InfoStor : Caringo upgrades CAS software, July 2007 Page 1

By Kevin Komiega

July 31, 2007—Start-up Caringo rolled out the second version of its CAStor content-addressed storage (CAS) software this week with new local and wide area replication features, better performance, and expectations that its software can change end-user perceptions of not only CAS technology but also the entire concept of information lifecycle management (ILM).

CAStor 2.0 has a number of new features, including asymmetric local and wide area replication, which ensures that redundant data resides on separate sub-clusters locally or at a remote site. Version 2.0 also includes features that allow administrators to build larger storage clusters with added support for network booting, centralized configuration files, and SNMP enhancements that improve the monitoring and management capabilities of CAStor clusters.

"We initially thought about CAS as being synonymous with deep archives. Now with version 2.0, there is more applicability to things like active archives and enterprise applications," says Mark Goros, Caringo's CEO.

Goros claims CAStor 2.0 delivers the access and performance specs of primary disk storage, enabling continuous data availability. The software comes on a USB flash drive, can be installed on any hardware, and can take advantage of high-capacity, low-cost SATA drives. The release includes performance increases for small and medium file writes and reads. Additionally, network protocol enhancements allow for the creation of CAStor clusters in excess of 1,000 nodes.

Goros claims the CAStor architecture is "future-proof" because of its standards-based design. "It's one large fixed-content store that's organically upgradeable. As soon as a standards-based disk drive comes out you can plug it into CAStor that day," he says. "The system will become less expensive over time as commodity CPU and disk prices fall."

Caringo is also attempting to alter traditional views of ILM with a new concept called single-tier ILM. CTO and founder Paul Carpentier says the ILM process is supposed to help move information from real-time to nearline to archive, but traditional approaches are complex. And, until now, storing everything on fast, online disk has been too expensive.

Carpentier claims CAStor 2.0 eliminates the need for traditional storage tiers used for ILM systems, where less-frequently accessed files are moved to lower-cost, lower-availability storage. To create the foundation for a single-tier ILM platform, CAStor software creates a single, large, flat address space without hierarchies. Each object gets a unique identifier for its entire lifetime and, according to Carpentier, data is not vulnerable to attacks because the system transparently upgrades the integrity hash functions during operation, without changing identifiers or reloading data.

CAStor 2.0 is available via Caringo's partners, and pricing varies. Existing customers can upgrade to the new release without migrating data by obtaining a new USB key, uploading the software, and rebooting the storage node.

Page 1 of 1