and acted like a leased linebut allowed you to pay for whatever portion of that leased line you actually used. That isessentially what packet switched networks do. You pay for whatever you use. FrameRelay provides a packet switched network to many different customers at the same time.It is based on the assumption that all customers will never need to transmit constant dataall at the same time. An example to throw more light on the “assumption” could be thatin a bank, there are number of customers who deposited money. The bank can divert themoney for some other investment purposes on the assumption that all the customers willnot come to withdraw the money at the same time. Like that, the Frame Relayconnectivity is provided to you on the assumption that all the customers will not senddata constantly so that some portion of the bandwidth will always free and can be used byother customers.Frame Relay is normally implemented with standard unshielded twisted pair (UTP)cabling from the Service provider. Devices on the subscriber premises are calledcustomer premises equipment (CPE). This is again connected to a device known asChannel Service Unit / Data Service unit (CSU/DSU) and then connected to your
Compiled by A.K.Asokan. (E-mail:email@example.com) Page Nos.9 of 16company’s router with the serial interface. The CSU/DSU (Modem) generally takes theoriginal signal from the provider and converts its electrical properties into the serialformat for your router. It also sometimes act as a channel bank, or multiplexer, whichbreaks the channelized connection apart so that the voice and data can be separated.The following diagram explains the cabling of frame relay from your premises till theservice provider.
Compiled by A.K.Asokan. (E-mail:firstname.lastname@example.org) Page Nos.10 of 16Frame relay works by providing a portion of dedicated bandwidth to each user and alsoallowing the user to exceed their guaranteed bandwidth if resources on the Telco networkare available. The providers allow their customers to buy a lower amount of bandwidththan what they really use. This is possible because of multiplexing. There are two typesof multiplexing. 1. Frequency Division Multiplexing (FDM) 2. Statistical Division Multiplexing (SDM)In frame relay, we use statistical division multiplexing. This means that if all thecustomers are transmitting at the same time, then you are assured with the CIR (assuredbandwidth). Depends on some customers are not transmitting, you get a better bandwidthto that extent. Anything beyond your committed information rate (CIR) is a bonus foryou and it is known as ‘burst’. The following terminologies are important.Access rate and Committed Information Rate (CIR)The maximum speed at which the frame relay interface can transmit is known as Accessrate. CIR the maximum bandwidth of data guaranteed to be delivered. The customers canfix the CIR it can be 256kbps or 512 kbps etc. depends on their requirement. However,in reality this is the average amount that the service provider will allow you to transmit. Ifthese values are the same, (access rate and the CIR) then the FR connection is prettymuch like a leased line. However, they can also be different values. Let us say that youbuy an access rate of T1 (1.54mbps) and a CIR of 256 kbps. By doing this, the first256kbps of traffic will always be available to you and it is guaranteed to be delivered.Anything beyond that is called a ‘burst’ which is the transmission that exceeds yourguaranteed 256kbps, and can be any amount upto the T1 access rate (if that amount is inyour agreement). The delivery of the 256 kbps is guaranteed. The “burst” will bedelivered on something called best effort delivery. Or may be not delivered. If yourTelco’s equipment doesn’t have the capacity to deliver at the time you transmitted, thenyour frames will be discarded and the DTE (your router) will be notified so that it will re-transmit the same. Timing is everything – you can scream data out at six times yourguaranteed rate of 256kbps (T1) only if your Telco has the capacity available on theirequipment at that moment!We have previously discussed that multi-access networks require identification (LANCommunication notes) and point-to-point communication doesn’t need identification ofdestination. Frame relay is multi-access in nature. But it will not broadcast so we can saynon-broadcast multi-access (NBMA). (remember the example of a telephone exchange?It will not send a call to all the lines it is connected to; but it can access multipledestinations. In other words, it will not ‘broadcast’ but multi-access is possible.) So framerelay is an NBMA network and since it is multi access, identification of the destination isnecessary. Non-broadcast multi access network (NBMA) means that it does not send any
