How does MFR stack up?

As we continue our discussion of using multiple parallel transmission links at Layer 2, let’s explore how multilink frame relay (MFR) stacks up against the two technology competitors we discussed last time: inverse multiplexing over ATM (IMA) and multilink PPP (MLP).

MFR actually has two implementations: an end-to-end version and a version for the user-to-network interface (UNI). Technically, they’re quite similar. The major difference is that end-to-end MFR, as described in the Frame Relay Forum’s FRF.15 Implementation Agreement, does not require the frame relay service provider to do anything special. By contrast, FRF.16 for the UNI requires MFR-compatible equipment both on the customer site and in the service provider’s network.

In many ways, MFR was built to take advantage of the strengths of IMA and MLP, both of which predate MFR.

Because MFR is frame-based, it uses bandwidth more efficiently than IMA. This can be quite important, given that the whole reason for using multiple links is to increase throughput, and wasted bandwidth decreases throughput. Combining frame relay fragmentation with MFR can give a performance level that approaches that of IMA.

MFR compares quite favorably with MLP. In fact, a good deal of MFR is based on the concepts in MLP. The major advantage in using MFR is support of multiple protocols. Frame relay is designed to be protocol-agnostic. MLP theoretically could support upper-layer protocols other than IP, but in actual practice, it is seldom – if ever – used for protocols other than IP.

The major problem MFR faces today is one of availability. The most useful form of MFR is an FRF.16-based service, but there are few services available. What’s the best way to accelerate service availability? Ask your service provider for it, and give preferential consideration to suppliers that are moving forward.