US 7,680,998 B1Granted

Journaled data backup during server quiescence or unavailability

AssigneeEMC Corporation

InventorsScott Auchmoody, Eric Olsen, Scott Ogata

Life of this patent

Estimating the term…

Loading patent…

Abstract

A backup is performed by a client at a time when a backup server is unable to process the backup. The client maintains a cache including a root tag vector and hash entries. The client begins a backup by writing the root tag vector to a journal file and breaking files into pieces. For each piece, the client performs a hash and compares the resulting hash to entries in the cache. If the hash does not match any entries, the client records a request in the journal file to add the corresponding piece of data to an archive. After completing the backup, the journal file can be sent to the server. Before processing the journal file, the server validates the root tag vector. If the root tag vector is valid, the server processes each of the requests to add data. Otherwise, the server discards the journal file.

Claims16

Dependent claims

A method of generating a backup file on a client, the method comprising:

determining by a client that a hash file system server is not currently able to contemporaneously process a backup from the client, the client including a computer device, wherein the backup is a backup of a file directory accessible to the client;
breaking each file of the file directory into one or more pieces;
writing by the client a root tag vector to a journal file accessible to the client, wherein the root tag vector includes one or more root hashes stored in one or more caches accessible to the client and wherein the one or more caches additionally store one or more entries representing data protected by the one or more root hashes, wherein each of the one or more entries includes a tag field that is n bits long and a hash of a piece of data, wherein the n bits in the tag field correspond to the one or more root hashes stored in the one or more caches of the client, and each tag bit indicates whether a corresponding root hash protects the entry;
determining by the client that the one or more pieces of data of the backup are not identical to data represented by the one or more entries by performing a hash on each of the one or more pieces of data and comparing each hash to the one or more entries;
and
recording by the client, in the journal file on a storage device, one or more requests to store the one or more pieces of data, wherein the one or more requests to store the one or more pieces of data can be processed by the hash file system server at a later time when the hash file system server is able to process the backup.

3 more elements

A method of processing a journaled backup file generated by a client, the method comprising:

receiving, at a server, a journal file of a backup from a client, wherein the client generated the journal file in response to determining that the server was unable to contemporaneously process the backup from the client and the client stored the journal file until the server was able to process the backup, wherein the server includes a computing device and the client includes a computing device and wherein the journal file includes:
a root tag vector identifying one or more root hashes accessible to the client at the time of generating the journal file, wherein the one or more root hashes point to data accessible to the server in the past or currently;
and
one or more requests to store one or more pieces of data of the backup, the one or more pieces of data not protected by any of the one or more root hashes;
validating the root tag vector by determining that the one or more root hashes point to data currently accessible to the server in part by identifying bits in a tag field n bits long, wherein the n bits correspond to the maximum number of root hashes that can be stored in one or more caches of the client, and wherein each tag bit indicates whether a corresponding root hash protects the entry;
and
processing the one or more requests to store one or more pieces of data in a storage device.

4 more elements

Dependencies are inferred from the claim text.