1. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Chapter 18 Virtual Circuit Switching: Frame Relay and ATM

2. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.1 Virtual circuit wide area network Main difference between a circuit-switched and a packet-switched network is that in the latter the links are shared, channelized between different communication paths. Frame Relay and ATM [WAN technologies] use virtual circuit switching. Packet switching uses two different approaches: the datagram approach [used in network layer] and the virtual circuit approach [used in data link layer]. Global address: A source or destination needs to have a global address which is unique address over the global space. However, Global addressing in virtual circuit networks is used only to create a virtual circuit identifier.

3. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.2 Virtual Circuit Identifier (VCI) This identifier is used for data transfer. It is a small number. It only has switch scope; it is used by a frame between two switches. When a frame arrives at a switch, it has one VCI; when it leaves, it has another. VCI does not need to be a large number since each switch can use its own unique set of VCIs.

4. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.3 VCI phases To communicate, a source and destination need to go through three phases: Setup phase: Source and destination use their global addresses to help switches make table entries for the connection. Teardown phase: Source and destination inform the switches to erase the corresponding entry. Data Transfer phase: Occurs between these two phases.

5. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.4 Switch and table – Data Transfer Phase To transfer a frame from a source to its destination, all switches need to have a table entry for this virtual circuit. The switch table, in its simplest form, has 4 columns. Switch holds four pieces of information for each virtual circuit. A frame arrives at port 1 with a VCI of 14. When the frame arrives, the switch looks in its table to find port 1, and VCI 14. When it is found, the switch knows to change the VCI to 22 and send out the frame from port 3. The procedure is same for all frames from the source to destination.

6. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.5 Source-to-destination data transfer The figure below shows how a frame from source A reaches destination B and how its VCI changes during the trip. Each switch changes the VCI and routes the frame.

7. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Setup Phase Permanent Virtual Circuit [PVC]: Source and destination may choose to have a PVC. In PVC, the connection setup is simple. The corresponding table entry is recorded for all switches by the administrator (of course remotely and electronically). An outgoing VCI is given to the source, and an incoming VCI is given to the destination. The source always uses this VCI to send frame to that particular destination; the destination knows that the frame is coming from that particular source if the frame carries the corresponding incoming VCI. If there is a need for duplex communication, two virtual circuits are established. PVC is like a leased telephone line. Costlier: A connection is need all the time even if not in use. So, two parties pay for the connection. A connection is created from one source to one single destination. If a source needs connections with several destinations, it needs a PVC for each connection.

8. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 SVC Setup Phase Switched Virtual Circuit [SVC]: SVC creates a temporary, short connection that exists only when data are being transferred between source and destination. SVC requires a connection setup phase. To setup connections between A and B, we need setup request and acknowledgement. Setup Request: Source A sends a setup frame to switch I. Switch I receives the setup request frame. It knows that a frame going from A to B goes out through port 3. The switch creates an entry in its table for this virtual circuit, but it is only able to fill three of the four columns. The switch assigns the incoming port (1) and chooses and available incoming VCI (14) and the outgoing port (3). It does not know the outgoing VCI, which will be found during the acknowledgement step. The switch then forwards the frame through port 3 to switch II. Switch II receives the setup request frame and does as switch I. Switch III receives the setup request frame and does as switch I and II. Destination B receives the setup frame, and if it is ready to receive frames from A, it assigns a VCI to the incoming frames that came from A, in this case 77. This VCI lets the destination know that the frames comes from A, and not from other sources.

9. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.6 and 18.7 SVC setup and SVC setup acknowledgment Acknowledgement: Destination sends an acknowledgement to Switch III. Acknowledgement carries the global source and destination addresses so the switch knows which entry in the table is to be completed. The frame also carries VCI 77, chosen by the destination as the incoming VCI for frames from A. Switch III uses this VCI to complete the incoming VCI column for this entry. Note that 77 is incoming VCI for destination B, but outgoing VCI for switch III. Switch III sends an acknowledgement to switch II that contains its incoming VCI in the table, chosen in setup phase. Switch II updates. Switch II sends an acknowledgement to switch I and switch I sends an acknowledgement to source A. The source uses this as the outgoing VCI for the data frames to be sent to the destination B. Teardown phase: After sending all frames to B, source A sends a special frame called a teardown request. Destination B responds with a teardown confirmation frame. All switches erase the corresponding entry from their tables.

10. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.8 Frame Relay network Frame Relay is a virtual circuit wide area network that was designed to respond to demands for a new type of WAN. Prior to Frame Relay, X.25 was used X.25 performs virtual circuit switching at network layer. X.25 has a low 64-Kbps data rate. X.25 has extensive flow and error control at both data link and network layers. X.25 requires ACK for both data link layer frames and network layer packets that are sent between nodes and between source and destination. X.25 has its own network layer, which differs from that of the Internet. To use X.25 in Internet, Internet must deliver its network-layer packet [called as datagram] to X.25 for encapsulation in the X.25 packet, which doubles overhead.

11. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Instead of X.25, people lease T-1 or T-3 lines from public service providers. Pay for the whole even if we use just 10%. Assumes that the user has fixed data rate. With bursty rate and the average is still within the limit, is not acceptable. Bursty data requires ‘Bandwidth on demand’ Frame Relay Can run at higher speed [1.544Mbps, and recently 44.376 Mbps] Operates in physical and data link layers. Thus, can run with the network layer of the Internet. Allows bursty data. Allows a frame size of 9000 bytes, which can accommodate all local area network frame sizes. Less expensive. Error detection at data link layer only. No flow or error control. No retransmission policy. Flow and error control should be handled at higher layers. Provides PVC and SVC.

12. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.9 Frame Relay layers VCIs in Frame Relay are called (Data Link Connection Identifier) DLCIs. The table on the switches match an incoming port-DLCI combination with an outgoing port-DLCI combination. Physical Layer of Frame Relay No specific protocol at Physical layer. Supports any protocol recognized by ANSI. Data link layer of Frame Relay Simplified version of HDLC at data link layer because HDLC provides extensive error and flow control fields that are not needed in Frame Relay. Frame Relay has features for congestion control and quality of service.

13. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.10 Frame Relay frame Frame Relay frame format is similar to that of HDLC, except that control field is missing because this field in HDLC is used for flow and error control. Address field defines the DLCI as well as some bits used to control congestion and traffic. Address (DLCI) field: First 6 bits is used by part 1 of DLCI. Second 4 bits is used by part 2 of DLCI. DLCI is of 10 bits. Command/Response: Allows upper layers to identify a frame as command or response. Not used by Frame Relay protocol. Extended Address: Indicates whether the current byte is the final byte of address. EA = 0 means another address byte is to follow. FECN: Indicate congestion in the direction in which the frame is traveling. Informs destination. BECN: Indicate congestion in the direction opposite to which the frame is traveling. Informs source. Discard Eligibility: Indicate priority level of the frame. DE =1 means discard this frame if there is congestion. Can be set by the sender or by any switch in the network.

14. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Extended Address: To increase the range of DLCIs, Frame Relay address has been extended from the original 2-byte address to 3- or 4-byte addrseses. Frame Relay Assembler/Disassembler (FRAD): FRAD assembles and disassembles frames coming from other protocols to allow them to be carried by Frame Relay frames. FRAD can be implemented as a separate device or part of a switch.

15. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Voice Over Frame Relay (VOFR): Sends voice through the network. Voice is digitized using PCM and then compressed and sent as data frames over the network. Quality is not as good as circuit-switched telephone network. Local Management Information (LMI) is a protocol added to Frame Relay to provide management features. Keep-alive mechanism to check if data are flowing Multicast mechanism to allow a local end system to send frames to more than one remote end system Mechanism to allow an end system to check the status of a switch.

16. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.13 Multiplexing using different frame sizes Asynchronous Transfer Mode (ATM) is the cell relay protocol. Design Goals Work with high data rate by using the features [large bandwidth, less susceptible to noise] of optical fiber. Interact with existing systems and provide interconnections without lowering their effectiveness or requiring their replacement. Cost effective. Able to work with and support existing telecommunications hierarchies (local loops, local providers, long-distance carriers,…) Connection-oriented to ensure accurate and predictable delivery. Move as many functions as possible to hardware and make the software simple and faster.

17. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Problems in existing systems: As networks become complex and so headers need more information in them. Header size becomes larger. All network devices should read and process a lot. Variable sized frames. Scenario One line uses large frames (data frame) and other uses small frames (audio and video information). If the large frame arrives a moment earlier, then the smaller frames need to wait a long time in the multiplexer before being transmitted. Thus data frames create unacceptable delays for small frames.

18. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.14 Multiplexing using cells A cell network uses the cell as the basic unit of data exchange. A cell is defined as a small, fixed-sized block of information. As frames of different sizes and formats reach the cell network from a tributary network, they are split into multiple small data units of equal length and are loaded into cells. The cells are then multiplexed with other cells and routed through the cell network. High speed of links coupled with the small size of the cells means that, despite interleaving, cells from each line arrive at their respective destination in an approximation of continuous strem. Cell network can handle real-time transmissions, such as phone call, without the parties being aware of the segmentation or multiplexing at all.

19. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.15 Asynchronous Transfer Mode (ATM) multiplexing ATM uses asynchronous time-division multiplexing. Multiplex cells coming from different channels. ATM multiplexers fill a slot with a cell from any input channel that has a cell; the slot is empty if none of the channels has a cell to send. ATM is a cell-switched network. The user access devices, called the endpoints, are connected through a user-to-network interface (UNI) to the switches inside the network. The switches are connected through network-to- network interfaces (NNIs).

20. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.17 TP, VPs, and VCs Connection between two endpoints is accomplished through TPs, VPs and VCs. Transmission Path(TP): Physical connection (wire, cable, satellite, and so on) between an endpoint and a switch or between two switches. [All highways] Transmission path is divided into several virtual paths (VP). It provides a connection or a set of connections between two switches. [Highway] Virtual Circuits (VC): All cells belonging to a single message follow the same virtual circuit and remain in their original order until they reach their destination. [Lanes]

21. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.18 Example of VPs and VCs The first two VCs seem to share the same virtual path from switch I to switch III, so it is reasonable to bundle these two VCs together to form one VP. The other two VCs share the same path from switch I to switch IV, so it is also reasonable to combine them to form one VP.

22. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.19 Connection identifiers Virtual connection is defined by a pair of numbers: the VPI and the VCI. Virtual Path Identifier (VPI) defines the specific VP. Virtual Circuit Identifier (VCI) defines the particular VC inside the VP. VPI is same for all virtual connections that are bundled (logically) into one VP.

23. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.20 Virtual connection identifiers in UNIs and NNIs Length of VPIs for UNIs is 8 bits and that for NNIs is 16 bits. Length of VCIs [16 bits] is same for UNIs and NNIs. Idea behind dividing a virtual connection identifier into two parts is to allow hierarchical routing. Most of the switches in a typical ATM network are routed using VPIs. The switches at the boundaries of the network, those that interact directly with the endpoint devices, use both VPIs and VCIs.

24. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.21 An ATM cell Cell is only 53 bytes long with 5 bytes allocated to header and 48 bytes carrying payload (user data may less than 48 bytes). Header contains the VPI and VCI. ATM has two type of connections: PVC and SVC. PVC: A permanent virtual circuit connection is established between two endpoints by the network provider. VPIs and VCIs are defined for the permanent connections, and the values are entered for the tables of each switch. SVC: Each time an endpoint wants to make a connection with another endpoint, a new virtual circuit must be established. Request for making virtual circuit is made using network layer address and services of another protocol (such as IP).

25. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.22 Routing with a switch ATM uses switches to route the cell from a source endpoint to the destination endpoint. A switch routers the cell using both the VPIs and the VCIs. Scenario: A cell with VPI of 153 & VCI of 67 arrives at switch interface (port) 1. Switch checks its switching table, which stores six pieces of information per row: arrival interface number, incoming VPI, incoming VCI, corresponding outgoing interface number, the new VPI, and the new VCI. Switch finds the entry in the table and discovers that the combination corresponds to output interface 3, VPI 140 and VCI 92. It changes the VPI and VCI and sends via interface 3.

26. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.23 ATM layers Three layers: Application adaptation layer, ATM layer and physical layer. Physical layer ATM calls can be carried by any physical layer carrier. SONET is preferred because: High data rate. Using SONET, the boundaries of cells can be clearly defined. SONET specifies the use of a pointer to define the beginning of a payload. If the beginning of the first ATM cell is defined, the rest of cells in the same payload can easily be identified because there are no gaps between cells. Just count 53 bytes ahead to find the next cell. Can run on wireless also. But, the problem of cell boundaries must be solved. One solution for the receiver is to guess the end of cell and apply the CRC to the 5-byte header. If there is no error, the end of cell is found, to a high probability, correctly. Count 52 bytes back to find the beginning of the cell.

27. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.24 ATM layers in endpoint devices and switches ATM layer Provides routing, traffic management, switching and multiplexing services. Processes outgoing traffic by accepting 48-byte segments from the AAL sublayers and transforming them into 53-byte cells by the addition of a 5-byte header.

28. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.26 ATM headers ATM uses two formats for this header, one for user- to-network interface (UNI) cells and another for network-to-network (NNI) cells. GFC: 4-bit GFC field provides flow control at the UNI level. It is not needed in NNI and so these fields are added to VPI. Long VPI allows more virtual paths to be defined at NNI level.

29. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 VPI: 8-bit for UNI and 12-bit for NNI. VCI: 16-bits for both UNI and NNI. Payload Type (PT): 3 bits First bit defines the payload as user data or management information. The interpretation of the last 2 bits depends on the first bit. Cell loss Priority (CLP): Provided for congestion control. A cell with CLP as 1 should be retained as long as there are cells with CLP as 0. Header Error Correction (HEC): Code computed for the first 4 bytes of header. It is a CRC with divisor x 8 +x 2 +x+1 that is used to correct single-bit errors and a large class of multiple-bit errors.

30. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.27 AAL1 Application Adaptation Layer (AAL) should be to accept data frames (data payload) and streams of bits (multimedia payload). AAL has two sublayers: SAR & CS. AAL sublayer SAR (Segmentation and reassembly) does segmentation [divide the data frame or stream of bits into 48-byte segments to be carried by a cell] at the source and reassembly at the destination. Before data are segmented by SAR, they must be prepared to guarantee the integrity of the data. This is done by a sublayer called Convergence Sublayer (CS). ATM defines four versions: AAL1, AAL2, AAL3/4, AAL5.

31. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 AAL1 Supports constant bit rate applications such as audio and video, Uses existing digital telephone networks such as voice channels and T-lines. Bit stream of data is divided into 47-byte chunks and encapsulated in cells. CS sublayer divides bit stream into 47-byte segments and passes them to SAR sublayer. CS sublayer does not add a header. SAR adds 1 byte of header and passes 48-byte segment to ATM layer. Header has two fields: Sequence Number (SN): 4-bit field used to order the bits. First bit is sometimes used for timing, which leaves 3 bits for sequencing (modulo 8) Sequence Number Protection (SNP): 4-bit field protects the first field. First 3 bits automatically correct the SN field. Last bit is a parity bit that detects error over all 8 bits.

32. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.28 AAL2 Originally designed for variable data rate but now used for low-bit-rate traffic and short-frame traffic such as audio, video or fax. Ex. Mobile telephony. AAL2 allows the multiplexing of short frames into one cell. CS overhead consists of five fields: CID: 8-bit defines the channel (user) of short packet; LI: 6-bit indicates how much of the find packet is data; PPT: defines the type of packet; UUI: Used by end-to-end users; HEC: last 5 bits used to correct errors in header. SAR overhead is start field (SF) that defines the offset from the beginning of the packet.

33. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.29 AAL3/4 AAL3 was intended to support connection-oriented data services and AAL4 to support connectionless services. But, they evolved to have the fundamental issues as same. Thus combined as AAL3/4. CS header and trailer has six fields: CPI: Defines how the subsequent fields are to be interpreted. The value at present is 0. Btag: This value is repeated in each cell to identify all the cells beginning to the same packet. Value is same as Etag. BAsize: 2-byte tells the receiver what size buffer is needed for coming data. AL: 1-byte included to make the rest of the trailer 4 bytes long. ETag: 1-byte serves as ending flag. L: 2-byte indicates the length of the data unit.

34. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 SAR header and trailer consists of five fields: ST: 2-bit identifier specifies the position of the segment in the message: beginning (00), middle (01), or end (10). A single-segment message has an ST of 11. SN: sequence number. MID: 10-bit field identifies cell coming from different data flows and multiplexed on the same virtual connection. LI: Defines how much of the packet is data, not padding. CRC: last 10 bits of trailer is a CRC for the entire data unit.

35. McGraw-Hill©The McGraw-Hill Companies, Inc., 2004 Figure 18.30 AAL5 AAL3/4 provides comprehensive sequencing and error control mechanisms that are not necessary for every application. AAL5 assumes that all cells belonging to a single message travel sequentially and that control functions are included in the upper layers of the sending application. ATM has provided a fifth AAL sublayer, called Simple and efficient adaptation layer (SEAL). Four trailer fields in CS layer are: UU: Used by end users. CPI L: 2-byte indicates the length of the original data. CRC: last 4 bytes are for error control on the entire data unit.