Diaz points to an emerging IBM-backed Oasis standard known as Topology and Orchestration Specification for Cloud Applications (TOSCA). It's worth quoting Diaz at length, as he explains TOSCA's potential benefits for both customers and software vendors:
It's a way of describing a workload application that's independent of the actual infrastructure that's running it. I'll give you a real example: You can imagine encoding, say, an SAP application or SugarCRM -- along with its data store, its application server, whatever -- its LDAP, it's connections, everything. And you can describe what you need to stand one of these things up, how you start it, how you stop it, how you manage the lifecycle, how you do patches, etc. That kind of stuff.
It's a declarative model. And what that allows you to do is take a workload that's running, say, in and IBM environment and move it somewhere else -- say from a private cloud to a public cloud. What it also allows folks like SAP -- and this is why they're really involved -- they want to have their stuff running in everybody's cloud. But the development costs for them -- let's say that there are six clouds that they care about -- they've got to figure this out six different times and that's expensive. So they can reduce that cost a bit by using standards like this.
After years of covering Web services standards wars, I have to take TOSCA with a grain of salt, although Diaz claims the Oasis group is very active and includes the likes of Google, HP, NetApp, Red Hat, and SAP, as well as IBM and perhaps a dozen others. Either way, TOSCA shows the direction IBM is working in -- and, says Diaz, it overlaps with a similar OpenStack project known as Heat, a template-based orchestration engine.
The competitive OpenStack landscape
Of course, IBM's plan to make the latest, unvarnished OpenStack bits core to all its cloud offerings is only one approach. Piston Cloud, Rackspace, Red Hat, and others will continue to deliver packaged OpenStack versions -- and soon the first OpenStack appliance will arrive from Nebula, whose CEO Chris Kemp was CTO of NASA when the agency began developing the Compute portion of OpenStack.
Moreover, IBM doesn't have an OpenStack public cloud yet, while HP and Rackspace do. Somehow, it's difficult to imagine IBM going into the commodity public cloud business. Last time I looked, to use the current IBM public cloud, you need to contact IBM and wait for someone to get back to you. It's not exactly the self-service model.
But for private cloud engagements, particularly with enterprises rather than service providers, IBM would appear to have an edge. Professional services are the company's bread and butter. Plus, IBM claims that it had 5,000 private cloud customers as of last year. I have a feeling that claim may rest on a rather loose definition of the private cloud, but no doubt some major portion of those customers may be running SmartCloud software of some kind, which will soon be upgraded to include the OpenStack bits.
Diaz puts it bluntly: "Our objective, frankly, through this layer of cloud technology, this infrastructure-as-a-service layer, is that we want a ubiquitous open source infrastructure as a service layer, period, end of story. That's what we want. OpenStack was headed in that direction when we helped create the foundation. It's even more in that direction now."
This article, "What IBM's embrace of OpenStack really means," originally appeared at InfoWorld.com. Read more of Eric Knorr's Modernizing IT blog. And for the latest business technology news, follow InfoWorld on Twitter.