One of the things I was pretty pleased about at HP Discover, was HP’s focus on OpenStack.
OpenStack is a tech that I’m pretty damn interested in. And along with Docker is a technology that I think will be massively influential in the next few years. May be even as influential as Linux has been in the last ~15 years – though I totally appreciate that’s a big BIG ask.
HP Are All Over OpenStack
Anyway…. For good or for bad, HP seem like they’re all over OpenStack. I mean they’ve even have own distro (of sorts) in HP Helion OpenStack.
But the messaging from HP – according to my interpretation – is that HP want to be for OpenStack, what Red Hat are for Linux. So the enterprise version of OpenStack, or the OpenStack company that customers go to when they need a rock solid enterprise-class version.
And to that end, they’ve piled an absolute ton of cash behind it – to the tune of a billion dollars in R&D over the next two years. Not only that, they’ve also got *something like* three quarters of the OpenStack technical leads working for them. So that part of the message is crystal clear – HP are employing key figures at the heart of OpenStack, and are committing code and cash to it like crazy! Check out the stats below (only Red Hat is more committed – pun intended).
(image source http://activity.openstack.org/dash/releases/ – June 2014)
HP have also built their own HP Public Cloud offering on top of OpenStack. So they know a thing or two about deploying it, and at pretty decent scale too! And they’re hoping to leverage that experience when it comes to helping customers deploy (but that assumes that the guys who work on the HP Public Cloud service actually talk to the guys in the rest of the organisation, which isn’t guaranteed in an organisation like HP).
And that’s all well and good. But…… it’s one thing to deploy OpenStack when you’ve got an army of passionate developers who know it inside-out and are working on it full time. It’s altogether a different thing to deploy it on a shoe-string-budget with a bunch of developers and IT guys who also have to concentrate on running the core business. Sure, HP are looking to wrap enterprise grade services around their OpenStack Helion, and that may help customers…. but… in my personal opinion, we’re still a ways off it being ready for production in the enterprise (and by that, I mean running it in live production environments at large established companies).
Ticking Time Bomb
One of the biggest challenges I see to OpenStack deployment in the enterprise is the rapid clip at which new releases are published – every six months a new major release comes out. That’s gonna be a challenge for enterprises and large organizations. And it’s not helped by the fact that there’s currently no upgrade path.
Then there’s this small ticking time bomb…… HP will only support a particular release for 18 months!!!! Seriously.. only 18 months!
That’s no time at all at a large company! And will be a major barrier IMHO. It’s like a ticking time-bomb…. You deploy on day one of release (tick tick tick) and get 18 months support (tick tick). When that 18 months is up (tick tick tick) there’s no upgrade path other than a fresh clean install (BOOM!!!). You’d have to be insane right!?!?!?
I was personally expecting HP to take the same route Red Hat have traditionally taken with Linux – take a stable kernel and run with it for a long time. For example, up until earlier this week, the latest version of Red Hat Enterprise Linux (RHEL) was RHEL6.5, which was based on a 2.6 kernel from 2010 (that’s a ~4 year-old/~44 month-old kernel, during which time we’d get nearly 8 OpenStack releases). And this was despite the fact that the mainline and stable Linux kernels were at around version 3.14 – it’s not like the Linux kernel is slow with the changes! Oh and of course Red Hat would backport important features to the 2.6 kernel if deemed necessary. And that, in my opinion, is how enterprises work! They
want need rock solid reliability and long term support. That’s what I was expecting from HP on the OpenStack front.
So I personally see this scarily short support deal being a major sticking point for enterprises. That said, Red Hat are the same with their support of OpenStack – they’re only offering 18 months too. Hmmmmmmm.
And I get that 18 months is an age in the world of cloud, where new stuff is coming out all the time. But……. for an enterprise, 18 months no time at all. I bet most enterprises would probably take 18 months to get it installed!!
And it would look a lot different if there was a smooth and simple upgrade path between versions, but that’s not the case. And although HP say they’re working on that…. every developer I know wants to work on cool new features, and not boring stuff like simplifying upgrades. So I’m not holding my breath….
Anyway….. All of that complaining out of the way, like I said, I think OpenStack has a bright and influential future ahead of it. And HP getting in relatively early is a great move. They really look to be positioning themselves well for an OpenStack future. And I’ve gotta give them kudos for that!
Right now though, I see a gap between it’s production readiness, and the kinds of steady-as-she-goes customers that normally partner with HP.
Still…. the futures bright, the future’s OpenStack!