Managers need to do homework to pick the ideal next-generation switch

It seems all network infrastructure vendors these days have boxes that do everything. Not only do most routers (or switches or whatever other aliases they might be marketed under) now provide wirespeed routing, switching and port densities to die for, but almost all can perform some level of traffic prioritization.

It’s not too difficult to imagine a day when vendors will begin pushing boxes that transform into SUVs at night and continue to act as routing and switching platforms by means of wireless transceivers, while their network managers cruise along in comfort, monitoring the network via a dashboard console. (Okay, maybe it’s a bit of a stretch.)

The bottom line is that on the surface at least, it is getting harder to differentiate one vendor’s infrastructure wares from another’s. But below the surface, there are differentiators such as price, QoS capability and interoperability. How are network managers supposed to determine which box is best for their particular network?

That, essentially, was the topic of discussion at a NetworkWorld (US) Town Meeting held in Toronto late last month. The event brought together representatives from 3Com, Alcatel, Extreme Networks, Foundry Networks, Lucent and Nortel to face an audience of approximately 250 network managers. Kevin Tolly, president of the Tolly Group, a N.J.-based consulting and testing firm, and John Gallant, editorial director of Network World (US), moderated the discussion and provided some vendor-neutral context.

The answer to the question of which box is best is still unfortunately the same as it’s always been – it depends on the network and the network manager’s goals for the network. If all a manager is looking for is raw bandwidth, then the lowest price per port may be the most important consideration. But if a manager wants to run voice over IP, QoS mechanisms will be a critical consideration.

While each vendor at the town meeting touted their own company’s boxes as being ideal, the “best of breed”, “first of class” and “this kicks everyone else’s butt” rhetoric was kept to a minimum.

With prompting, the vendors doled out some useful advice audience members could use in shopping for their next-generation LANs.

Some of the pearls of wisdom included:

•implementing management and maintenance tools to cut down on staffing.

•writing tighter RFPs, because as one vendor noted, a loose RFP tends to spin out of control.

•planning to use more than one vendor in the network.

The last point is somewhat self-serving from the vendors’ perspective, since all of them are aware that most of the shops they’re trying to sell into already have Cisco equipment installed. But, as Tolly noted, it also makes sense. If a network manager doesn’t force a vendor to compete for the manager’s business, what incentive is there for the vendor to offer a good deal?

For network managers, knowing exactly what they want and how it will perform in their network can make a big difference. Tolly noted his firm had recently completed some price/performance tests of vendor gear and found the results varied significantly.

For instance, the cost of a gigabit of throughput on a gigabit Ethernet Layer 2 connection ranged from a low of US$1,249 to a high of US$3,500. The difference for fast Ethernet was even more marked with a low of US$956 and a high of US$3,200.

The bottom line is unless network managers know exactly what they want their networks to do now and into the future, what’s available and what they’ll need to do the job, they’re either going to end up paying more than they should, or they’ll end up with a network that doesn’t do what they want it to do.

Savvy managers need to do their homework, pick out several contenders and bring them in to fight it out head-to-head on the managers’s home turf. This guarantees solid competition which should lead to lower prices. It also ensures any new gear being brought on-board will work with existing networks.