Google Compute Engine Enters General Availability
Search giant Google announced this week that its Infrastructure-as-a-Service (IaaS) offering, Google Compute Engine (GCE), has gone into general availability. The news comes about 18 months after the service was launched at Google’s I/O conference in June 2012. The announcement of general availability brings several new features to the product, including expanded OS support, new compute instances and transparent maintenance, as well as lower prices.
Google vice president Ari Balogh delivered the details of the announcement in a blog post, noting that the company had cut prices 10 percent for standard server instances in all regions, lowered storage fees (aka Persistent Disk) by 60 percent, and dropped charges altogether for storage input-output. Google reports that its largest persistent disk volumes now offer up to 700 percent higher I/O capability.
During its beta period, GCE only supported two Linux distributions, Debian and Centos, but effective immediately, developers have the option of running any standard Linux distribution, including SELinux and CoreO. The company is also announcing support for SUSE, Red Hat Enterprise Linux (in Limited Preview) and FreeBSD. Tools such as Docker, FOG, xfs and aufs will also be supported.
In a move that should be attractive to the HPC crowd, Google has added three new 16-core instances to its portfolio. The 16-core instances are on par with Amazon’s largest instances (xlarge) and come in the following configurations: standard (with 60 GB of RAM), high-memory (with 104 GB of RAM) and high-CPU (with 14.4 GB of RAM). The company expects that applications will “range from silicon simulation to running high-scale NoSQL databases.”
Currently, the new instances are only available in limited preview, which means that users must request access to the machine types by filling out the Limited Preview request form.
Google has also added support for transparent maintenance, which “combines software and data center innovations with live migration technology.” This means the system can perform proactive maintenance while the virtual machines are running, which translates to less downtime and fewer mandatory reboots. Another added feature brings VMs back online in minutes in the event of a failure.
Director of Product Management at Google, Greg DeMichillie, shares further details about the announcement in this video: