Is it too late for OpenStack?


There’s been a lot of news about OpenStack recently — notably a conference dedicated to the open-source cloud-computing platform this week and IBM and Red Hat  signing on to the effort. And yet there is a feeling in some quarters that  it may be too late for the project to take hold.

Two years after Rackspace and NASA launched OpenStack in part to counter Amazon Web Services, AWS keeps getting bigger and broader — with new, increasingly enterprise-focused services coming out all the time. There also is fear — even among some OpenStack proponents — that too many cooks might spoil the effort. Sure OpenStack could become the “Linux of the cloud,” but it could also get fragmented as each vendor adds its own secret sauce to the OpenStack underpinnings. The downside scenario is that OpenStack ends up more like Unix than Linux.

Taking on AWS and VMware isn’t easy

AWS clearly took note of OpenStack ambitions to take on public, private and hybrid clouds, making a series of moves to make AWS more palatable to businesses that are wary of putting their data and applications into a shared public cloud. AWS’s API alliance with Eucalyptus — an open source private-cloud player that competes with OpenStack — is one example. AWS’s Storage Gateway and Simple Workflow Services are two others

Robert Shear, president of Greystone Solutions, a Boston development shop, is a big AWS user and sees no reason to change that. “Maybe OpenStack will be the ‘Linux of the cloud’ but every vendor playing catch-up is in favor of open standards. The reality is that each vendor selection is a commitment. The nature of the AWS stack is that the customer has a wide range of choices when deciding portability versus functional capacity. And, with AWS, the benefits of scale and cost-leadership are palpable and immediate,” Shear said via email.

VMware is another target for many of the OpenStack players, who hope to use the cloud platform to prevent the virtualization giant from replicating the power it holds in server rooms to the cloud at large. Hewlett-Packard’s new OpenStack-based public cloud relies heavily on KVM virtualization, for example.

And many see Citrix’ CloudStack move — the former OpenStack backer recently launched its own competitive project  within the Apache Software Foundation — as being motivated by Citrix’s desire to push Xen as the virtualization of choice in the cloud. Citrix points to slow OpenStack development and an unwillingness to embrace the AWS APIs as reasons for its decision to launch the competitive effort, but says it’s far more concerned with competing against VMware than with OpenStack.

The Amazon API quandary

Whether you think OpenStack is done before it gets started also depends on whether you think AWS’s APIs are the be-all and end-all of cloud computing. There is healthy debate on that topic. Some in the OpenStack camp, including Rackspace, don’t think Amazon API support is the name of the game. In an interview last week, Rackspace CTO John Engates said Rackspace’s OpenStack implementation does not support the AWS API set nor should it.

“We don’t need to be compatible with the AWS APIs,” he said. “OpenStack itself does have some compatibility with the Amazon APIs but Rackspace is not exposing that compatibility in its public cloud … mapping to AWS APIs restricts innovation. If you have to wait for Amazon and reverse engineer [what it does], where’s the innovation? We want to spur innovation.”

Still, even those interested in OpenStack wonder how it is that Rackspace is still not offering a production version of the platform. As Keith Townshend commented on a GigaOM story on the Rackspace implementation: “After 2 years in the making, you would think they [would] have a product ready for beta. This market will not sit back and wait for the ‘good guys.” He later clarified: “I don’t think it’s too late, but it is late … Most are late including CloudStack. If I needed a solution today, what are my real options? I know it’s not vaporware but it feels like it.”

Randy Bias, CTO of Cloudscaling, another OpenStack player, said the talk about production or not production is a tempest in a teapot. For him, a production system is one that is up and running with customers’ applications even if it is in “beta” or “trial phase.”  If defined that way, he would characterize HP’s existing cloud, AT&T’s private OpenStack-based cloud as well as Internap and other OpenStack implementations, as being in production, Bias said via email.

Enterprise cloud inertia helps Openstack

The fact that many businesses are still kicking the tires on cloud plays to OpenStack’s favor. For some companies, the specter of one big player — AWS — dominating key cloud functionality and services is disturbing. They weren’t wild about Microsoft’s dominance on the desktop and they certainly don’t want to see that situation replicated in the cloud.

OpenStack players are doing their best to play on that fear. This week, Rackspace CEO Lanham Napier told CRN that “Amazon’s proprietary system cultivates customer lock-in.”

All this means is that there’s still a window for OpenStack to succeed — if the various parties get their clouds out in good shape and don’t fracture the code base.

OpenStack “definitely still has an opportunity because so much of the market hasn’t moved [yet],” said Forrester VP and Principal Analyst James Staten. And, he notes, the market wants open source options just as it wanted an open-source alternative to Microsoft Windows. “While [OpenStack] may still not be ready for production they are definitely getting closer,” Staten said.

The problem is, he added, OpenStack’s perceived tardiness “leaves open a door Eucalyptus and Citrix will be more than happy to drive their trucks through.”

Photo courtesy of Flickr user Earls37a

You're subscribed! If you like, you can update your settings


Comments have been disabled for this post