More Details on Google’s Cloud Offering; Many More Details Needed
June 28, 2012 by Leave a Comment
Google has released a few more details about Compute Engine, the infrastructure-as-a-service offering it debuted at Google I/O today. In a company blog post, Craig McLuckie, its product manager, sketched some of the details that didn’t make it into Urs Hölzle’s remarks during today’s keynote address at Day 2 of Google’s developers conference. The heart of the offering is the ability to run scores of Linux virtual machines on Google’s services. Now, if that doesn’t mean anything to you, it’s helpful to be reminded of one key tenet of cloud computing: Virtualization. Basically, one physical computer has so much computing capacity — because processors are now so powerful — that it can act like two or four or 10 computers all at once, all sharing the same hardware. Google is counting that its experience in harnessing huge swaths of inexpensive computing hardware will give it a competitive leg up against the likes of Amazon Web Services and Microsoft’s Windows Azure, but it’s still available only in a limited preview, so there are going to be a lot more questions. One piece of the puzzle that Google has going for it is that it is working with several partners that already have a history of helping customers manage their various cloud services. Google named RightScale, a management platform; Puppet Labs; OpsCode, which specialize in automation; Numerate, which uses computing to design drugs; and Cliqr, a cloud applications manager among its partners. Another partner stood out to me: MapR, whose funding I covered last year, offers a version of Hadoop, an open source, big data technology that’s designed to run in the cloud. It’s already the Hadoop of choice for running on Amazon’s cloud, and now it seems to have an inside track on Google’s. Then there’s the matter of performance. Amazon has been known to have stability issues: Fifteen months ago it suffered a crash that brought numerous Web services down with it. Google is promising “strong and consistent” performance that customers can rely on and thus tune their applications based on that expectation. Google is also planning to compete on price. According to its price list, prices are beginning at 14.5 cents per hour for the use of a single virtual processor core with 420 gigabytes of local storage. I’m not familiar enough with all the services to compare it with this price list for Amazon Web Services. There’s also the question of security. CIOs have often tended to view cloud services with a suspicion based on the perception that the cloud isn’t secure. And while I haven’t read it yet, Google has written a white paper on the subject. Obviously, more details will be coming out in the days and weeks ahead as Google moves this from a preview to a full-fledged service. It has made a big point today in bragging about its capabilities. Now it remains to be seen if those bragging points can stand up to scrutiny and daily use.