Chapter 8: WAN Cell Switching

In Depth

WAN switching is defined as the process of forwarding data traffic across a wide area network. WAN switching uses cell relay technology to multiplex all network traffic across WAN trunk links without a predefined timeslot for each type of connection. Cell relay networks use small, fixed−length packets called cells to send control information in a header attached to the user’s data. Using a common cell format for the encapsulation and transport of all network traffic, voice, video, and data over the WAN results in simplified, efficient, and quick routing and multiplexing of data.

These cell relay networks provide for very high throughput, short delays, and very low error rates. The industry standard for cell switching at Layer 2 is Asynchronous Transfer Mode (ATM) and LAN Emulation (LANE).

ATM was developed by the ATM Forum, which is part of the International Telecommunications Union Telecommunication Standardization Sector (ITU−T). Cisco—which is a leading member and one of the original founding members of the ATM Forum LAN Emulation Sub−Working group—has implemented LANE in most of its Core layer products.

The following Cisco WAN switches support ATM:

∙BPX 8600 series wide area switches (8620, 8650, 8680)

∙IGX 8400 series wide area switches (8410, 8420, 8430)

∙MGX 8220 edge concentrator

∙MGX 8800 wide area edge switch

These switches, which are also called nodes, fall into three Cisco WAN switched architectures:

∙Feeder nodes—The MGX 8220 concentrator shelves, which are used to aggregate narrowband UNI connections and multiplex traffic onto a single trunk link to a BPX switch or routing node.

∙Hybrid nodes—The IGX 8400 and the MGX 8800 switches, which are used to aggregate UNI connections. These switches are also used to route and switch packets to the trunks that lead to the final destination.

∙Routing nodes—The BPX switches that actually render the switching decisions and forward packets to appropriate trunk links.

In the Cisco LAN and Catalyst switching line, you can use the Cisco Catalyst 5000 ATM module or the LightStream series of switches for ATM cell switching. The LightStream series of switches is covered in Chapter 9.

ATM Overview

ATM is a cell−based networking technology designed to be a high−speed, efficient method of supporting multiple types of traffic, including voice, data, and video. ATM’s characteristics allow it to effectively support today’s networking requirements.

Some of the major benefits of ATM are:

∙Efficient bandwidth—ATM efficiently supports most transmission requirements of the network and allocates bandwidth as necessary. One of the primary reasons ATM is such a great protocol is its ability to accomplish this task without any manual intervention.

160

∙Scalability—ATM is highly flexible, accommodating a wide range of traffic types, traffic rates, and communications applications.

An ATM network includes two types of devices: ATM switches and ATM endpoints. One type of ATM interface, called a user−network interface (UNI), connects an ATM device to a switch; a second type, called a network−to−network interface (NNI), connects an ATM switch to another ATM switch.

ATM has built−in support for Quality of Service (QoS), which is used to guarantee a level of service for networks that use ATM. This guarantee includes bandwidth utilization and data throughput. This type of service is critical when dealing with newer multimedia technologies.

LANE

LANE is a method used to provide backward compatibility to legacy Ethernet and Token Ring networks. LANE makes an ATM interface look like an Ethernet or Token Ring network interface, so no modifications to existing network drivers or applications need to be made to support ATM environments. LANE allows ATM networks to emulate Media Access Control (MAC) broadcast networks. Before the implementation of LANE, a proprietary emulation device was needed to connect ATM to a LAN topology.

ATM LANE works with a client/server architecture to create an emulated LAN (ELAN). An ELAN is very similar to a VLAN, in that it limits local broadcasts and multicast traffic to the ELAN. LANE devices can be either clients or servers. The LANE Emulation service (LE service) consists of several different components:

∙LAN Emulation Client (LEC)—Resides in every ATM device and provides a LAN interface to higher layer protocols.

∙LAN Emulation Server (LES)—The centerpiece of the LANE architecture. A single LES is responsible for address registry and resolution for an ELAN.

∙Broadcast and Unknown Server (BUS)—The means by which ATM provides broadcasting support for an ELAN.

∙LAN Emulation Configuration Server (LECS)—Contains the database of LES/BUS pairs for all the configured ELANs.

LANE is discussed in much more detail later in this chapter.

ATM—Easy to Learn?

Nothing in ATM makes it easy to comprehend and learn. It defies a lot of what today’s network administrators have learned. Telling you that ATM is used as a backbone protocol in the network makes you think that you do not need to worry about packet−based broadcast LANs trying to communicate with cell−based ATM networks (which will be discussed in the following sections). In this chapter, I discuss how to connect ATM—which is a connection−oriented, point−to−point protocol—to the Layer 2 addresses of the broadcast domains in the LAN.

