Google Compute Engine is now generally available and facing the three headed cloud monster: Amazon's AWS, Rackspace and Microsoft's Azure. How will Compute Engine fare against the three established public IaaS players? Will it become the next big thing in enterprise IT?
Now that Google has officially opened the Google Compute Engine to general availability, developers and enterprises have yet another well supported, full featured public IaaS. Google made its name by dominating search, but has since become a major force in mobile computing with the Android operating system. Can it do the same with cloud computing? How will it fare against three of the other big public IaaS offerings?
The relative success of Google Compute Engine will depend on the quality and cost of compute and storage services, its ability to integrate with enterprise IT infrastructure, and Google's willingness to offer more services to support application development and management.
The Basics: Compute and Storage Services
Compute and storage are the major resources used by cloud computing customers. Other services, from search and message queues to load balancing and performance monitoring, are valued by some customers, but the bulk of revenue for an IaaS provider will stem from selling time on virtual machine instances and storage devices. These services will compete on cost, performance and reliability. Google cut the prices of its compute services by 10% and committed to a 99.95% SLA. It also expanded its operating system options to include virtually any Linux distribution.
Microsoft shops will still find their best option in Windows Azure or Amazon AWS where Windows Server platforms are supported along with Linux. Google's SLA is comparable to Amazon's and Microsoft's and nominally better than Rackspace's 99.9% availability. Details matter when measuring availability. The 0.05 differences between Rackspace and competitors should be considered in terms of how a vendor measures availability. For example, if one vendor includes regularly scheduled maintenance in downtime while the other doesn't, then SLA measures are not comparable.
Google has found a way to avoid the question of how to measure maintenance downtime -- don't have any. The Google Compute Engine provides transparent maintenance. If the machine running your application has a hardware failure or needs a software patch, your virtual instance is automatically migrated to another server.
- Google IaaS Goes GA
- Google PD: Cheaper, Faster Storage For Compute Engine
- Getting Started with Google Compute Engine