Modern Infrastructure

Cloud computing, data center trends: If you can't beat 'em, join 'em


Evaluate Weigh the pros and cons of technologies, products and projects you are considering.

One-on-one with Randy Bias: Cloud standards and the battle for control

When discussing open source cloud and standards, Randy Bias argues it's only a matter of time before dominant cloud standards prevail in IT.

As cloud computing lurches toward greater maturity, issues such as common standards and interoperability remain hurdles -- even for users encouraged by the cloud's promise of better infrastructure management and reduced costs.

In this one-on-one interview, Randy Bias, co-founder and chief technology officer of open source cloud infrastructure provider Cloudscaling, talks about the lack of cloud standards among providers and how today's cloud computing service provider patchwork will ultimately give way to a more uniform cloud universe.

Many major cloud providers have application programming interfaces that aren't compatible with others' APIs. Will this change?

Randy Bias: What I've seen is that infrastructure that's important to the business ultimately drives toward [the establishment of] standards and APIs and homogeneity. Without those things, it's difficult to create a plethora of standards and APIs higher up the stack.

Which cloud player might win the battle for dominance?

Bias: You can complain about whether Amazon Web Services or VMware is the right model, but we can say with great assurance that they both dominate their respective parts of the market, and that is driving a certain amount of customer perceptions.

The thing that we're really missing is a set of solution providers that can package and deliver an Amazon equivalent that is more than just an API.

Randy Bias,
co-founder and CTO, Cloudscaling

Still, some people feel uncomfortable with [this] dominance … and they have to make a case for there being other standards, platforms and APIs. But with 90% of public cloud capacity on Amazon and 90% of enterprise virtualization clouds on VMware, it's hard for me to believe that another standard is going to come out and clock those -- that just doesn't make sense.

Is it a problem that public and private cloud providers' APIs haven't been compatible with one another?

Bias: There are two different architectural approaches. The VMware pattern is what we call 'enterprise virtualization clouds,' which is the ability to take existing enterprise stovepipes -- all the little silos of different hardware, software and network and storage architectures -- and recreate and virtualize them. But [with that approach], you don't get the economies of scale.

So this approach has been driven by the need for greater server consolidation and greater utilization rates.

The Amazon approach is more reminiscent of what Internet giants have been doing. Instead of building stovepipes, they build layer cakes; they have a whole stack of services that support each other, from the concrete to the software. And then at the software layer, they have a whole bunch of applications that leverage that whole stack. We refer to this as an 'elastic infrastructure cloud.' It caters to greater scalability for the apps and the infrastructure, and doesn't try to manage lots of stovepipes, but it makes the applications reflect the underlying infrastructure.

How do these approaches to cloud differ in practical terms?

Bias: There is a very big difference in the types of applications and workloads supported on each.

VMware customers have database servers like Oracle that are old and don't have any basic replication. They rely on the infrastructure to have high amounts of availability and the application is designed to assume that the infrastructure never goes down.

We see a very different [experience] for customers like Netflix that adopt Amazon. They are redesigning their applications for the underlying assumption that servers always go down, and it's just a question of when. So the question is, How do you design your application to handle that?

There is a sea change of perception; people now see the value of being able to scale applications horizontally as the large platforms like Facebook have done.

Will companies coalesce around a common standard?

Bias: There is some amount of coalescing already, but customers are all over the place, frankly. It's still early days. And there is a disconnect between the understanding of these two fundamentally different patterns. Customers think they are recreating the Amazon environment when they create a VMware environment that allows them to turn on virtual machines on demand -- except that it doesn't provide the scale or the elasticity. And they find out that it's not good enough for their application developers to be successful internally, who then continue to go to Amazon. People will continue to fail and through the failures learn the hard way.

The thing that we're really missing is a set of solution providers that can package and deliver an Amazon equivalent that is more than just an API.

What role, if any, does open source play in cloud standards?

Bias: With this particular disruption, open source and open standards are leading the charge. If you look at a lot of those guys who are really big, like the Googles and Facebooks, all of their systems largely use open source. Yes, for cost economics, but in many ways, it's a control-your-own-destiny thing.

Simultaneously, you're starting to see a backlash among enterprises that have been beholden to hardware and software vendors and are starting to get sick of being locked into these vendors and then getting milked. Enterprises are starting to see that open source is at a maturity level at which they can embrace it.

How is the cloud different from its predecessors?

Bias: It's pretty much disrupting the de facto standard, the client-server and enterprise computing paradigm, which displaced the mainframe paradigm.

We've gone from a gigantic, single iron box that everybody logs in to that runs one or a handful of applications to something that is more distributed -- with shared responsibility between IT staff who run centralized servers and users running laptops or desktops -- to an even more fully distributed model that is always on.

About the author:
Lauren Horwitz is the executive editor of the Data Center and Virtualization group at TechTarget. Horwitz joined the Data Center group following a two-year tenure at CIO Decisions magazine, where she also served as managing editor and contributing writer. Previously, Horwitz served as senior editor at Cutter Consortium, an IT research firm, and editor at The American Prospect, a political journal. In 2009, she received a Silver award from the American Society of Business Publication Editors (ASBPE) and a min Best of the Web award. In 2007, her work won awards from the ASPBE as well.

Article 1 of 7

Dig Deeper on Cloud computing standards and compliance

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.

Get More Modern Infrastructure

Access to all of our back issues View All