Compiled by A.K.Asokan. (E-mail:email@example.com) Page Nos.11 of 16broadcasts like RIP updates (periodic updates) across the network by default. Let us nowsee what is the identifying mechanism by frame relay.If you remember, we also have discussed previously that Ethernet on a LAN environmentworks based on MAC address. Likewise frame relay is based on a number known asDLCI numbers (Delsi numbers). DLCI stands for Data Link Connection Identifier.(Technically, unlike MAC address, DLCIs do not specify the physical port. They specifythe logical link between two systems. It is only virtual and not physical)LAN – MAC address (Media Access Control)WAN (FR) – DLCI (Data Link Connection Identifier)Every VC (virtual circuit) is identified by DLCI numbers. The Telco gives you yourDLCI numbers. DLCI 100, DLCI 150, DLCI 200 etc. DLCI has only local significance.The Telco’s frame relay switch maintains the DLCI numbers. There cannot be twoidentical numbers in a switch. If you need to call Singapore office, you need a telephoneconnection with ISD facility which you may get from a Telco. Like that, in order to setup network communication with Singapore office, you must get a frame relay connectionwith a committed information rate (CIR) (your required bandwidth) from the ServiceProviders. Refer the following diagram.In the above diagram, router RA is in Chennai and router RB is in Singapore. Both theoffices are connected via a frame relay connectivity. Refer the cloud, it has an ISP switch,which provides a PVC (Permanent virtual circuit) till the other side ISP which provided aframe relay connectivity to the Singapore office. The PVC has DLCI numbers 102 forchennai office and 201 for Singapore office. (Think of the PVC as a virtual pipe) WhenRA wants to send a packet to RB it checks its routing table whether there is any route
Compiled by A.K.Asokan. (E-mail:firstname.lastname@example.org) Page Nos.12 of 16specified to RB whose destination IP address is 126.96.36.199. See the routing table of RAwhich says that any packet you want to sent to IP address 188.8.131.52, send it to Serial 0 (S0)interface of the same router (IP address 184.108.40.206). Accordingly it sends the packet to S0interface. Since the serial 0 interface of RA is encapsulated with frame relay, it hasanother table called frame relay MAP table. Hence it looks into the Frame relay maptable as to what to do with the packet whose destination IP address is 220.127.116.11. The framerelay map table tells yaar, for 18.104.22.168, you put the packet into DLCI number 102. It putsthe packet into DLCI number 102. Since the line is always up, it reaches the Telco’sframe relay switch and the switch, switches the packet to DLCI number 201 of the otherside ISP through the frame relay cloud, who in turn delivers the packet to IP address22.214.171.124 which is RB router.Similarly, when RB router wants to send a reply packet, it sends the same to the routersserial interface and since the interface is encapsulated by frame relay, it has a fame relaymap table which tells, to put the packet into DLCI number 201 and it reaches ChennaiDLCI 102 and delivers the same to the chennai router 126.96.36.199 (please understand thatthis is the way it works and in order to understand fully, we need to understand the LMItype and other configuration details which is omitted in this note.)Refer the above diagram, in the frame relay cloud, there are hundreds of frame relayswitches. Frame relay is a synchronous communication means it synchronises with theclock speed. The clocking is provided by the data communication equipment (DCE)which is present in the Telco’s infrastructure. Frame relay specifies how to operate in thelocal look which is not fully described here and it doesn’t specify how the frames arecrossing the cloud. Refer the following diagram to understand the cloud a little more.
