Today the network operator industry descends on San Diego for NANOG 66. Carriers, vendors and other industry players will convene to share insights with one another on how to run a massive global IP network, as well as get a much needed shot of Vitamin D from some early Spring California sunshine.
Standing here at NANOG reminds me of the last time I was in America’s Finest City when I got to pay another visit to San Diego’s world-renowned zoo. I’m sure that carriers and business customers attending the conference looking for DCI solutions will find it’s every bit as noisy and colourful as a wildlife exhibit. And the similarities don’t end there.
At the zoo you can see the cheetah who runs at 70 miles an hour. At NANOG you’ll find a vendor or two who moved lightning-fast to ensure their DCI-optimized platform was first to market. But remember, cheetahs are only quick for short bursts of speed. Eventually they are easily outdistanced by the ungulates they chase. Similarly, point-products rushed to market cannot solve the needs of end customers who are in business for the long run. Only a solution that’s available today and scales indefinitely for the future will ultimately win the chase.
Some equipment vendors are treating the DCI space like a symbiotic species, like egret birds riding on the backs of hippos or rhinos. They couple a few DCI-like features with their router bundles: Buy a hippo from us, and get your egrets for free! But DCI customers deserve more than to ride on the backs of their router brethren. The DCI space is unique and it needs dedicated products optimized for the application.
Then you’ve got your bottom-feeders who focus on a race to the bottom with no value add. These vendors cobble together bent sheet metal and pluggable optics and leave it to the customer to make it work. With no need to support integration testing, services or warranties, they can afford to underbid everyone else. But buyer beware, as there is more to a product than just metal and silicon.
In recent years, as the spend of data center operators has increased, so has the presence of vendors hoping to catch their attention. I chuckle to think the scene is not unlike the sea lion exhibit, where the hungry animals waddle around flapping their flippers and doing tricks like balancing balls on their noses, all in hopes of being thrown a fish or two. (Fair disclosure: I like fish too.)
And that brings us to funding. Plenty of zoos struggle with underinvestment and insufficient revenue to cover debts they have accumulated. Simply put, they hit problems when there are more animals on display than customers coming through the gates. Likewise, the DCI space is going to find that the number of customers out there simply cannot support the rapidly rising number of equipment vendors. And then there will be a culling of the herd. Just like at the African watering hole and across the desert plains, the winners will be the species most evolved with attributes for long term survival.
And what might these attributes be? DCI features like sufficient client granularity with support for 10, 40, and 100Gbit/s interfaces of multiple types. Sufficient network side flexibility with support for 100, 200, 300, and 400G coherent interfaces. The indefinite scalability that line cards and non-switched backplanes provide. A truly open architecture, rather than pretending to be open. On the hardware side, that means the transponding and optical line shelf functions can be separated and seamlessly intermixed with other vendors’ solutions. Or they may be all installed in a single chassis for “system-in-a-box” usability advantages. On the software side, that means CLI, REST, NETCONF, RESTCONF interfaces, and the ability to be controlled by any hypervisor from any vendor, whether OpenFlow, OpenDaylight, or OpenConfig.
Yes, chances are there will be more than one zoo in San Diego this week, and both will be entertaining. And if you see me at the show, balancing a ball on my nose, feel free to throw me a fish or two.