International Business Machines Corporation

Denis A Frank

Carlos Francisco Fuente

Joanna Katharine Brown

Christopher B E Beeken

Christopher B Beeken

Lee Jason Sanders

David Adam Sinclair

Roderick Guy Charles Moore

Payal Mehta

Disclosed are techniques for defragmentation in deduplication storage systems. Machine language determines using deduplication metadata that at least some of an incoming input/output stream is a duplicate of at least part of a source volume whose physical locations of its stored data are fragmented in backend storage. Subsequently, defragmentation is carried out on the stored data by using the incoming input/output stream to write the data into sequential chunks at new physical locations in the backend storage and updating the source volume location mappings to the new physical locations.

Defragmentation in deduplication storage systems

Managing storage volumes in a point-in-time copy cascade. The cascade includes a first source volume, a first target volume and a second target volume. The method detects a corruption of the first source volume and, if a first restoration of the first source volume from the first target volume is in progress, creates a third target volume from the second target volume and directs I/O operations for the first source volume to the third target volume. The first source volume is then restored from the third target volume.

Point-in-time copy restore

Point-in-time backup copies of a production volume that include volume maps between adjacent source and target volumes can be created in a cascade of data volumes. An empty volume map having the backup copy as the target can be created. The empty volume map can include a first bitmap used to indicate that data is stored in the source or the target, a second bitmap for use in managing and optimizing background copying, and an index for tracking a status of background copying. The volume map can be populated so that they first bitmap indicates data is stored in the source, and the index set to a start value. Production volume data regions can be sequentially background copied to the backup copy, while tracking using the index. Upon backup copy completion, the values of the volume map having the backup copy as the target can be stored.

Managing backup copies in cascaded data volumes

Point-in-time backup copies of a production volume that include volume maps between adjacent source and target volumes can be created in a cascade of data volumes. An empty volume map having the backup copy as the target can be created. The empty volume map can include a first bitmap used to indicate that data is stored in the source or the target, a second bitmap for use in managing and optimizing background copying, and an index for tracking a status of background copying. The volume map can be populated so that the first bitmap indicates data is stored in the source, and the index set to a start value. Production volume data regions can be sequentially background copied to the backup copy, while tracking using the index. Upon backup copy completion, the values of the volume map having the backup copy as the target can be stored.

Managing storage volumes in a point-in-time copy cascade. The cascade includes a first source volume, a first clone point-in-time target volume, and a second point-in-time target volume that is a point-in-time copy of the first point-in-time target volume. Grain background copies from the first source volume to the first clone point-in-time target volume are tracked, and, if the first clone point-in-time target volume is corrupted, a third point-in-time target volume is created from the second point-in-time target volume. I/O for the first clone point-in-time target volume is directed to the third point-in-time target volume. The first clone point-in-time target volume is cleaned to the second point-in-time target volume. The first clone point-in-time target volume is restarted from the third point-in-time target volume and each grain of the first clone point-in-time target volume that has not been background copied is copied from the cascade to the first clone point-in-time target volume.

Restoring a clone point-in-time copy

Managing storage volumes in a point-in-time copy cascade. The cascade includes a first source volume, a first target volume and a second target volume. The method detects a corruption of the first source volume and, if a first restoration of the first source volume from the first target volume is in progress, creates a third target volume from the second target volume and directs I/O operations for the first source volume to the third target volume. The first restoration is stopped. The first source volume is restored from the third target volume, and the third target volume is deleted.

A data receiver module receives, at a storage device simulator, a data transmission from a storage controller being tested. The data transmission includes data and metadata. The metadata is associated with the data. A signature receiver module receives a signature from the storage controller as part of the data transmission. The signature is used to distinguish the metadata from the data. A data/metadata determination module examines the data transmission and determines data from metadata using the signature. A metadata storage module stores the metadata of the data transmission on the storage device simulator in response to the data transmission including metadata. The data storage simulator includes a data storage device. A data discard module discards the data of the data transmission in response to the data transmission including data.

Growing community of inventors

Winchester, United Kingdom

Denis A Frank