Google Compute Engine: What took them so long?

Finally, we’ve come full circle. The popular notion of cloud computing really began with Google: A sea of cheap white Google servers shimmered before the eyes of C-level execs everywhere, igniting the dream of commodity computing. Pool everything! Allocate resources wherever you need them! No waste! No downtime!

IT managers stared in horror as they were asked the unintentionally hilarious question, “How can we be like Google?” Anyone, it seemed, could move from the crippling complexity of the average data center to elegant yet economical simplicity in one swift imaginary maneuver.

[ Stay on top of the current state of the cloud with InfoWorld’s special report, “Cloud computing in 2012.” Download it today! | Also check out our “Cloud Security Deep Dive,” our “Cloud Storage Deep Dive,” and our “Cloud Services Deep Dive.” ]

Now, some four years after Google launched its App Engine development platform, Google has finally decided to get in the public cloud IaaS game — currently dominated by Amazon — and rent part of its infrastructure in the form of Google Compute Engine. The announcement comes at a time when customers have grown more sophisticated about cloud computing, with a better understanding of how hard and expensive it is to build your own cloud and of the limitations of outsourcing workloads to IaaS providers.

Where will Google Compute Engine fit in the IaaS mix? It’s hard to tell, given the lack of detail in the announcement. Last week I had a brief conversation with the product manager for Google Compute Engine, Craig McLuckie, who seemed utterly confident that the sterling reputation of Google’s world-class infrastructure would bring out customers in droves.

My main question for McLuckie: Why now? Google inspired the whole cloud juggernaut years ago. Why not sooner? Here’s his reply:

It has taken us a little while to get this product to market, and the heart of that has been around the core technology. Our target here is not to deliver just another multitenanted cloud. Our goal is to provide our customers infrastructure in a way that feels familiar to them, like their own data center. It’s about predictability of performance; it’s about having access to clean infrastructure. So it’s taken us awhile to produce a solution that we’re proud of and that we think compares favorably not only to other multitenanted clouds, but to on-premise data centers.

Note the line about “predictability of performance.” That’s a slap at Amazon Web Services, which has earned a reputation for being a little erratic in that department. How can Google deliver on its promise? Basically, McLuckie said, because Google’s infrastructure is better.

Last week in an interview with InfoWorld, Michael Crandell, CEO and founder of RightScale, provided some insight into the nature of that infrastructure, citing among other things the high-speed network that spans Google’s global data center locations. But with the launch of Compute Engine, Google is clearly trading on its own brand and the excitement its advanced infrastructure stirred years ago.

Back to my original question: What took Google so long? McLuckie’s “no wine before its time” explanation seemed a little off. For one thing, Compute Engine is not yet available even in beta, but in “limited preview.” If Google infrastructure as it stands is so great, what required all that time to be carefully crafted? It could only be the software that will deliver the compute cloud to customers — and perhaps the service and support, which has not been a Google strong suit for, say, the paid version of Google Apps.

Why might the software take a while to develop? Last fall, a Google engineer by the name of Steve Yegge unintentionally shed some light on that subject. Yegge, a former Amazon employee, wrote a revealing internal blog post for Googlers only, but accidentally made it public on Google+ (it’s still there if you want to read it in full).

In that post, Yegge detailed how the CEO of Amazon, Jeff Bezos, had mandated in 2002 that “all teams will henceforth expose their data and functionality through service interfaces.” That is, Bezos forced Amazon to become service-oriented. When SOA is properly implemented, shared services can drastically reduce development times because you call existing services through APIs rather than coding from scratch and duplicating functionality. When you do it right, you turn a company’s application infrastructure into a platform.

By contrast, said Yegge, Google “doesn’t get platforms.” Referring to the company’s internal development practices, he says:

It’s a big stretch even to get most teams to offer a stubby service to get programmatic access to their data and computations. Most of them think they’re building products. And a stubby service is a pretty pathetic service. … The problem we face is pretty huge, because it will take a dramatic cultural change in order for us to start catching up. We don’t do internal service-oriented platforms, and we just as equally don’t do external ones.

It’s pure speculation on my part, but the problem Yegge outlines may have been part of the reason it took Google so long to produce the software for Google Compute Engine.

It seemed clear from my conversation with McLuckie that he doesn’t feel Google needs to build an IaaS brand. After all, it’s Google, which has the most advanced infrastructure on earth, and everyone is going to want a piece of it, right? At least, they will whenever the Google Compute Engine goes from limited preview to beta and finally to production and actual SLAs, which McLuckie says Google will eventually offer.

Make no mistake, I think Google will be a player to be reckoned with in the IaaS game, and quite a bit of the attraction may be Google’s mystique, which persists despite various outages and missteps. Yet this is a more competitive cloud world than the one in which App Engine launched four years ago. Google will have not only Amazon and Rackspace to contend with, but also HP, Microsoft, Dell, and even the telcos (Verizon/Terremark, anyway). Google is going to have to prove itself in the IaaS space like any other provider.

This article, “Google Compute Engine: What took them so long?,” 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.