Compiled by A.K.Asokan. (E-mail:email@example.com) Page Nos.13 of 16The Frame Relay ‘cloud’ contains hundreds of Frame Relay switches. There are twotypes of devices exist. The DCE and DTE. DCE stands for Data communicationsEquipment and DTE is for Data Terminal Equipment. The DCEs are the frame switchesin the cloud. And the DTEs are the routers. The DCEs are capable of handling lot ofVirtual Circuits (VCs) simultaneously. Apart from that the DCEs provide “clockingsignal” to the DTEs. Clocking signal is needed because Frame Relay is a synchronousprotocol and the frames are ‘synchronized with the clocking signal’. There is no start bitand stop bit needed in this case. Hence it is more efficient and faster.For those who find it is difficult to understand please assume that your Telco will giveyou a frame relay connectivity for data communication (as a telephone connectivity forvoice communication), and you must require some equipment like the router and Modem(CSU/DSU) unit. CSU is Channel Service Unit and DSU is Data Service Unit to set upthe connectivity. For a telephone call to land in Singapore office, the Singapore officealso must have equipped with a telephone connectivity from any Telco available inSingapore. As you don’t bother about how the Telcos communicate with each other, thesame scenario exist here as to how the Telco does the data communication through framerelay technology, in a user perspective. The idea here is to understand that tocommunicate with a remote place which is geographically and administratively doesn’thave any boundaries, and to set up a WAN connectivity, we have certain technologiesand one of such technologies is Frame relay.
Compiled by A.K.Asokan. (E-mail:firstname.lastname@example.org) Page Nos.14 of 16There are various types of frame relay connectivity. 1) Full mesh topology 2) Partial mesh topology, and 3) Hub and spoke topologyThe following diagram explains the full mesh topology.In the above diagram, the Head office (HO) is connected to 5 branch offices. All the firstbranch offices are connected each other. This is known as full mesh topology and itprovided lot of redundancy and fault tolerance. Even if one link fails, the connectivity isnot completely disconnected. However, look at the diagram how many PVCs are requiredto buy?! It will cost exorbitantly!. You can have partial mesh also which means may bebranch office 3 and 5 alone can be fully meshed. The alternative is Hub and spoketopology. Refer the following diagram.
Compiled by A.K.Asokan. (E-mail:email@example.com) Page Nos.15 of 16In the above diagram, the Head office (HO) is connected to 5 branch offices. However,all the five branch offices are connected only to Head office and not each other. So weneed to get only five PVCs. It is less costly to that extent. The only problem with hub andspoke topology is that, in case the router at the HO is failed, then none of the branchoffices can communicate each other. Otherwise, it is an excellent way to connect remotelocations. The hub and spoke topology is a simple and generally the easiest to set up. In ahub and spoke topology, every device is connected to a central hub device, (a router)which performs the routing work.In the following diagram a hub and spoke topology is described. See that if the branchoffices have to contact each other, they have to connect through the Head office.
Compiled by A.K.Asokan. (E-mail:firstname.lastname@example.org) Page Nos.16 of 16In the above diagram one new concept we are going to discuss before we close. Take aclose look at the router RA. It has only one serial interface. But it has been connected tothree branch offices located in Singapore, USA and Canada respectively. All the threebranch offices are (encapsulated with frame relay) connected to a single serial interface.This is possible due to a new concept known as “sub-interfaces´.What are sub-interfaces? Think of a sub-interface as a logical interface. Several sub-interfaces will share a single hardware interface (physical interface) which is calledmultiplexing. You can define sub-interfaces in the serial interface of the router. Beforecreating sub-interfaces, the physical serial interface must be encapsulated and then youcan define sub-interfaces. Generally one sub-interface per PVC will be created. In ourabove diagram, there are three sub-interfaces. Each sub-interface must be givenindividual IP addresses. See the above diagram, the sub-interfaces are given ip addressesviz. 172.16.1.1, 172.16.2.1 and 172.16.3.1. Sub-interfaces are of two types. 1) Point-to-point and 2) multipoint. It is very very important to note that when you create sub-interfaces, the physical serial interface must not be given an IP address but it has to beencapsulated with Frame Relay.Dear Friends, the WAN concepts described above is only to the point. You are requestedand encouraged to read lot of materials from standard publications, journals andperiodicals in order to get good understanding of the concepts. Hope this note was usefulto you. Best of luck. – asokan. (email@example.com). ____________________________________________