A Brief History of Frame Relay
Originally Frame Relay was conceived to run over ISDN. The initial proposals went to the Consultative Committee on International Telephone and Telegraph (CCITT) in 1984. As mentioned in the initial chapters, CCITT is now known as ITU-T, for international standards, whereas American National Standards Institute (ANSI) is still known for American standards.
Standards
ITU-T approved what is known as Recommendation I.122, the framework for additional packet mode bearer services back in 1988. This was part of a series of ISDN specifications where Link Access Protocol D Channel (LAPD) carried the signaling information on the D channel. I.122 outlined how LAPD might be used in other applications besides ISDN. ANSI rapidly progressed on this recommendation and T1.606 was approved early in 1990 with complete approval in 1991. The ITU and ANSI standards for Frame Relay are in alignment with one another. ANSI T1.606 is equivalent to ITU-T I.122 for architecture, and ANSI T1.616 is equivalent to ITU-T Q.922 for data transfer.
In 1990 the Gang of Four consortium developed LMI. LMI is further discussed in the signaling section. The Gang of Four included the following:
Cisco
StrataCom (later acquired by Cisco)
Northern Telecom
DEC (later acquired by Compaq)
LMI popularized Frame Relay and the Gang of Four later formed the Frame Relay Forum that has grown to more than 300 members. The Frame Relay Forum in Figure 8-5 is at www.frforum.com. It is an excellent resource for Frame Relay.
Frame Relay is a bandwidth-on-demand technology where you share bandwidth with others in the cloud on a packet-by-packet basis. Although in a PVC the logical path is up and running, no bandwidth is actually consumed until needed. This is perceptibly more cost-effective than paying for leased lines in many business applications. Frame Relay not only provides a low cost of ownership, but it is standards-based, has low overhead along with high reliability (depending on your service level agreements), and internetworks well with other services such as ATM.
Frame Relay is based on sort of a KISS principal (“Keep It Simple, Stupid!”), by letting the higher-level protocols worry with the problems. Although the technology includes signaling and congestion-notification mechanisms, they are optional. This does not affect compliance with the standards but does affect performance. Your best bet is to subscribe to the Committed Information Rate (CIR) that is right for you. This and your maximum burst rate (less than the line speed) is primarily your service level agreement (SLA) with the provider. When your frames are above the CIR, the provider can set the Discard Eligible (DE) bit to 1. This just means that DE traffic is discarded prior to frames with the DE bit set to 0 or non-DE. However, the reality of all this is all bits are Discard Eligible by the nature of Frame Relay. It is up to the upper layers to do the error correction. On the other hand if you have a fat enough pipe, there are times you will be able to burst to more than your CIR depending on your SLA, but not more than your physical capacity. For example, your physical pipe may be a T1 or T3, but your end-to-end CIR may only be 56 kbps.
Frame Relay is a connection-oriented data-link protocol. The virtual connection or connection identifier for the PVC is the DLCI. It offers statistical multiplexing by switching variable-length frames. Obviously, this means that traffic delays vary according to frame size. However, Frame Relay is even optimal for carrying delay-sensitive traffic such as voice. Traffic shaping and quality of service prioritization mechanisms are discussed in CCNP Practical Studies: Remote Access (Cisco Press).
It is getting pretty old to say Frame Relay is a more efficient X.25 replacement because it has been alive for better than 10 years now. However, Frame Relay really is an updated X.25 that leaves the slow error correction and flow control to the upper layers so as not to burden things at Layer 2. You could say Layer 2 can switch it or pitch it and let the upper layers recover anything that has been discarded, for X.25 is the only Layer 2 protocol to offer error correction (retransmittal). Frame Relay typical speeds are 56 kbps to 44.7 Mbps (DS3).
Terminology
Like other technologies, many terms and acronyms are associated with Frame Relay. Use Figure 8-6 to help you review them.
Figure 8-6. Frame Relay Terminology
Frame Relay is used between the customer premises equipment (CPE) and the Frame Relay switch, but the complete path is known up front. Figure 8-6 illustrates the LMI signaling (keepalives) that occurs from your router to the local frame switch. The Frame Relay connections from r1 to r2 and r3 are through PVCs. Local DLCIs are the Layer 2 connection identifiers assigned by the service provider.
Autosensing LMI and traffic shaping are among the many significant features that have been available since 11.2 code. Congestion control may be through forward explicit congestion notification (FECN) and backward explicit congestion notification (BECN) if in fact Frame Relay is used within the service provider cloud. The DE bit is a priority discard bit, but packets within your CIR take priority. On the other hand, you may burst higher than your CIR if you have the physical capacity to do so. The following list provides a quick review of the main Frame Relay terms:
DLCI— Data-link connection identifier is a number that identifies the logical local circuit between the router and the frame switch. My discussion of DLCIs assumes local significance, which is the norm in practical application. Think of these as circuit identifiers that are provided by the service provider.
PVC— Permanent virtual circuit is a virtual circuit that corresponds to an end-to-end path going through a Frame Relay cloud. It is permanently established compared to a switched virtual circuit (SVC), which is established on demand.
LMI— Local Management Interface is the signaling between the local router and the local Frame Relay switch. The three types include Cisco (LMI), ANSI (Annex D), and Q933A (Annex A).
CIR— Committed Information Rate is the delivery during normal conditions—minimum acceptable throughput.
Committed burst— Guaranteed delivery under normal conditions.
Excess burst— Bytes outside the CIR accepted by the frame switch and marked as DE eligible.
DE— Discard Eligible is really a priority discard bit in case the network becomes short of resources.
BECN— Backward explicit congestion notification is set in frames traveling in the opposite direction of the congested path.
FECN— Forward explicit congestion notification informs the DTE device receiving the frame that congestion was experienced in the path between the source and destination.
SLA— You have a service level agreement with your service provider, which includes such things as response time, availability, restoration of service, throughput, and SLA reporting. For example, physically you may have a T1 or T3 but you only pay for the subscription you need, yet you can burst to the maximum burst rate within your physical capacity if available. On the other hand, frames flagged as DE are dropped when congestion occurs. The ingress frame switch optionally performs the policing. It can drop all frames in excess of CIR plus burst or it can just mark them DE and let them proceed. This is a service provider policy choice.
This terminology may vary according to the service provider, but these are some of the most common terms and acronyms used in the Frame Relay environment.
Now I want to look at Frame Relay frames to analyze the details of some of the terminology mentioned.