IT Is Power. Try Living Without A Single Breadboard For A Day.

Don MacVittie

Subscribe to Don MacVittie: eMailAlertsEmail Alerts
Get Don MacVittie via: homepageHomepage mobileMobile rssRSS facebookFacebook twitterTwitter linkedinLinkedIn

Related Topics: Continuous Integration, DevOps Journal

Blog Feed Post

If You Look Close, You Can See Ops’ Future By @DMacVittie | @DevOpsSummit #DevOps

We’ll have to wait a bit, there is always a lag between promise and reality, but you can feel the change now

The evolution of IT operations is upon us. There has been a lot of change in both direction and momentum over the last several years, and at this point, it is not hard to see the future of Ops.

Perhaps, having worked for a networking firm and a server automation firm, it is easier to see than for the average operations person, so let’s just take a look at the burgeoning ops trends and technologies and see where they ultimately lead us…

  • APIs for Infrastructure gear.
    • The days of memorizing ten different command line interfaces are rapidly coming to a close. While GUIs exist for most modern infrastructure equipment (switches, routers, load balancers, etc), increasingly every one of them also has a working API. That means they can be controlled at a very functional level from third party applications.

As one example of this trend in useful action, check out AppViewX. A third party “agnostic” management tool for ADCs, Firewalls, Cert Management, and more – through APIs.

  • Automated installation of servers – both physical and virtual.
    • Today there really is no reason that a busy operations team has to manually install every server that comes through the door. No matter what your operating system or hardware choices, this process can be largely – or completely – automated. The number of choices for this automation is larger than you think, and they’re all growing more and more capabilities by the minute. The larger appeal is repeatability – the chance to configure server 1 and server 2 essentially the same, with only differences the Ops team needs.

GearSlideMy own employer sponsors the Stacki project, where enterprises are automating systems installations to free up operations’ time. The family that Stacki grew from is responsible for installing millions – yes millions of servers. That’s just counting one of the solution sets available.

  • IT Operations Management Automation.
    • Better named “Software Configuration Management Automation”, this includes a bundle of well-known products from Puppet to Chef, and several lesser but still impressive names like Ansible and SaltStack. Installing and configuring software is now automatable and repeatable. Another space with a lot of entrants, it too is growing rapidly as each solution set adds features to make Ops’ life easier and grow their market share. This market is serving millions of servers also. Today.
  • SDX.
    • Let’s face it, we’ve gone a bit “Software Defined” crazy over the last few years, but it is reaping benefits in some spaces. Software Defined is intended to imply more agile, and in some cases it is. Let’s just list the ones that come to mind for me – note that these are ones that I have had reason to pay a little attention to, there are more…
      • SDN – Building a software network on top of your hardware network (still need wires and devices, so yeah, that is the definition).
      • SDS – Essentially, unbundling management from the hardware. This is not a terribly new concept, when I was covering storage, it was a market by a different name – Storage Management. But because hardware vendors didn’t offer support, it was a stagnant market (see “APIs for Infrastructure Gear” above)
      • SDDC – take all of the automation available and centrally manage it for the entire data center.
  • Virtualization/Cloud.
    • No need to define these, but they incorporate elements of all of the above. Networking in both VMware and OpenStack is software defined, once the feed is hooked in. Same with storage and spinning up servers.
  • Continuous Integration.
    • Once Dev is done with a given change/set of changes, they go live. Most (though not all) of the time, there is no tweaking required by ops.
  • Containers.
    • Lightweight “sort of” virtualization. Containers will have an impact… But they seem to be too deep in the hype cycle to determine how much. For example, if deconstructing the code into small functional bits and spreading them about the network is the direction containers go, that would seem to indicate a lot more work in the networking space to support. And continuous integration would be upstaged by containers, perhaps not much else will be – only time will tell.

The overall trend that is showing here (and among other up-and-coming trends/technologies) is a move toward less daily operations work.

And that’s where we’re headed. Remember the “self healing network”? It and the servers on it are coming to life, and the rate is accelerating. That’s not to say there will be no need for operations staff… Who do you think will install, configure, manage, and script all of the above? But the role of operations will be changing even more than it has over the last five years. Less reliance on day-to-day configuration tasks, and more reliance on architecture and design roles. Essentially, “Meta Operations”.

And that is a good thing. Some of the network/systems/servers people I’ve worked with were geniuses, and their life was relegated to “Okay, now reconfigure the VLAN…”. Not the best use of those brains, and not the most exciting role for them. Taking away a significant amount of the drudge work and putting them in a position to do astounding things will reap benefits that we are only beginning to see today.

The one side question that we should be considering is increasing standardization. As tools become standardized to manage total automation, there will be increasing pressure on hardware vendors to standardize – interfaces or entire stacks – which will squeeze out specialty hardware improvements. But that’s been going on (though slowly) for a while anyway, so does it matter?

Meta Operations will be a cool thing for ops, though someone will likely invent a better name for it, or orgs will simply use “infrastructure” or “Architecture” Operations to define it. It doesn’t matter what it gets called, we’ll have operations people increasingly capable with automation tools and a wider selection of scripting languages/tools out there helping the organization achieve goals in a more proactive manner. That’s worth applauding.

We’ll have to wait a bit, there is always a lag between promise and reality, but you can feel the change now, and you can see the future if you look. And smart Operations folks will be looking for their preferred place in the new world, making certain they’re still relevant and that the organization doesn’t rush too fast into a future still being made.

But I probably didn’t need to tell you that. Keep knocking out the org’s needs and automating bits at a time, and enjoy the stress reduction that comes from less day-to-day firefighting.

Read the original blog entry...

More Stories By Don MacVittie

Don MacVittie is founder of Ingrained Technology, A technical advocacy and software development consultancy. He has experience in application development, architecture, infrastructure, technical writing,DevOps, and IT management. MacVittie holds a B.S. in Computer Science from Northern Michigan University, and an M.S. in Computer Science from Nova Southeastern University.