IBM's recent decision to hand over its network hardware operation to Cisco raises some very interesting issues about product support in the enterprise.
The deal between the two industry giants is a complex one, and focuses far more on access to markets than on who actually puts the boxes together. Make no mistake, both companies did very well out of it. The basics are pretty well known: IBM gets $US2 billion from Cisco for a whole range of components over five years. Cisco gets most of IBM's switching and network patents; IBM stops making routers and switches (apart from Token Ring), but gets access to Cisco's service providers. Cisco joins forces with IBM's hugely influential Global Services operation to promote Cisco-based network technologies (e-commerce, voice/data integration and so on) in competition chiefly with Lucent and Nortel. This is a huge market, and both partners look set to do very nicely, thank you. And let's face it, IBM never did get its act together in the internetworking business, so this seems a great solution, doesn't it?
Well, yes and no. There were some intriguing comments in the press from the IBM network hardware marketing folk at the time of the announcement, to the effect that IBM had lost out to Cisco in the router race by taking its eye off the ball. "Hey, if only we'd realised that SNA and Token Ring were yesterday's technology and that the future belonged to IP we wouldn't be in this sorry state now," they seemed to be saying. This is the kind of "mea culpa" rhetoric that IBM chairman Lou Gerstner and his shareholders love to hear. Owning up to past mistakes, playing down the significance of legacy technologies, and looking forward to a future that is bright and cheery and begins with the letter "e-".
After all, Lou Gerstner is trying -- generally very successfully -- to re-brand IBM as an e-commerce services organisation. And why not? Cisco was able to don the Internet mantle without too much of a struggle. But of course, Cisco is a very focused company with just the right kind of products to exploit the Internet revolution. IBM has considerably more baggage to carry -- and long-standing products to support. So its e-persona is developing a little more slowly. Having said that, Gerstner was no doubt more than happy to shed one of his last unprofitable product units (network hardware) and focus on his twin enthusiasms of component supply and services.
But wait a minute -- let's not forget why so many of IBM's products were SNA-focused for so long. Many users have been reluctant to abandon SNA applications, despite the ubiquitous nature of TCP/IP. Why, I wonder. Possibly because SNA is the most functionally rich and stable networking architecture yet devised. Possibly because those users have spent 20 years building applications that take full advantage of its strengths.
Which begs a question. What does the IBM/Cisco arrangement mean for enterprise support? Well, for the most part, things shouldn't change too much. IBM has given the usual assurances to continue providing support for its network devices, and it has a pretty good reputation for supporting products long after their "best before" date. There are some understandable concerns about IBM ceding its channel-attached bridge/router business to Cisco, as this is becoming a strategic point of integration between SNA and TCP/IP in many installations. But, as network analyst Anura Gurug points out, IBM may steer its larger users via a different route -- OSA Express, announced back in February and due to be fully functional in the new year, will take advantage of facilities in OS/390 2.7 and 2.8 to plug straight into the G5 server's Self-Timed Interconnect bus, supporting Gigabit Ethernet, Fast Ethernet, and ATM. This will be a very fast alternative to traditional channel-attached bridge/routers, and, though it doesn't address existing support concerns, it suggests that IBM still retains control over the next wave of network technology for mainframe users who have all the latest kit. And IBM's still responsible for Token Ring, too.
The writing has been on the wall for IBM-proprietary network standards for some time. Though recent OS/390 announcements (such as the Triple DES security for SNA and SSL client authentication to the TN3270 server) show continuing enhancement of business-critical network resources, the SNA-to-IP migration path gets wider all the time, with the lure of IP-dependent e-commerce functionality. And there can be no doubt that, with Cisco in the driving seat, this trend will accelerate. In the meantime, though, it should be business as usual -- but keep an eye on your support contracts.
Mark Lillycrop is Director of Research at IT market analyst Xephon (www.xephon.com)
Join the CIO Australia group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.