Like many of you, I've been tracking the formation and evolution of OpenStack as a potentially viable approach for building clouds.
While the OpenStack momentum appears to wax and wane over time, one thing is clear: many people find the design principles and philosophies behind OpenStack quite appealing, myself included.
Not to mention, it all can be quite entertaining at times :)
As part of the VMware/EMC/Pivotal federation, I routinely get asked for the internal perspectives on OpenStack. I think people might be expecting a healthy dose of FUD, but instead they hear quite the opposite: a strong pattern of support, endorsement and investment in the evolving OpenStack landscape and community.
Digging deeper, it's clear to people that Pivotal should be cloud-agnostic: they play higher up the stack. EMC has long supported customer choice (hypervisors and cloud frameworks included), so that's not entirely surprising either.
But there's some visible head-scratching when we start digging into VMware's extensive support for OpenStack. It seems counter-intuitive to most people.
That is, until I explain the motivations.
Patterns Repeat Themselves
From my perspective, OpenStack is evolving on roughly the same trajectory we saw with Linux over a decade ago. Great technology, open source models, a vibrant community, many distributions, etc. My inevitable conclusion is that OpenStack will eventually evolve to hold a similar role in enterprise IT: a key component of many landscapes.
But there's a difference with OpenStack: its real strength is as a framework for integrating and orchestrating all the different technologies that go into building and running a production cloud.
Yes, there are default open source components for much of the environment, but it's also assumed that implementers will want to mix and match -- sort of like running Oracle on your Linux server.
For most enterprise IT shops, the ultimate appeal of the OpenStack approach is clear: the importance of being in control of your own destiny. But it's not as easy as it might look. I think it's fair to point out that -- at this point in time -- OpenStack is attempting to "cross the chasm".
There are plenty of pilots around, but few production environments to study.
I think that part of that has to do with the technology components found in OpenStack: some of them not quite mature yet, many of them incompatible with existing production environments. And that is where VMware fills a few gaps nicely.
The Recent News
There is a significant OpenStack effort at VMware these days -- it's not window dressing, it's deep and substantive.
The most recent evidence comes from VMware's announcement supporting Havana, the latest OpenStack release.
In addition to the substantial role that Nicira (now NSX) has played in the Neutron networking project, there's also announced integration support between vSphere and Nova (the compute controller), and a new Cinder storage interface that supports the VMware VMDK format. The latter opens the door to potential use (among other things) of things like VSAN as part of an OpenStack cloud.
And, of course, the DynamicOps-based VMware management products have always been comfortable with OpenStack, AWS, etc.
Behind this, VMware has established a customer-facing OpenStack team that's helping to integrate VMware components with specific OpenStack cloud projects: the all-important required technical firepower in front of customers.
While all this is good, there's even a more telling announcement ...
Those of us who are following this space were surprised (and more than a bit amused) to see the partnership announcement between VMware and Mirantis around integrating VMware technology into OpenStack clouds. Without getting into too much detail, it's safe to say that the principals at Mirantis have gone on record multiple times opposing VMware's involvement in OpenStack, including this famous post.
What changed? Boris Renski (co-founder of Mirantis) points to VMware's extensive contributions to the community, among other factors.
While that's inarguably true, I think there's a more powerful factor at work here: increased customer demand.
The Customer Perspective?
Let's say you're a decent-sized enterprise IT shop, and let's also assume that this isn't your first rodeo: you've built and operated successful clouds, and you know what you're doing.
You've worked through the integration challenges, new operational and financial models, etc. and you're pleased with the results. So maybe you're ready to take on something a bit more technically ambitious for your next cloud, and you decide to build something from OpenStack, knowing full well that you're signing up for additional technical effort.
While there's a lot to like in the various distributions, you quickly realize that not everything is ready for prime time quite yet. You find yourself wanting to use familiar components to not only improve the quality of the results and lessen the technical effort, but also provide some modicum of compatibility with existing environments.
Considering vSphere in that light seems reasonable, and the same argument might be expanded to network, storage, security, data protection, orchestration, etc.
From a strategic perspective, it's not a bad choice to mix and match proven components in the new framework. Thanks to your choice of OpenStack, you feel better about being in control of your destiny, albeit with additional integration and support effort.
I think we're now at a point in time where there's a critical subset of enterprise IT shops that fit this profile, and that's generating interest. And the good folks at Mirantis (among others) want to help.
The Road Ahead
So much of what you read in the enterprise IT world sounds almost religious, especially when it comes to open source. Sometimes I think they must be passing out Ken Kesey's kool-aid at these events.
But most enterprise IT professionals lost their religion years ago, and have become very pragmatic in dealing with the real world of impossible requirements, imperfect technology vendors and limited resources.
They make their choices carefully, because they'll have to live with them a long time.
In this context, more than a few have decided to make initial investments in OpenStack: if not for production, certainly to gain experience.
And -- if anything -- I think VMware's investment and support is giving them attractive new choices to consider as they move forward.
Like this post? Why not subscribe via email?