I’ve been an analyst now for a little over a decade. Prior to my role as an analyst, though, I held many technical positions, including a long stint as a networking engineer. When I was an engineer, my favorite trade shows were Cisco Networkers and Networld + Interop. The value of Networkers was to get a view into all the new stuff Cisco was coming out with, and N+I was key because if you ran any kind of multi-vendor environment, this was the show where all of the interoperability testing was done.
For those of you reading this who are still youngsters, networking wasn’t always this simple. Today we’ve standardized on IP and Ethernet so there’s no real protocol decisions to make when building a network, but back then there were some raging debates. Was 16 MB Token Ring really better than 10 MB Ethernet? Was FastEthernet a game changer? How well do the various vendors interoperate? How do you get IPX and IP to co-exist on the same network? There were plenty more questions, and there was a real need for a show like N+I. Over the years, though, we wound up with more and more industry and de facto standards and Interop turned into more of a marketing show. The show still has value, but it’s more around seeing new products and meeting with people and less about the technical side of networking.
Networking today is a state of change, however, and that has created the need for Interop to return to its roots and become the place to solve multi-vendor and/or interoperability issues. While the show isn’t going to solve all of the worlds network problems, there are some logical places to start, including the following:
- Network fabrics. Conventional wisdom is that Spanning Tree Protocol (STP) is long past its prime and needs to be replaced. Replaced with what, you may ask? Well, that depends on which vendor you talk to. Arista is using VXLAN, Brocade and Cisco both support TRILL, and Avaya, Alcatel-Lucent, and Huawei have adopted shortest path bridging (SPB). To date, there has been very little interoperability testing done except for the joint testing ALU and Avaya did about a year ago to demonstrate multi-vendor SPB. Ultimately the testing belongs with an independent body, like Interop.
- CloudsStack versus OpenStack. These are two competing orchestration protocols. OpenStack certainly has more media presence, but it appears that CloudStack is favored among telcos. Some vendors even have proprietary orchestration protocols. If we’re going to move into this world of fluid IT resources, a common orchestration protocol would certainly help accelerate the journey there. Independent testing of the two would be of great value to network professionals today.
- Programmability of the network. OpenFlow, OnePK, Python, EOS, iRules, Junos scripting… the list of ways to program a network device is continually growing. Some vendors offer APIs, others offer scripting tools, and some have developed communities to support the effort. The number of options available makes making a decision on which vendor or approach to use very difficult. Let’s come up with some standardized tasks and challenge the vendors to respond accordingly.
- Video Interoperability. According to Cisco, video is the new voice. Personally, I’d amend that statement — it could be the new voice but interconnecting systems from different vendors (sometimes even from the same vendor) is difficult, if not impossible to do without custom coding, gateway, or other form of video “middleware”. And the problem isn’t getting any simpler with the rise of WebRTC, HTML 5 and other forms of web based video protocols.
Networking is certainly going through a number of transitions and these transitions bring change. It’s great to see vendors do their own tests, but the industry would be best served to have an independent body manage this process and there’s no better place to demonstrate this than at Interop.