Cloud storage often is associated with object storage to allow massive scalability and geographic independence of data, allowing it to be accessed from any location. But object storage lags behind block storage for performance.
Cloud provider Rackspace last week began offering Cloud Block Storage, powered by OpenStack, to compete with Amazon Web Services' Elastic Block Store. Rackspace's Cloud Block Storage service holds active application data, such as database transactions, messages and files that are accessed regularly, where AWS S3 is meant to hold infrequently accessed file or data objects.
Cloud Block Storage requires the compute, storage and applications to reside in a data center and delivers the kind of high-performance that object storage can't provide. Cloud Block Storage uses solid-state drives (SSDs) for high performance applications and hard disk drives for applications that need more moderate performance.
"This is a services provider play," said Marc Staimer, president of Dragon Slayer Consulting. "And Rackspace is in perpetual catch up with Amazon. And I'm sure Google will have something very shortly."
Cloud Block Storage moves storage off the server so customers can provision storage independently of the server. Customers can pick the amount of storage they want to use with a specific compute instance. Currently, Rackspace is limiting 1 terabyte (TB) per volume, and customers can ask for multiple volumes. Like Amazon, Rackspace is offering two tiers -- the standard volume option for hard drives and SSD volumes for accelerated speed.
The standard hard drive option costs 15 cents per GB, while the SSD option costs 70 cents per GB a month. There is no added cost for the I/O throughput.
"You are not outsourcing storage," Staimer said. "You are outsourcing the entire infrastructure to support the application. It aligns storage with the application performance requirements. You need high performance, you use SSDs; you need lower performance, go to the disks."
Cloud Block Storage offers the flexibility to create public cloud-based storage volumes that interact independently of servers that are Windows- or Linux-based, said Mark Interrante, Rackspace's vice president of cloud products. It's a way to store large amounts of information that is not attached to specific cloud servers.
"Block storage is increasingly popular, designed for higher I/O. It allows you to have an architecture and design to support bigger and more scalable applications," Interrante said. "Object storage has a different design. It is optimized for assembling data and delivers millions of simple requests quickly."
Rackspace claims its offerings have less variability in performance compared to Amazon. "Based on their testing, they say they offer better service-level agreements because it's more consistent," Staimer said.