the future of networking

Predictions are dangerous things. We — humans, that is — are notoriously terrible at predicting the future, and continually get caught out by retroactively “obvious” developments.

Having said that, I’m reasonably comfortable that software-defined networking has a place in the future world of network equipment, and that as a result, the role of a network engineer will morph from being skilled in one or more vendor platforms to being skilled in one or more vendor platforms and a programming language, such as Python. In fact the best network engineers will likely have a role that looks more like business analysis than e-plumber. Python, and languages like it, will likely provide a means to that end.

Software-defined networking doesn’t really deal with moving bits fast: equipment from a range of vendors already does that, and does it pretty well. In fact, SDN doesn’t really provide anything that can’t be done with existing equipment; what SDN provides is (theoretically, at least) a simpler way to provide a given business solution.

I also predict that — as a side-effect of this trend — the average network engineer will have to become a lot more comfortable with commodity operating environments, and in particular, Linux. While SDN controller software runs quite well in a range of environments (that being part of its point, after all!), Linux is apparently becoming a default choice for a range of related environments, including OpenStack and several virtual (and real!) routing and switching platforms.

Comments are closed.