Optimization to permit block based incremental backup across system reboot or crash

Abstract

Techniques to back up data are disclosed. In various embodiments, a copy of a free block map as of a first time associated with a first backup is stored in persistent data storage. Writes made subsequent to the first backup to blocks not listed as free in the copy of the free block map as of the first time are tracked in a persistently-stored change block tracking log. A free block map as of a second time and the previously-stored copy of the free block map as of the first time are used to determine which blocks listed as free in the free block map as of the first time have been written to since the first time. At least a subset of blocks determined to have been written to since the first time are including in an incremental backup.

What is claimed is: 
     
         1 . A method, comprising:
 storing, by one or more processors, in a data storage a free block map, the free block map being stored in connection with a backup;   after the backup,
 tracking, in a persistently-stored change block log, a modification to one or more blocks that is modified subsequent to the backup based on whether the one or more blocks correspond to one or more blocks in the free block map as of a time of the backup that were not identified as free; and 
   performing, by one or more processors, an incremental backup based at least in part on the change block log.   
     
     
         2 . The method of  claim 1 , further comprising: resetting the persistently-stored change block log after the backup or the incremental backup. 
     
     
         3 . The method of  claim 1 , wherein the incremental backup is performed after a system crash or reboot that occurs after the backup. 
     
     
         4 . The method of  claim 1 , wherein the free block map is stored as of a first time associated with a first backup. 
     
     
         5 . The method of  claim 4 , further comprising:
 in response to the backup:
 premarking all blocks in the free block map that were identified as free as of a time of the backup as modified; and 
 premarking all blocks in a persistently-stored change block log that were identified as free as of the time of the backup as modified. 
   
     
     
         6 . The method of  claim 4 , further comprising;
 for each block write operation:
 determining whether a block associated with the block write operation is identified as modified in the persistently-stored change block log; 
 in response to a determination that the block is identified as modified in the persistently-stored change block log, maintaining an indication of the block in the persistently-stored change block log as being modified; and 
 in response to a determination that the block is not identified as modified in the persistently-stored change block log, updating the persistently-stored change block log to identify the block as modified. 
   
     
     
         7 . The method of  claim 1 , further comprising:
 determining whether the one or more blocks were identified as free in the free block map; and   in response to a determination that the one or more blocks were not identified as free in to the free block map, tracking the modification in an in-memory change block log.   
     
     
         8 . The method of  claim 1 , further comprising:
 determining which blocks listed as free in the free block map have been written to based at least in part on a comparison of a subsequent free block map and a previously-stored copy of the free block map;   adding one or more blocks to the persistently-stored change block log based on the comparison of the subsequent free block map and the previously-stored copy of the free block map; and   including in the incremental backup at least a subset of blocks indicated in the persistently-stored change block log as having had writes made to the subset of blocks indicated in the persistently-stored change block log subsequent to the first backup and at a least a subset of blocks included in the change block log.   
     
     
         9 . The method of  claim 1 , wherein tracking in the persistently-stored change block log the modification to one or more blocks that is modified subsequent to the first backup further comprises:
 intercepting a write to a sector or block;   checking an in-memory change block log to determine whether the sector or block has been marked as modified; and   marking the sector or block as modified if the sector or block has not already been marked as modified; and propagating the write to the persistently-stored change block log.   
     
     
         10 . The method of  claim 1 , wherein the free block map is stored in a persistent data storage, and wherein the persistent data storage comprises a local disk or other local drive. 
     
     
         11 . The method of  claim 1 , further comprising extracting the free block map from a file system with which the data to be backed up is associated. 
     
     
         12 . The method of  claim 1 , further comprising receiving an indication that a system with which the data to be backed up is associated has crashed or rebooted. 
     
     
         13 . The method of  claim 12 , further comprising reading from persistent data storage a previously-stored copy of the free block map; and using data from the previously-stored copy of the free block map to pre-mark as modified in the change block log at least a subset of blocks listed as free in the free block map. 
     
     
         14 . A system, comprising:
 a data storage device; and   one or more processors coupled to the data storage device and configured to:
 store on the data storage device a free block map, the free block map being stored in connection with a backup; 
 after the backup,
 track, in a persistently-stored change block log, a modification to one or more blocks that is modified subsequent to the first backup based on whether the one or more blocks correspond to one or more blocks in the free block map as of a time of the backup that were not identified as free; and 
 
 perform an incremental backup based at least in part on the change block log. 
   
     
     
         15 . A computer program product, the computer program product being embodied in a non-transitory computer readable storage medium and comprising computer instructions for:
 storing in a data storage a free block map, the free block map being stored in connection with a backup;   after the backup,
 tracking, in a persistently-stored change block log, a modification to one or more blocks that is modified subsequent to the first backup based on whether the one or more blocks correspond to one or more blocks in the free block map as of a time of the backup that were not identified as free; and 
   performing an incremental backup based at least in part on the change block log.