ATM is a difficult subject for most people, because they rarely are exposed to it on a day−to−day basis like Ethernet or Token Ring. In today’s networking environment, however, increased emphasis is being put on integrating data, voice, and video in networks, and ATM is a driving force. No other protocol today has ATM’s ability to ensure timely delivery of packets based on their type. In addition, ATM can be used on both LANs and WANs on almost any types of media, with speeds that can scale up to gigabits per second.

161

ATM Protocols

The protocols used in ATM have been specifically designed to support high−speed networks at speeds ranging up to gigabits per second (Gbps). Other physical LAN topologies, such as Gigabit Ethernet, provide high−speed networking and work very well in LANs. ATM, on the other hand, can handle network Gbps traffic in both LAN and WAN environments and could care less about the type of physical media being used.

ATM works on the theory that it is possible to expect upper−layer protocols to use a connectionless service to communicate with the lower layers. LANE is used to allow an upper−layer protocol to make connections to lower−layer ATM connection−oriented services. Thus, LANE provides a switching service that is transparent to the 802.x networks.

Traditional methods of transporting data use one of two ways to send data: character−based or frame−based. ATM is a cell−based switching technology that uses both circuit switching and frame switching to move packets through the network. Let’s take a closer look at ATM’s method of cell−based circuit switching.

ATM Circuit Switching

ATM is an efficient, high−bandwidth switching and multiplexing technology that also utilizes the benefits of circuit switching. Circuit switching is the process of using straight−through circuits between two points to ensure minimal transmission latency and guarantee equal bandwidth availability. Let’s take a look at the ATM technology components used in ATM circuit switching services:

∙Circuit emulation (CE)—A connection−oriented, constant bit rate ATM transport service. This service handles the heavy−duty, end−to−end timing requirements for the user’s chosen bandwidth and QoS requirements for establishing a connection. This is typically a dedicated line for applications such as video conferencing and multimedia.

∙Frame Relay—A widely used industry standard for WAN traffic that works by switching Data Link layer data. It uses multiple virtual circuits by implementing High−Level Data Link Control (HDLC) encapsulation between connected devices.

∙Switched Multimegabit Data Services (SMDS)—A high−speed, packet−switched, datagram−based WAN technology typically offered by telephone companies.

∙Cell relay services (CRS)—The basis for networking protocols, including ATM, SMDS, and IEEE 802.6. This networking technology uses small, fixed−length cells that can be switched in hardware at very high speeds.

Frame Relay, SMDS, and CRS are fastpacket transmission technologies used in today’s network. Most standard ATM platforms can support all three of these fastpacket technologies. Typically, these transmission technologies support two types of network connections:

∙Permanent virtual circuit (PVC)—A logical physical connection between two communicating ATM peers. This type of connection remains active (static) between two endpoints regardless of whether data is being transmitted over the connection. PVCs are typically used for interconnectivity between two fixed locations, such as a data center or company locations. This type of connection allows the network bandwidth to be predictable and constant.

∙Switched virtual circuit (SVC)—A switched connection that is established by means of a defined and standardized ATM signaling protocol. Such connections are set up dynamically and are activated only when data must be sent to the other end of the logical link. The connection is made on demand and is then terminated.

ATM Cells

ATM transports network data in fixed−sized units commonly called cells. Each cell is 53 bytes in length and is divided into a 5−byte header and 48 bytes of data. The 53−byte size of the cell, illustrated in Figure 8.1, is a compromise between the voice, data, and video advocates—one side wanted small cells (32 bytes) and

another wanted larger packets (64 bytes). The final decision was to add the defaults (32 + 64 = 96) and divide

162

the result by 2—and thus the data portion of the ATM cell contains 48 bytes.

Figure 8.1: The ATM cell.

The fixed size of the ATM cells provides some the following benefits:

∙Efficient bandwidth use of the physical medium

∙Ability of applications to share the network more fairly

∙Accommodation for bursty applications

∙Effective recovery of data loss on the physical wire

Note ATM is based on the switching and multiplexing techniques proposed by the ITU for Broadband Integrated Services Digital Network (BISDN) access.

Time Division Multiplexing

ATM uses a switching and multiplexing method called Time Division Multiplexing (TDM). This method places voice, multimedia, and data into fixed−length cells. These cells are then routed to their destination without regard to content.

TDM combines the information from different resources onto a single serial trunk link that dedicates a predefined timeslot on the multiplexed line for a piece of each resource’s data, as shown in Figure 8.2. If a source has nothing to send, then the timeslot goes unused, and the bandwidth is considered wasted.

Figure 8.2: Data from multiple switch ports (resources) is sent down a single multiplexed serial link.

163