Ishaya gave a talk on OpenStack this morning at LISA. He opened up with
a history of how OpenStack came to be, then moved into a summary of
each of the components.
OpenStack has experienced a very rapid growth from Big Idea to it’s current state. The OpenStack Compute portion had its first line of code written in April of 2010, a bare 32 months ago. I has moved from a in-house project at NASA, to a LLC owned by Rackspace, to it’s current parent organization in the OpenStack Foundation. In that time it has attracted several very large hardware makers, who are writing plugins to make networking and storage infrastructure more manageable for OpenStack.
This integration is central to the OpenStack vision. They envision, much like Vint Cerf mentioned in his keynote, a world with a standard for Cloud services which enables interoperability beyond anything we see right now. It’s a bold vision, and one being pursued by other groups as well.
They’re focusing on a 6 month release cycle, with the next iteration due out in April of 2013. They’ve only just reached the point where they’re talking about how to handle end-of-life code. It’s a tricky problem.
There are a few parallels between OpenStack and the CloudStack centric training sessions earlier this week. Both projects view their GUIs to be syntactic sugar on top of the real interface: their APIs. No one who is scaling beyond a few boxes will be using the GUI for anything but the broadest of metrics, all the guts will be handled through API calls.
Both projects are fully embracing programmable networking. OpenStack has an edge here since the major hardware vendors are involved with the project, but the same concept is being pursued by CloudStack. Networking as we know it is in for radical changes in the next few years, and storage isn’t far behind.