If you do look at many of the predictions being made you’ll notice a few common themes that I tried to capture here:
I’d like to re-focus this and say, data center networking is changing. Done. That’s the real trend simply put.
If you have extra time, keep reading to see my first rant of the year pertaining to networking of the future.
Dear Cloud. Thank You for this opportunity.
Cloud is driving this change. Cloud needs automation. Clouds are large. They are virtual. What does the network need to look like to support the evolution of the cloud, furthermore to offer the same network services in the public cloud as in your local data center? The network needs to also exist in software. It needs to be virtual. With a virtual network, does this downplay the physical network? Not at all. Stop thinking it does. We still need to map logical virtual networks to physical networks just as is done with compute resources today. If links are being oversubscribed, then particular flows should be moved. Can we call this DRS for the network or DRS-N? Combining these worlds will be key, but the root of this all may very well mean simple layer 3 physical networks with overlays riding on top. This has been talked about for a while now (thanks to Nicira). For those that can manipulate the physical network based on the virtual network needs will be able to provide significant value.
When you were growing up, didn’t you want to learn new skills?
As we enter 2013, many are interested in the skills of the future network engineer. They will evolve over time, but the good news is if you are searching about this today, you will likely not have to worry about it in the future because you are already ahead of the curve. Network Engineers need to understand how large scale systems environments are being automated today and ask themselves, how inefficient are their daily routines compared to their peers in the systems world. It’s not a matter of systems folks taking over the network or vice versa, it’s about being able to take a step back and analyzing the best way to operate a network, or better yet, operate a data center or cloud. This can be a large or small data center. Can a hypervisor manager be used to manage a virtual network? Of course. Can Linux tools be used to automate a physical and/or virtual network? Sure, why not? If so, does that take away from the network protocols being used? Not at all. Network Engineers will still be needed. The only thing that would be different is the way the network is managed and the UI that is presented to the operator. Embrace the change. Expand your horizons. That’s all you can do is to keep learning. Okay fine, the CLI will likely be around for years to come, but what’s the downside of learning new interfaces?
The Network Industry is Pregnant. SDN is in the womb.
Some IT veterans say SDN is still 3-5 years away. 3-5 years away from what I ask? That’s the question. Are you afraid of not being needed anymore? Do you fully understand the potential of a network being software defined or even true network automation (without SDN)? If it’s 3-5 years of really seeing negative impact on major incumbents’ bottom line, then I can agree with that, but next generation networks are here now. Any individual or company still has time to re-invent themselves to take advantage of this shift in networking to realize reductions in OpEx, CapEx, and potentially to even see new revenue streams. If you don’t think SDNs are here already and won’t be for years to come, that’s great and better for the revolutionary companies and individuals. Maybe you are a network manager that won’t take meetings around SDN because you’re too busy fighting fires. Maybe you’re an engineer that loves the CLI and think applications don’t have a place in the network. Maybe you’re a tech and don’t think anything should change because you just got your first job as a network admin. I’d be willing to bet that with the acquisitions we’ll likely see in the next 12-18 months, many perspectives on SDN will change. Start-ups just don’t seem to get real “street cred” across the Enterprise until they are acquired.
Had many of these thoughts in my head for a while, so I just wanted to get them out for now, but I do plan on re-visiting the topic of skills for the future network engineer in upcoming posts.
Happy New Year.
Follow me on Twitter: @jedelman8