- You should never admit that you don’t watch Star Trek.
Ever. But you especially shouldn’t make this admission when you’re at a Star Trek-themed party taking place during an OpenStack Summit. It’s a conversation stopper. Your Star Trek-loving companions will be so taken aback by this confession they won’t know what to say to you about it. There will be an extended awkward silence. But then (and this is when you’ll realize what an epic mistake you’ve made), they’ll take it upon themselves to recount everything that happened in all of the episodes that you missed (which, as implied by “I don’t watch Star Trek,” means ALL of them). There will be talk of tribbles and The Borg and you’ll have no idea what is going on. Just nod and smile and remember to lie convincingly next time you’re asked about it.
- Containers are all anyone wants to talk about right now (well, besides Star Trek).
They can’t help themselves. I mean, in four solid days of breakout sessions, I did not attend one that didn’t mention containers at some point—or (more commonly) dedicate 10 or 15 minutes to them. And these are the sessions that weren’t even meant to be about containers. If you really wanted to get into them, there were at least one or two sessions a day dedicated exclusively to the subject.
What’s interesting is that the vast majority of attendees seem to like them—or love them (“They’re so easy to deploy! They take up so little space! They make applications portable!”)—but no one seems sure exactly what is the best way to leverage them within the OpenStack ecosystem. Yes, Magnum looks like it’s going to address most user needs (click here to see a Magnum demo), but I heard of a number of different ways companies are using containers, and there was vigorous debate about which was best. Stay tuned for a better answer on that one.
- Most companies still haven’t figured out what to do with their legacy apps.
And there’s no easy answer. Do you leave them alone and let them continue to exist as part of a bi-modal IT system, or do you rewrite them to be cloud-aware? Bi-modal IT has certainly gained mindshare as an option in the wake of some recent Gartner research, but to go that route is to forfeit all the goodness that comes with cloud-aware application design. Cloud-aware apps offer massive scaling, the ability to burst, and high resiliency. Plus, they’re secure, location-independent, and elastic.
So what to do? A panel of enterprise IT leaders (that are currently wrestling this very question), readily acknowledged that there is no single answer that will work for every company, but they did have a few solid suggestions for those looking to get started:
- Educate your developers: Share the characteristics of cloud aware apps, create design patterns for devs so they have examples, host internal hackathons to convert existing apps to cloud-aware, and hold tech talks that explain what OpenStack is.
- Have compassion for your developers: A lot of things are changing for them right now. They’re doing agile design and CI/CD, and now we’re asking them to look at the cloud aspect of things as well. There are a lot of new tools to deal with. Show them that these things can make life easier for them rather than trying to force them to do it. Think carrots, not sticks.
- Make sure all the new applications your team is working on now are, in fact, cloud-aware so you don’t have to go back and rewrite them later.
- Prove that some legacy apps can successfully be rewritten: Choose an entire legacy app, rebuild it quietly, and use it as a demonstration for your executives. Show them what you were able to accomplish. Once they’re excited, momentum will grow quickly across the infrastructure, operations, and development teams—and funding for the project will follow (at least that was the experience of the panelist that made the suggestion).
- Change parts of an application to make it more resilient: If it’s clear that you can’t rewrite an entire application, consider rearchitecting parts of it. For instance, one of the panelists said her company wants all their apps to run on all devices, so they’re writing HTML 5 front-ends to work with traditional back-ends. The result is not a fully cloud-aware app, but one that delivers a much better end user experience than it had in the past.
Great point regarding Legacy Apps – I’m a big believer that the term is abused slightly and many more applications then we think can be moved to OpenStack based deployments then is given credit. It might be as you comment as simple as splitting out the tiers and running in a hybrid deployment mode, or challenging the ISV to port, the later likely faces investment resistance.