International Business Machines Corporation

Tal Sivan

Michael Edward Factor

Danny Harnik

Joseph S Glider

Elliot Karl Kolodner

Yoram Novick

Erez Webman

Dalit Naor

Demyn Lee Plantenberg

Yael Petruschka

Systems and methods for optimizing deduplication in a data storage system are provided. The method comprises associating a first name with first data blocks including first content stored in a data storage system, wherein the first name is associated with the first data blocks by way of a reference to a first meta file that points to a data file which points to the first data blocks; storing a first signature derived from the first content in an indexing data structure, wherein the first signature is used to associate the first name with the first data blocks and as means to verify whether a second content is a duplicate of the first content, based on value of a second signature derived from the second content.

File system object-based deduplication

System for the storage and maintenance of data sets updates in a storage subsystem, that comprises direct access storage device(s) that serves as the main storage of the storage subsystem and on which the data sets are originally stored; a Cache memory storage device that enables fast interaction with the storage subsystem, on which a copy of the data sets is stored; a non-volatile storage device partitioned into a plurality of fixed size non-volatile memory pages. The system employs an update process, in which the changes to the data sets are applied to the data set copy stored on the Cache memory device; a journal of the changes that are being made to the data sets stored on the Cache memory device, utilizing the non-volatile memory pages to store update records reflecting the changes in the data sets; a reconstruction process in which the data sets are reconstructed utilizing the update records stored on the non-volatile memory pages, and the data sets stored on the direct access storage device; and a process for freeing arbitrary non-volatile pages from their prior association with data sets that are stored in the Cache memory.

DASD-free non-volatile updates

Growing community of inventors

Yokneam, Israel

Tal Sivan