EMC announced its Centera Content Addressed Storage system, an integrated hardware and software solution, targeted...
for what EMC refers to as the "fixed content" market. Fixed content documents include items such as financial records, photographs, medical x-rays and MRIs, video, seismic data, books and manuals. EMC classifies this as a new storage category that it has named Content Addressed Storage (CAS). The basic architecture for Centera was acquired by EMC as part of its acquisition in April 2001 of a Belgian software company.
EMC refers to the Centera hardware architecture as RAIN (Redundant Array of Independent Nodes). Each Centera cabinet can contain 16, 24 or 32 nodes providing up to 9.6TB of usable mirrored capacity in a 19-inch rack. At a minimum two of the nodes are "access nodes providing external API access. The other nodes are "storage nodes." Each node contains a 850MHz Pentium III with 256MB roam, four 160GB EIDE drives and three 10/100 NICs. The disk drives in the access nodes are used for cache. The disk drives in the storage nodes are used for storage.
Up to 16 Centera cabinets can be installed per Centera cluster which can provide up to 153.6TB of usable capacity. Additionally, up to seven Centera clusters can be configured per Centera domain providing a little more than 1PB of usable capacity.
CentraStar is the management software component of Centera. It incorporates a content-based addressing scheme, based upon Message Digest 5 (MD5), that assigns a unique address for each stored object. This address, or digital fingerprint, is 27 characters (128 bits). Some of the characters are available for users to add application specific metadata such as the originator's name and identification, date, characteristics, etc. The original content cannot be overwritten or changed. Any change in an object will result in a new content-based address.
Centera is self-configuring and self-managing. It will automatically configure the internal IP addresses of its nodes and switches. It will also determine the physical placement of objects and perform load balancing across the active nodes as well as when new capacity is added or deleted. Centera also supports asynchronous remote replication over IP networks.
Centera initially supports APIs for applications running under Windows NT, Windows 2000, Solaris and Linux. EMC has stated that support for HP-UX and AIX will be available latter this summer.
EMC has established partnerships with more than thirty independent software vendors (ISVs) covering: backup/restore, check imaging, content management, digital content management practice, document management, edge storage/caching, e-mail management and archiving, government research, knowledge management practice, media/entertainment, medical imaging, server/content management, and workflow mangement.
Deployment and migration
Several methods are available for deploying Centera and migrating data to it. Software from ISVs that have incorporated the Centera API will direct data to it. Users can also engage professional services from both the ISVs and EMC to develop customized migration plans. EMC also provides software developers kits for those customers that would like to modify current applications to utilize Centera's APIs.
The first release of Centera utilizes a standalone management console. The Evaluator Group is not aware of any plans EMC may have to integrate it into its AutoIS management scheme.
EMC stated that the price for a Centera cabinet configured with 4.8TB of mirrored usable capacity with sixteen nodes would be about $205,000, which is about $.04 per MB.
Evaluator Group comments:
The fixed content market is very large and according to the UC Berkeley study, represents about 75% of all new digital data. With Centera, EMC has once again taken an innovative approach to storage. However, one must question how much of this very price sensitive market will be willing to pay $.04 per MB for sub-second access and any improvements that Centera may provide in the area of management of this data. EMC's success with the Centera product will depend primarily upon its channel partners (ISVs and VARs).
To view all Storage Evaluator analyses and commentaries, go to Storage Evaluator Tips ===========================================================
The Evaluator Series and Evaluator Series On-Line (ES/OL) are trademarks of Evaluator Group, Inc.