Happy third birthday, OpenStack: Ready to dominate all clouds now?
'A testament to the power of bourbon and unit test coverage'
The inevitable criticism and sour grapes
None of this means that there has not been plenty of criticism heaped on OpenStack software, the foundation that controls it, and the developers who code it.
"The older OpenStack gets, the more I feel like a parent, and I have two kids of my own outside of OpenStack," McKenty says with a laugh. "It is amazing that everyone who is not a parent wants to tell you how to raise your kid. Every single person will criticize what you are doing because it is not their job.
"So this week has been punctuated by 'You're doing it wrong.' And they can't agree on what we are doing wrong, which is what makes this funny. Maybe I am just a parent who feels this criticism personally."
The OpenStack ecosystem of partners continues to expand
The criticism comes from two camps. There are those who are looking for "the POSIX of the cloud" and accuse the OpenStack Foundation of not creating this, which essentially means not making OpenStack a cloud computing standards body instead of product development community.
"A standard is not a product, and it is not even a framework," continues McKenty, and he is on a roll now. "Moreover, the best standards we have ever had were the Internet standards, and the entire Internet RFC process was about taking what was already working and then calling it a standard, not sitting around in a committee four or five years ahead of time inventing something in the abstract.
"And that is specifically what we refuse to do – which is to make stuff up before we build it – with OpenStack. The people who really like to be on those committees are really offended because we don't give them a place to live."
You're not... compatible
The other criticism is that OpenStack should have just cloned Amazon Web Services, and McKenty is having none of that talk.
"This one makes me crazy," says McKenty. "The reason that people wanted something like OpenStack to exist is because they wanted something that Amazon could not deliver. And the example I always give is the first reason we broke AWS API compatibility is because we wanted to be able to have an instance that you could turn off and then you could turn it back on.
"You cannot do that with Amazon. If you turn it off, it is deleted, it is destroyed. Goodbye. Sayonara. Having AMI images you can reload is not the same thing as having a pause button."
The other problem with strict AWS compatibility is that OpenStack is meant to be used in private, public, and hybrid clouds and there will be certain APIs that have to be exposed that do things that AWS simply does not expose externally (even if it might have such functions internally).
The example that McKenty gives is that the corporate auditors have to know the physical location of a VM, and there needs to be an audit trail of where is has been running as it flits around the cluster of servers inside the firewall or hops back and forth between public and private clouds.
The dominance of Rackspace over OpenStack
The whole point of starting OpenStack as far as NASA was concerned was to get the space agency out of the software development business yet give it a cloud controller that was totally open source and that it could contribute to as a means of getting the features it needed. As soon as NASA's cloud coders left to do their own riffs on OpenStack, that left Rackspace in the driver's seat as the biggest contributor and at this point the largest user of OpenStack.
With the establishment of the OpenStack Foundation last September, Rackspace let go of OpenStack and it was then that Red Hat could join up and formally adopt OpenStack as its official cloud control freak for peddling to its Linux shops.
IBM followed suit and is now adopting OpenStack as its cloud controller for both private clouds built from its systems (including System z mainframes, believe it or not) and for its SmartCloud Enterprise public cloud.
"Rackspace is consciously not worrying about whether we are the top contributor in terms of commits to a release of OpenStack, and we are looking at other things such as patches," explains John Igoe, vice president of cloud at Rackspace. "We are also looking at what important projects, from an operational standpoint, that should be started up."
Don't fork this up, man...
Igoe has been involved with OpenStack even before the project was formally started, having spent time with Rackspace executives four years ago when he worked at Dell's Data Center Solutions custom server business. Igoe ran a company called Silverback that created remote network monitoring and management software that was bought by Dell.
He was eventually tapped to handle the software part of this unit, and among other things, he shepherded the creation of Dell's "Crowbar" OpenStack setup utility and became the point person at Dell for hooking into the OpenStack community. Three months ago he joined Rackspace, and it looks like he is eager to assert himself, on behalf of his new employer, inside the OpenStack community. This will probably annoy other contributors and member companies.
"I think Rackspace, since it has transitioned the ownership and the management of OpenStack to the foundation, an independent entity, is in the process of reasserting its leadership position in the community," Igoe tells El Reg.
"What I saw when Rackspace was the major sponsor of the foundation was a little bit of timidness on the part of Rackspace to actually to lead and demonstrate a direction in the community. I think that what we will see in year four is that Rackspace is going to be much more visible. And that is important because the OpenStack Foundation board is facing a number of challenges."
The first challenge is that users are becoming an important part of the community along with developers and corporate sponsors, and their needs have to be addressed. So, for instance, there needs to be a shift from features and functionality being added to making OpenStack a better operational tool, making metering, billing, orchestration work better. And this is something Rackspace execs involved in OpenStack have been saying for the past year.
But Rackspace is also quite aware that it cannot push OpenStack too hard and that it is not in charge of the project any more.
"We need competition in the OpenStack community, and we need cooperation, too," says Igoe. "If we don't encourage competition and really solid collaboration, we are going to develop forks.
"Organizations are going to feel that in order to drive their business and a return on their investment, there will be forks in the OpenStack community where people try to differentiate themselves based on proprietary capability. And that will not be good for the overall growth of OpenStack."
That is always the challenge with any open source project. And only time will tell if OpenStack can hold it all together. Perhaps it is best to just keep the bourbon flowing, then. ®