The myth of single-vendor networks

By Kevin Tolly
Network World, 10/18/99

When the big guys go up against the start-ups over campus LAN strategies, the vendors almost always wind up arguing over the merits of a "best-of-breed" strategy vs. a "single-vendor" one. Whatever the benefits of working with one large vendor, it is time to explore, expose - and explode - the single-vendor myth.

Best-of-breed vendors argue that the ideal network can be built by choosing the best gear for each key area. The best LAN switch combined with the best WAN router complemented by the best IP telephony product will yield the best value. But the behemoths posit that the inherent synergies of gear provided by a single vendor result in the best possible overall solution.

The essential single-vendor myth consists of a "big picture" illustrating not only the core and edge switched infrastructure needed for your next-generation campus, but also the WAN connectivity, heterogeneous network support and, increasingly de rigeur, telephony. The vendors imply that purchased from one vendor, homogeneously branded products somehow will work better together. In other words, these products were designed to work together.

The problem with this picture is that most of the products were not, in fact, designed to work together.

Because Cisco has its own allure and is not governed by the rules that govern other vendors, I'll leave it out of this discussion. Nortel Networks, Lucent and Alcatel, however, warrant attention. All three chant the single-vendor mantra. But all three have achieved behemoth status through major acquisitions. When you have tens of thousands of employees to deal with, the integration of products and anything else is not a simple feat. At best, you might hope to end up with a patchwork quilt - at worst, a Frankenstein monster.

Today, when you buy a core-edge campus switch combo from Nortel, you get dramatically less consistent functionality than you would from Extreme or Foundry. Why? Because the Nortel edge and switch products were designed by two different companies (the core Accelar came from Rapid City by acquisition), while Extreme, for example, engineered its edge switches to run the identical code as the firm's core switches. Lucent, whose edge switches came from Madge/LANNET and whose core switches were built by acquired company Prominet, is in the same position.

Where, you may ask, is the synergy? If the core and edge switches can be shown to be so different, it is ludicrous to think that the emerging IP telephony products will have some special affinity for like-branded gear. After all, this gear is supposed to comply with industry standards rather than proprietary vendor protocols - only Cisco is allowed to break that rule.

While there are clear advantages to dealing with a large vendor, getting inconsistent product architectures and implementations may be the biggest drawback - and a heavy price that some net managers may not be willing to pay.