Hewlett-packard Development Company, L.p.

Hewlett-packard Company

Mohamed Belhadj

David Kiyoshi Umberger

Rodger Daniels

Douglas L Voigt

Siamak Nazari

Douglas James Cameron

Charles Franklin Randall

Optimization of a RAID volume is described. In an example, a method of optimizing a redundant array of independent disks (RAID) volume includes: identifying a RAID level used for a plurality of disks in the RAID volume; defining a relation among rebuild rate, scrub rate, and annualized data loss event rate (ADLER) for the RAID volume using constant values and the RAID level; selecting a value for the ADLER; applying a policy to the relation to choose a value for one of the rebuild rate or the scrub rate; and determining a value for the other of the rebuild rate or the scrub rate from the relation using the value of the rebuild rate or the scrub rate as determined from the policy.

Optimizing a RAID volume

Improved host I/O performance and availability of a storage array during rebuild is obtained by prioritizing I/O requests. In one embodiment, rebuild I/O requests are given priority over host I/O requests when the storage array is close to permanently losing data (for example, failure of one more particular disk in the storage array would result in data loss). Rebuild I/O requests continue to have priority over host I/O requests until the storage array is no longer close to permanently losing data, at which point host I/O requests are given priority over rebuild I/O requests.

Host I/O performance and availability of a storage array during rebuild by prioritizing I/O requests

Host I/O performance and availability of a storage array during rebuild by prioritizing I/O request

A method of managing data in a hierarchical data storage system employing data redundancy schemes includes prioritizing a data rebuild based on a most vulnerable data redundancy scheme in the storage system. A data storage system embodying this rebuild prioritization method is also described. Prioritizing the data rebuild includes enabling a rebuild of the most vulnerable data redundancy scheme prior to enabling a rebuild of any other data redundancy scheme in the system. The most vulnerable data redundancy scheme is determined by comparing a probability of losing data that can be prevented by a rebuild for each data redundancy scheme with respect to the potential for one or more next storage device failures in the data storage system. The probability of losing data for each data redundancy scheme is determined by considering characteristics associated with the storage system and disk drives in the storage system, including number of storage devices, mean time between failure, mean time or calculated time to rebuild, and failure dependencies.

Raid rebuild using most vulnerable data redundancy scheme first

Growing community of inventors

Eagle, ID, United States of America

Mohamed Belhadj