Beyond Datacenter Geography: Cloud-era Careers

Last week, “Datacenter Geography” observed that it now often makes sense to site new facilities in places, like Oklahoma or Ontario, that seemed exotic to the IT (information technology) world of just a few years ago.

The destination that truly is dominating plans for the near future, though, is … nowhere. DevOps, get comfortable having your hands only on virtual assets.

That’s much of the message, at least, of “Data center facility die-hards versus the cloud“. Christina Torode makes the case that cost and speed are sending “large enterprise IT departments” to the cloud.

Deep, down-to-the-metal expertise will always have its place–but that place is shrinking. Sure, there are inefficiencies in the cloud, but all the scale advantages are there. Moore’s Law and relatives tell us that the balance will only tip farther in the future. Those of us comfortable handcrafting a loaded rack, or rationalizing the color scheme for a wiring closet, are mostly better employed designing and implementing higher-level architectures.

The “speed” that Torode writes about is business speed: how quickly can IT provision assets for a new project, re-site BC (business continuity) processes, spin-down and liquidate unused infrastructure, roll-out short-term mobile apps, and so on. This kind of nimbleness and agility is at more of a premium now than the networking latency or spindle speed that technicians generally have in mind when discussing “performance”.

Cost and quickness aren’t cloud’s only strengths: there are plenty of reasons to agree, for example, that “Cloud providers will be better at security than you can ever be“.

None of this means, however, that DevOps will disappear. It’s simply time for us to operate at a higher level–to think more abstractly about computing resources than we did when we expected to have a “personal” relationship with every motherboard and disk array that came through Receiving. It’s also time to tackle tough challenges like matching cloud’s scalability advantages to the difficult requirements of marginalized co-workers operating out of branch offices, in the field, or other unusual computing environments.


  1. Raman Garg says:

    @Cameron, very valid points. We are seeing the trend where # of virtual servers are growing at a significant clip across enterprises, whether on the public or private cloud. So your data center may or may not be managed by you, and job is changing a bit, work is significantly more. Data center management software like device42 and automation tools like puppet, chef are being deployed more and more to help with this growth.

    • Cameron Laird says:

      Raman, you reasonably write that “… automation tools like puppet … help with this growth.” For me, they’re essential to management of these changes–it goes beyond “help”. I’d been planning to write about DCIM in 2013; it occurs to me now that there are more business issues involved than I was early calculating.

      Raman, what’s the incidence in your experience of hybrid operations? Do you see organizations “mixing it up” with a combination of public cloud, traditional datacenter, private cloud, …?

      • Raman Garg says:

        As we are talking to customers, we realize that the approach to hybrid operations varies quite a bit. Budget, compliance, security(or perception thereof) are some of the factors here.

        For example, one user have added some critical components to their operations to the public cloud for DR/BC and still manage traditional data center. Another user has compliance issues and reservations to go to public cloud. So they are migrating some of their QA/Development environments to the public cloud to get their feet wet.

Leave a Reply

Your email address will not be published. Required fields are marked *