Richard Soley was among those at the center of the move in the 1990s to the standard distributed object computing architecture that came to be known as CORBA. The Common Object Request Broker Architecture led to a massive paradigm shift in enterprise architecture, one that sowed the seeds for more shifts that followed.
Soley is uniquely prepared to analyze the path of highly-touted cloud computing and less-publicized cloud computing standards. He can also mark the path of SOA, which more or less preceded cloud computing in the hyperbole parade.
How does Soley, the long-time chairman and CEO of the Object Management Group (OMG), an organization now looking at modeling extensions related to cloud computing compute loads and SLAs, view the paradigm shift represented today by cloud computing architecture? Cloud computing's value may be obscured just now by all the surrounding hyperbole, according to Soley. Yet, for certain companies it is already a game changer.
Richard Soley: My take on cloud computing is it’s a great idea – but not a particularly new idea. It is unfortunately being overhyped right now, which masks the fact that it’s a great idea.
Cloud computing is the nexus of a bunch of ideas that have been around for a long time, impacted by the fact that computing costs have plummeted, especially for storage. Meanwhile, Internet access is ubiquitous.
But we have this tendency to overhype things in the industry, and then at the end of the hype period people assume that [the overhyped item is] worthless, and then move onto the next hype thing. That would be very unfortunate here.
Cloud is not a paradigm shift for a lot of businesses, but there are a couple of huge cases for which it’s a major shift, especially in the startup world. Since cloud computing is about moving capital costs to operating costs, it is a major paradigm shift for startups.
They can startup, although they are capital poor, with little or no computing equipment. They can buy more access - elastically as they need it - to computing power on a cloud platform as they go.
One of the definitions of whether a standard becomes successful is when people don’t talk about it anymore and they just do it.
The green field of the startup is a far cry from the established enterprise with a lot of long-working systems sometimes called 'legacy.' The cloud experience here is different. Soley continues.
Soley: There is another reason why cloud computing is more of a paradigm shift for startups than for bigger companies. The issue is the legacy. And it’s not just moving your legacy applications to the cloud - it’s how to integrate what’s running on public cloud with the computing resources that are still in your own shop, either because of the legacy or the security concerns, or you just haven’t had the time to move them. So you have more connections between the cloud implementation and the in-house implementation. There are starting to be available software solutions to help with that move, but we’re not entirely there yet.
Cloud computing as first described had the feel of interchangeable Lego pieces. That impression has proved deceptive. Yes, there are many existing standards that cloud computing taps into. But you can't yet move from cloud to cloud without big changes. How does Soley view the cloud standards' firmament?
Soley: I think that the cloud standards picture is somewhat muddied by the fact that a lot of standards are already in place that were not developed specifically for the cloud, but which are relevant to cloud computing. If you look at the cloud platforms that are out there, they tend to have very different interfaces. So it is not trivial to move your cloud implementation from, say Microsoft to Amazon to Rackspace to Salesforce. They use different programming languages, and they use different APIs for controlling the platform, and so forth. There are already some very high quality standards done for cloud computing, notably the Distributed Management Task Force (DMTF's) OVF for cloud computing program loads, but they are not widely supported yet by all of the vendors. So portability is difficult.
And there are simply missing standards for managing cloud implementations – moving from one to another, starting and stopping [jobs], and so forth.
The good news is, rather than having 97 different standards for everything, the cloud computing standards organizations are working together to avoid stepping on each other's work. And you have the Cloud Standards Customer Council actually providing prioritization and requirements for cloud standards organizations.
This is not a standards' organization but is a group of primarily end users that are working to develop a practical guide to introduce cloud computing into an organization, and to develop a set of priorities and requirements for standards organizations – not just the OMG, but The OpenGroup, the World Wide Web Consortium, the Internet Engineering Task Force and DMTF and whoever else is interested to get directly from the horse's mouth – the buyer's mouth – the priorities and requirements and what they care about, and how they think about public, private and hybrid clouds.
SearchSOA.com sees affinities between cloud computing and SOA. One affinity is a bit on the negative side: that is the penchant for both cloud and SOA to ride the hype rollercoaster. We asked Soley what he saw here. Is it mainstream?
Soley: This is a great example of what I was talking about right at the beginning. And that is that SOA is a great idea but it was overhyped as something you could buy and put in place and your computing world would be ten times better. At the end of that hype cycle, that caused a lot of pain.
There are still plenty of organizations that are using SOA and still plenty of organizations that are moving to SOA, but it’s the hype pathology making it look like it has died. It hasn't. For our part, the [OMG] SOA Consortium was very successful and is now working with our Business Process Improvement Community of Practice and as part of our business ecology initiative. Plenty of organizations are moving to SOA – which they should be – because it’s the right way of thinking about building a set of services that will implement your business.
If you ask if SOA has become mainstream, the answer is ''yes.'' One of the definitions of whether a standard becomes successful is when people don’t talk about it anymore and they just do it.