It’s Not Highlander: There Can Be More Than One Open Source Cloud
The last few days have been, to say the least, interesting for cloud watchers. If you care at all about the open source cloud, it's been a major week. OpenStack pushed out its Essex release, but not in time to avoid having to share a news cycle with the announcement that Citrix wants to donate CloudStack to the Apache Software Foundation (ASF).
Predictably, the stories have pitted CloudStack and OpenStack as rivals. CloudStack has been painted as a troubled project and Citrix has been painted as "walking away" from OpenStack. The press usually gets the blame for the "us vs. them" stories, but the truth is that companies involved in OpenStack can take some credit here. I had several pitches from OpenStack-backers immediately following the CloudStack announcement, looking to chime in and deride the project. One of the folks I intended to talk to about the OpenStack release couldn't be bothered to schedule a call for that - being too busy pitching press to talk smack about CloudStack.
Truth is, if you paid attention you'd see that Citrix hadn't been very active in OpenStack Essex development in the first place. Then again, neither have a bunch of the so-called OpenStack partners. The OpenStack community claims 155 companies for the OpenStack ecosystem, but only 55 companies participated in development according to the OpenStack press release.
Lighten Up MacLeod
To hear them go at it, you'd think that open source cloud was a zero-sum game. It's a bit like Highlander, except with much better acting and special effects. Seriously, have you watched Highlander? After more than 20 years of hearing references to Highlander and "there can be only one!" I finally broke down and watched the movie a few months ago. I wish I hadn't. But I digress...
Turns out, I wasn't the only one thinking Highlander. I pinged RedMonk's Donnie Berkholz, who has been a Gentoo developer for many years as well. Berkholz says, "cloud stacks haven't done a great job of delineating all the niches in their space – many of them still think they can be everything for everyone. This creates the same Highlander-style perception that there can be only one."
Find Your Niche
Instead of trying to fight to be all things to all people, Berkholz says they should try to "focus on deciding and communicating the missions of each project along with how they can fit together into the broader cloud ecosystem."
This, eventually, happened in the Linux market. In the early days, every distribution tried to fit all niches. That didn't work. Red Hat, for example, finally decided that it really didn't see much point in pursuing the desktop. I hear that's worked out relatively well for them. Novell/SUSE have carved out a pretty good niche on IBM's hardware.
Citrix's announcement should be seen as unalloyed good news for everybody involved. Yes, really. A vendor-sponsored project is going to a vendor-neutral organization that has a track record of doing well by open source projects. CloudStack has a number of deployments and paying customers, and a lot of companies have pledged to support it.
Assuming even one-third of the companies that pledged to support CloudStack become serious contributors, CloudStack should have a bright future. It takes little away from OpenStack for CloudStack to succeed.
Likewise, CloudStack already uses OpenStack's Object Storage (Swift). There's no reason that the two projects can't share components and find ways to differentiate themselves for different markets. (The same is also true for Eucalyptus, OpenNebula and any other open cloud projects out there.)
After all the dust settles, it would be best for everyone from the open source cloud communities to figure out how they can work together. The "there can be only one" approach isn't going to help anybody, and may benefit Amazon, VMware and other proprietary vendors who can market their wares without any drama.