Rackspace Adds Block Storage, Answers Amazon

Rackspace users starting today will get what Amazon users are long accustomed to: the ability to order the amount of storage their cloud server needs instead of the amount the service provider assigns.

Rackspace announced this morning Cloud Block Storage, intended to compete with Amazon's Web Services Elastic Block Store. Both are the type of storage assigned to a working server running applications. This storage is used to hold active application data such as database transactions, messages, or files that are accessed regularly, compared to AWS' Simple Storage Service, or S3, which is meant to hold infrequently accessed file or data objects.

Rackspace's block storage will come with the performance characteristics of either disk drives or solid state devices, with the latter best for high-performance applications. Solid state is also more expensive, at $.70 per GB per month. The standard service, based on disks, will be priced at $.15 per GB per month.

Amazon charges $.10 per GB per month but adds an I/O fee of $.10 for each million I/O operations. Rackspace's pricing includes no I/O fees. "We don't charge for I/Os. We think it makes sense to simplify the pricing," said John Engates, CTO of the San Antonio, Texas-based cloud service and hosted service provider, in an interview Monday.

[ Rackspace is part of the trend to reduce cloud service prices. Read more at Cloud Pricing: Amazon, Microsoft Keep Cutting. ]

Engates said Rackspace didn't want to launch a cloud storage service until the framework was ready from OpenStack, the open source code project that is producing a set of core elements for private and public cloud computing. OpenStack's Cinder Project is responsible for producing the API and storage service that offers access to permanent block storage in an OpenStack cloud.

A Cinder service, Engates explained, "exposes block storage to users through a standard framework. The Cinder API call will cut across all clouds based on OpenStack [open source code]." That's a key element of the Rackspace approach. It wants its services based on readily available open source code to give its cloud interoperability features either with a customer's private cloud or another public cloud service.

Rackspace pioneered micro servers in the cloud, coming up with a 1.5-cent per hour configuration that undersold both Amazon Web Services and Microsoft. But the company was slow to provide a service that adapted, storage-wise, to the applications customers wanted to run. The micro server came with 10 GB of storage. A small server was accompanied by 40 GBs. If that wasn't enough, customers could move up to a medium server (160 GBs), a large server (320 GBs), or an extra-large server (1200 GBs). Each option was connected to disk drives available in the server rack, and thus certain limits couldn't be breached.

Amazon's Elastic Block Store, on the other hand, has since 2008 allowed users to select the amount of storage needed from its ESB storage service.

Rackspace's standard service comes with more flexibility than previous Rackspace server storage. Users can create snapshots and store them for $.10 per GB per month. They can also attach up to one TB of volumes to a cloud server, and they can detach storage volumes from one server and reattach them to another in a matter of seconds, Engates said.

Find out the nine questions you must ask before migrating apps to the public cloud in the Cloud Ready? special issue of InformationWeek. Also in this issue: It's time to lay to rest two common myths of the cloud computing era. (Free registration required.)