Regardless of where you stand on the future of data networks and SDN and the prevailing idea that dust made from ground up rainbows will be powering our pipes – there is something that I believe is being overlooked and should definitely be addressed.
There is a rate of unprecedented change going on in the world of technology. Every day we’re cramming more and more transistors into a square millimeter and pumping more and more feature sets over these circuits. Enterprises and consumers are demanding more from the technology that we interact with and setting never before seen expectations of the way of ‘always on’. This is great news for anyone with a job in the industry, its job security. When people want data, they want it now and they’re not happy when they don’t get it. So we’re here to give them what they want and when they want it! Except therein lies the problem. Someone has to be there to give it to them.
The hottest topics in the networking world today are SDN, and automation and what implications these have on our distributed systems of today. There’s a lot of heresy as to how this will all unfold over the next few years.
I would like to take a step back and talk about logistics. Talk about the fact that Cisco is touting that there will be 250 connections, per second, to the internet by the year 2020. That’s a big number. A 150% increase to the numbers of today. Almost staggering when looking at it through the eyes of an engineer that has to capacity plan for something like this. And not necessarily capacity plan from an infrastructure perspective. My biggest concern, is man power. The work force is rapidly shrinking and there are no bodies to fill seats. This is a bigger problem, I think, than facing how to connect such a vast number of devices. Not having people to do the actual connecting, negates having devices to connect. This brings me back around to SDN and automation, or what I like to call the networking industry’s abstraction phase. Currently we as network engineers tend to toil away changing VLANs on interfaces and piping output from configured devices planning for our next designs or changes, and that’s worked for the last 20 or so years. But this tidal wave of connectivity coming down the pipe is going to be a tsunami the industry isn’t ready for unless we take the proper steps to eliminate tedious tasks that typical engineers do on an every day basis. Simply because 10 years from now, we won’t have enough people to perform those functions at the scale that they’ll need to be done.
This abstraction phase is going to be how we are able to do that. Allowing a controller or APIs to handle the every day tasks of modifying where and when a flow can propagate within a network will free up our limited staffing to design the next generation connectivity. With that abstraction will come a tool set that will allow us to automate away the simple tasks and create a platform for us to deliver the next generation of services on. I don’t know about you, but I could do without modifying the allowed VLANs on a trunk and do more with deciding what types of services I can build to ride that trunk.
Deciding how the traffic propagates through the network should be as much of a worry to us as a software engineer worries about how the raw I/O scheduling works with the CPU on the platform that they’re designing a piece of software on. Read, it doesn’t. That abstraction will give us the flexibility to take connectivity to the next level. Intelligence doesn’t even begin to explain what could come of this next generation of connectivity, but I can’t wait to see what 2023 will look like.