Will convergence require additional disaster recovery?

At a recent Network World "State of the WAN" town meeting, some network managers expressed concerns about convergence - specifically, about the downtime risks and resulting revenue loss potential of "putting all their eggs in one basket" (the eggs being voice, data and video).

It is certainly true that the more stuff you have riding on a single facility, the greater the impact on your business if that facility should fail. Appropriate disaster recovery plans have always been important for critical traffic - whether voice or data - and the nagging dilemma has been how much to spend on network diversification vs. the degree of risk if you skimp here and there.

As you converge more types of traffic over a packet network - be it frame relay, ATM, or IP - you have several options for diversification. Since most outages tend to occur in the access link (local loop), using inverse multiplexing to aggregate several lower-speed links to achieve greater bandwidth, instead of buying a single fat pipe, might make sense for protection in this network segment.

If you are using an ATM access link, for example, a standard way to do this is to use Inverse Multiplexing for ATM (IMA), which is a feature that must be supported in your access equipment. Its counterpart for frame relay, Multilink Frame Relay, is still under construction. Both allow you to aggregate multiple diverse access links to accumulate the required amount of bandwidth, rather than buying a single, large network pipe. Traffic can fail over to a remaining live individual link or links, if one should fail. Generally, to assure this type of resiliency, you should purchase your diverse access links from more than one access provider.

In addition, equipment is available that will let you dial around your frame relay network automatically in case of a failure. Some of these products are embedded in frame relay access devices (FRAD) or smart DSU/CSUs. For example, AdTran offers smart DSU/CSUs and FRADs that enable you to dial up to 96 simultaneous connections over ISDN at 64K bit/sec or Switched 56 to bypass your frame relay service. If a failure occurs either in the frame network or on the local loop, the DSU/CSU or FRAD detects the failure and initiates the backup call either at the remote or host site.

Such capabilities are also available in some routers. It wouldn't hurt to ask your router vendor what kind of performance hit you might take by bundling such functionality into it. If the performance degradation happens only in the event of failures, this may be a negligible issue. What you do lose by going the router route, however, are the network statistics gathered by your smart frame DSU/CSU.

It is important to remember that you cannot build complete redundancy into your entire frame relay network while still using services from just one carrier - unless, possibly, the carrier is the product of a recent merger and actually runs two separate networks. We feel that such carriers are ideally poised to leverage their situation to offer inexpensive back-up services by offering diversity across two separately managed networks. So far, we are not aware of any that are doing so, though some have acknowledged to us that it doesn't sound like a bad idea.

Installing back-up permanent virtual circuits (PVC) - two predefined paths from point A to point B through the network - helps in case a network element along the first path should fail. Even so, any data residing in the faulty switch at the time of failure or en route to that switch will be discarded, possibly affecting your communications. Also, if the carrier's core backbone dies, redundant paths through the broken network won't help much.

In terms of cost, reinforcing only part of your network - such as the local loop or with a back-up PVC- is generally the least expensive option. The "dial around" options in FRADs, DSU/CSUs and routers can actually be free from an equipment perspective if you already have the proper interfaces in your equipment. But of course you still have to pay for the back-up service. And buying two separate services from two separate carriers will cost the most - but could still pay off if you are the type of firm that loses millions to a minute of downtime.

---

Steven Taylor, consultant and broadband packet evangelist, and Joanie Wexler, an independent networking technology editor and writer, team up to bring you this analysis and commentary. Taylor specializes in education and market analysis, and Wexler adds incisive reporting and research. For more detailed information on most of the topics discussed in this newsletter, connect to www.webtorials.com, the first Web site dedicated exclusively to market studies and technology tutorials in the Broadband Packet areas of Frame Relay, ATM, and IP. Feedback and additional topic ideas are welcome. Please contact taylor@webtorials.com or joanie_wexler@mindspring.com>

ADTRAN whitepaper on disaster recovery and frame relay networks

Veritas rolls out slew of data protection software
Network World, 06/25/99

Adtran puts emphasis on the enterprise
Network World, 05/10/99

Security action plan
Network World, 02/22/99

Archive of Network World on Frame Relay newsletters