Just What Is 'Content-Addressed Storage'? Page 2


Want the latest storage insights?

Download the authoritative guide: Enterprise Data Storage 2018: Optimizing Your Storage Infrastructure

'Searchability' May Be Key to CAS Growth

Content-addressed storage is object-oriented, not unlike the Java programming language conceptually. It also uses disk-based technology, which is more easily searchable than removable media such as tape, but also more expensive. Searchability may be the key to finding more applications for CAS.

"There's great future value for search by content approaches, particularly for images," maintains MacFarland. For example, a hospital or medical research center can construct an application that will search for tumor growth in images stored using CAS.

Kevin Daly, CEO of Irvine, Calif.-based Avamar Technologies, readily admits that EMC coined the term content-addressed storage, even though his company is using the same concept for a disk-based back-up system called Axion. But that's just fine with him.

"The technology has been around in academic circles for a while," says Daly, "but it hadn't really found applications until EMC brought Centera to market."

Axion is a hardware and software appliance that is going where few backups have ever gone -- magnetic disk. By using CAS, Avamar has come up with an approach that reduces the amount of disk needed, and in turn brings disk-based backup closer to the price point of tape.

Daly says that Axion does use the same CAS approach as EMC, but Axion uses it at a level he calls "very small grain." Once Axion sees an object when creating a backup, it never stores that object again. This is made possible because each object, as in EMC's Centera, has a unique logical address related to its content.

"We don't have to do fancy things like pattern searching," adds Daly. And because no object is stored twice, it uses one-tenth the capacity of a normal backup. An organization with a 10 TB environment, for example, will require 100 TB of tape for a normal tape backup. With Axion, Daly said the backup will require just 10 TB.

Both EMC and Avamar claim their CAS systems are nearly self-managing and much more autonomous than traditional file systems because the objects are fixed and randomly placed in the addressed space.

Avamar has just raised $13 million in its fourth round of funding, and Daly believes his company's application of CAS will take hold as disk-based backup grows in popularity.

"I believe the entire back-up world will in time move to disk-based," he says. But right now it has to start small. Disk-based backup may not be efficient for customers with an environment less than 2 or 3 TB, according to Daly. CAS requires a great deal of processing, as it recomputes an object's unique identifier whenever it reads, and thus is more attractive to larger systems.

Regardless of the system, CAS is very effective at making sure the data you saved is the data you need because of the unique identifier.

"It gives you a very powerful way of making sure the object you need hasn't changed," says Daly. "That's very hard to do in a traditional file system. This is a real change in the storage business."

This story originally appeared on the CIO Information Network.

Back to Enterprise Storage Forum

Submit a Comment


People are discussing this article with 0 comment(s)