3-4 A Frame Relay Primer
EDWARD BURSK
Virtually everyone who makes data communications their business has heard of frame relay. Many people understand, to one degree or another, that frame relay is a form of packet switching. Many have heard that frame relay can save them a lot of money.
Although companies are running their business applications over frame relay networks today, there is still much about this technology that is not well understood. Where did it come from? What is a DLC? What are the potential pitfalls? How can it be used for Internet access? What effect does the emergence of asynchronous transfer mode (ATM) have on frame relay services?
This chapter is devoted to providing answers to these questions. The chapter also offers an update on the current state of public frame relay service (often used for Internet/Intranet access), which has been available commercially for several years now.
HISTORICAL AND PRACTICAL DESCRIPTION OF FRAME RELAY
Packet switching is well-known today, having been introduced into business communications some 20 years ago in various formsnotably as Systems Network Architecture (SNA) from IBM Corp., the X.25 standard of the International Telecommunications Union-Telecommunications Standards Sector (ITU-TSS, formerly known as the CCITT), and ARPANET, which was the progenitor of the Internet. With the widespread deployment of packet-switching devices around the world, this technologyin which streams of data are broken up into discrete blocks of data (called frames or packets) and then statistically multiplexed over a transmission pathis well-proven and considered mainstream.
The form of packet switching called frame relay emerged from a convergence of several factors, including:
- The need for higher-speed packet technology driven by LAN internetworking.
- The availability of a definition for a streamlined packet technique from integrated services digital network (ISDN).
- The desire of equipment vendors and communications carriers to deliver a high-speed packet service as rapidly as possible.
Combined Characteristics of OSI Layer 2 and Layer 3
As a packet-switching technique, frame relay has its roots in X.25 and ISDN standards. ISDNs D channelthe packet-switching control channel used by ISDN equipment to set up callsprovided a basis for the definition of frame relay. The protocol used over the D channel is called link access protocol for ISDN D channels (LAPD). It is based on link access protocol balanced (LAPB), the protocol used to run X.25 connections over point-to-point links.
LAPB is a link layer, or layer 2, protocol according to the Open Systems Interconnection (OSI) model from the International Standards Organization (ISO). Layer 2 is responsible for establishing and maintaining a reliable connection across a link, guaranteeing that frames of data (as they are called at this level) are successfully transferred across a link without the presence of undetected errors.
X.25 is a layer 3 protocol, wherein network connections (i.e., calls) are made, often traversing many such links. These calls are referred to as virtual circuits, because many can be multiplexed across shared links to many destinations.
Given that it is possible to establish virtual circuits across a frame relay network, many people think of frame relay as a Layer 2.5 protocol, because it has characteristics of both levels: the ability to establish a point-to-point connection between two devices and the ability to multiplex calls over many links (see Exhibit 3-4-1).
Exhibit 3-4-1. Multiplexed Virtual Circuits in Frame Relay
Faster Implementation
However, there are some important assumptions made to streamline frame relays speed of operation and ease of implementation as compared with other layer 2 and layer 3 protocols such as LAPB and X.25. First, it is assumed that the end devices (i.e., the devices that need to communicate over some distance) are intelligentthat is, they employ a higher-level protocol that guarantees the integrity of the communications between them, negating the need to have the network perform that function itself, as it must with X.25.
Second, it is assumed that the physical and electrical connections (OSI layer 1) are cleanthat is, they have a very low error rate, as with the gigabit-speed fiber optic connections used increasingly by long-distance carriers. The result is that few errors occur as a result of using the lines. When errors do occur, frame relay can identify the situation, but does not retransmit the error frame itself. Error frames are simply dropped, with the expectation that the end devices will detect the missing frame and retransmit the data.
A common protocol stacka grouping of protocols running together, each at different levels of the OSI modelis TCP/IP (Transmission Control Protocol/Internet Protocol). TCP establishes robust transport connections over a network and IP carries data packets across a network. This combination handles retransmission of data if errors occura perfect fit for frame relay. Therefore, most of the error-handling and retransmission functions in X.25 and LAPB did not need to be implemented in frame relay, allowing for smaller, faster implementations as compared to X.25.
Exhibit 3-4-2 compares protocol stacks for the OSI model, X.25, and frame relay.
Exhibit 3-4-2. Comparison of Protocol Stacks
|