What Google Compute Engine means for cloud computing

Google’s new infrastructure-as-a-service offering Compute Engine is a big, big deal in the world of cloud computing. My colleague Om Malik covered the details in a post earlier today, and we’ve both offered our pre-emptive views in posts breaking the news that Compute Engine was coming and then confirming it would be announced at IO. Now that it’s a reality, here are five things I think Google Compute Engine means for the cloud industry.

Platform as a Service is still too ambitious. Microsoft couldn’t make a cloud business out of Windows Azure as it was originally built, and apparently Google couldn’t with App Engine, either. That message was made clear in my discussion with Microsoft cloud boss Satya Nadella at Structure last week, and during a chat with Google’s cloud leaders last night. Trying to get ahead of Amazon Web Services, they also got too far ahead of the market in terms of PaaS readiness. While both companies still sing the praises of PaaS as the future of application hosting, they understand that IaaS is still the name of the game if they want real applications.
OpenStack is more important than ever. I cannot tell you how many times I’ve heard that OpenStack is too immature and how the development process isn’t fast enough to keep up with AWS. But I can also tell you how scared many larger enterprises, especially, are of getting locked into a particular cloud computing platform. This is especially true as they move toward hybrid cloud environments that span public and on-premise clouds. Rackspace ends up looking better than ever, but it will have to get moving to build out its OpenStack-based cloud into something more comparable to AWS, Windows Azure and Compute Engine (Google’s whole cloud suite, really) than it currently is in terms of higher-level services and support for specialized workloads such as Hadoop.
The scale business is closed. Between AWS, Microsoft and Google, there’s little use trying to compete in the IaaS world on price or scalability. They have the economies of scale to drive prices down ever lower (Google claims Compute Engine is half as expensive as other clouds, but I doubt that will be the case for long given AWS’s penchant for cutting prices) and sheer computational power to handle high-performance applications. Lots of researchers are already running HPC apps on AWS. At IO, Urs Hölzle, Google’s senior vice president of technical infrastructure, demonstrated a genome-analysis application running at 600,000 cores. There’s plenty of room for other cloud providers — especially those targeting traditional enterprise applications — but they’d better have some real differentiating value to justify paying what’s certain to be a higher price point than the big boys will charge.
Facebook has to counter with its own IaaS offering. Although, granted, it will probably function a lot differently considering the different set of developers Facebook would be targeting. I don’t suspect anyone is looking to Facebook for hosting mission-critical enterprise applications, but there’s certainly a business in letting companies build some serious apps that run on the Facebook platform, and IaaS is one way to do that. And Facebook definitely has the scale and infrastructure smarts to pull this off. Better yet, it could do this relatively inexpensively on its end by building atop low-power ARM processor-based servers, with which it has been experimenting.
Google is the force to be reckoned with in the cloud? While still immature, Compute Engine complements what’s already an expansive suite of cloud offerings around Chrome, Apps, and development tools ranging from Drive to big data services. Of course, Google also has App Engine, which Hölzle said is currently hosting a million apps. That’s still less than Heroku hosts, but not bad at all for a service still out of the reach of many developers and that angered many existing developers after pricing changes late last year. One could argue that Google currently has the most-expansive suite of cloud products available, with only Microsoft even close. AWS is presently taking a different tack, focusing on expansive IaaS capabilities and an app marketplace, so we’ll see if it feels compelled to move up the stack with a truer PaaS offering and more managed services a la DynamoDB. VMware has many of the pieces in place, but it’s a long way off from being considered a true cloud provider.