Emulex, LSI, Oracle and Seagate on Wednesday unveiled the Data Integrity Initiative (DII), a technology collaboration that implements and expands the T10 DIF (Data Integrity Field) standard for complete end-to-end data integrity for enterprise storage systems.
The issue is a critical one for high-reliability environments like financial services and healthcare coping with bit error rates and other sources of data loss, corruption and downtime.
The initiative began as a Seagate internal effort and became a standard after other companies became involved. Seagate engineering director Mike Miller said the technology addresses "a rare problem, but one that's important to attack."
The technology saves errors from being written to disk and sends the data back to the application for correction, said Miller. Each component seamlessly checks data all the way down the data path, correcting errors before it's too late to catch them.
"It's difficult to recover if you don't know when and how it happened," said Miller.
The DII technology uses standardized data checking mechanisms that allow each storage component to continuously monitor the integrity of data either in-flight and at rest. The technology enables rigorous data checking, starting with the software application, all the way through the storage and file system, and ending on the disk drive. The technology not only detects, but also isolates and reports the sources of error and data corruption, helping data centers avoid lengthy downtime.
There are no products yet, but the companies expect them to begin to appear next year. The standard directly benefits SCSI-based storage such as Fibre Channel, SAS and iSCSI, and can indirectly benefit SATA drives used in a Fibre Channel environment.
Oracle has a similar initiative called HARD (Hardware Assisted Resilient Data) that has a "bit of overlap" with the DII effort, said Oracle's Jim Williams. Emulex's BlockGuard technology, which ensures the integrity of user data as it is transferred from the application to the SAN, is also part of the demonstration.
There are different levels of data corruption, said Williams and Miller. Database metadata errors can be corrected immediately, while errors in relations between data can take a long time to catch, said Williams. And snapshots might be of little value data corruption at the right time in a RAID system can affect all data copies, said Miller.
The new effort is aimed at correcting such mishaps, they said.