Google finally designated its Compute Engine infrastructure-as-a-service offering as generally available on Monday, about 18 months after announcing the service at its I/O conference in June 2012. It’s not clear how, exactly, generally available is different than the “available for anyone to use” status Google donned Compute Engine with in May (perhaps it has to do with support and SLAs), but Monday’s announcement does come with a collection of new features.
The new features vary from standard stuff like bigger instance types and support for more varieties of Linux to higher-level capabilities such as live migration of virtual servers during scheduled maintenance and automatic restart of those servers should they go down. Google also lowered the price of computing resources by 10 percent across the board and lowered the price of persistent disk capacity by 60 percent. IO performance on the biggest persistent disk instances is up to 700 percent faster, the company claims, and IO is now included in the price of data storage.
Although the new features will certainly make Compute Engine more palatable to some potential users, it’s still nowhere near comparable to cloud leader Amazon Web Services in terms of breadth of capabilities and services. Gartner analyst Lydia Leong has written a post looking at the feature gap between the two platforms as they exist today and expresses an opinion that Google will ultimately start closing it by exposing some of the impressive technologies it has developed to run the rest of its services (e.g., search, Gmail, Apps and the like). I tend to agree with her assessment, and have said as much before.
Google probably does need to become feature-competitive with AWS sooner rather than later, but that doesn’t mean it necessarily needs to match AWS tit for tat. While AWS is rolling out new stuff pretty much weekly and has “two-pizza” teams focused on specific services, maybe Google can compete by focusing on the bigger picture and features that let it establish its own identity. Compute Engine’s live migration feature, for example, is a meaningful step toward the kind of resiliency that Netflix has architected itself on top of AWS but AWS doesn’t yet provide itself.
The cloud computing market is still very much a one-horse race, and it very likely will be — at least in market share and revenue at the top — for some time to come. If Google is ever going to make a big push to dent AWS’s lead (or at least duke it out with Microsoft for a chance to become the solid No. 2 in the space), perhaps the answer isn’t replicating AWS every step of the way but focusing on the stuff it does best. It does need to get some table stakes in place — better instance variety and maybe some more advanced data analysis services — but after that perhaps actually delivering on being Google can pay off.
For more on Google’s cloud plans, listen to our August podcast with Greg DeMichillie, director of product management for the Google Cloud Platform.