Pick a market – LAN switches, VPN access servers, Web switches. As soon as a solid market develops, aggressive players come out of nowhere and try to steal market share from market founders. These market crashers mostly start out with at least one potential strike against them: Will their products work with existing gear?
Prospective customers always consider this issue. Without proof of interoperability, the suspicion lingers that the devices might not work. Incumbent, market-defining vendors are in no hurry to prove that third-party gear works with their products. Surprisingly, though, the aggressive wannabe vendors seem oblivious to the benefits of proving interoperability.
Our recent Tolly Group/Network World LAN Switch Interoperability Study epitomizes some of these issues. For starters, the study – now in its second year – wasn’t instigated by the insurgent vendors but rather by The Tolly Group and Network World on behalf of network managers. This illustrates that proven interoperability is not high enough on the priorities list of switch vendors.
While it can’t be said that the studies have had massive participation, they have had respectable support from established and start-up vendors. In two years, 10 vendors have entered more than 15 Layer 2 and Layer 3 products.
What do customers think about interoperability and how does our study support or undermine those notions? Many customers believe that if vendors support the same standard, then they should be interoperable. In virtually every case where tests were made of basic standards, the switches worked.
On the other hand, our tests also succeeded in disproving this logic as well. We had to drop a test of industry-standard IEEE 802.3 flow control because we couldn’t get a consensus from the participants on how to devise a legitimate, multiproduct test.
In this case, the standard can be interpreted legitimately in such a way that even though A conforms to C (the standard) and B conforms to C, we have cases where A and B don’t work together. Legitimate implementations of the same standard are “unequal.”
The network manager can posit that in deciding whether to bring “foreign” gear into the organization. It will probably work, but might not. The incumbent gets the advantage in such a situation.
Cisco Systems Inc., the universal incumbent, did participate in the test last year and this year. For that, Cisco should be commended, especially because it has no strong motivation for promoting third-party interoperability.
Ultimately, things have to work. Unless aggressive switching newcomers can prove their gear can work with incumbent products, the incumbents will always have the upper hand.
