Google launched an Infrastructure as a Service (IaaS) offering, called Google Compute Engine at its I/O conference in San Francisco this week, but I’m not holding my breath for anything great, any time soon.
Google Apple Engine a flop. Yes Google has a giant, distributed back-end infrastructure, but that does not automatically qualify it to successfully monetize and sell that infrastructure to others. Google was early to market with a Platform as a Service (PaaS) called Google App Engine (GAE) which it launched in the spring of 2008, offering third party developers a way to build and run applications atop its infrastructure. Three years later in September 2011 GAE became an official product. Google changed the pricing at this time which irritated some developers but overall GAE has failed to rally the developer community in anything like the numbers that other PaaS offerings have. Google claimed 150,000 coders access the GAE console every month, but that’s small fry compared with the million+ developers on Heroku and at least that number on Amazon Web Services.
Enterprises want hybrid cloud. It took Amazon six years to accept that public clouds would not be the only model for building cloud services. And in the enterprise world especially, hybrid cloud architectures are currently the preferred model. Amazon eventually conceded this and partnered with a private cloud player, Eucalyptus, which cloned the AWS API, enabling connectivity between Eucalyptus clouds built in-house and the AWS public cloud. Google may partner with companies like RightScale and Opscode to make sure these cloud management tools, used by enterprises, will manage its IaaS, but Google has tried and largely failed at selling directly to the enterprise. It’s enterprise search appliance hardly set the world on fire.
Support will be a problem. Infrastructure is a fiddly business, no matter how much you abstract away hardware and design for failure, it still goes wrong in ways you don’t expect. And when outages happen you need rock solid support, which Amazon is only now really building out. When it comes to support, Google says “read our blog”, which is laughable if it has any intention of winning enterprise users to its IaaS.
Limited geographic understanding and sensitivity. Google has yet to embrace global availability zones. It is U.S. data center-centric and users can take it or leave it. Google will chop up data in its Gmail and Google Docs services, for example, and distribute this data in data centers around the globe, but the user has no idea where this data resides. For companies subject to EU laws and regulations around data sovereignty, where the data must stay in Europe, Google’s practice is unacceptable.
Market saturation and lack of standards. Google joins a shopping list as long as your arm of providers in the IaaS market. Hosting providers, communications service providers and pure play cloud providers are all fighting for a piece of the cloud infrastructure pie. But currently they are all offering proprietary services, meaning users cannot move workloads between IaaS providers. This is absolutely holding back the marketplace and will continue to limit adoption until there is a standard unit for compute, storage and bandwidth in and out, across all of them.
That’s not to say that Google won’t ever be successful in the IaaS business, but it took Amazon six years to get there with AWS and it will take Google at least that long. Google certainly has an opportunity to undercut Amazon on price which could be compelling for a certain community of users, but catching up with AWS is going to take a laser-like focus not typical from Google these days.