Data backup system with methods for stripe affinity backup to multiple archive devices
AssigneeSybase, Inc.
InventorTimothy E. Wood
Life of this patent
Estimating the term…
Abstract
A data backup system implements coordination between a Database Server and a Backup Server to produce a recoverable database dump. By utilizing a technique referred to as stripe affinity, a mechanism is disclosed for ensuring the integrity of a database backup made to multiple archive devices simultaneously. In addition, by utilizing stripe affinity, archived data may be reloaded from fewer archive devices than were used to make the original backup. A task scheduler mechanism allocates processor time among the tasks that comprise the backup system. In this way the I/O service tasks can process their event queues while the current set of allocation pages are also being processed.
Claims21
In a computer system comprising a plurality of storage devices and a plurality of archive devices, each storage device storing information as a plurality of data pages on a storage media, a method for transferring information stored on said storage devices to said archive devices, the method comprising:
- (a) for each storage device, grouping data pages for the storage device into a sequence of extent groups, each extent group comprising some of the data pages for the storage device;
- (b) for each storage device, associating said extent groups of the storage device with respective archive devices, so that each archive device has associated therewith a plurality of corresponding extent groups from multiple ones of said storage devices;
- (c) transferring information stored on the storage devices to said archive devices by transferring to each of the archive devices those extent groups associated with that archive device so that each storage device's extent groups are distributed among multiple ones of said plurality of archive devices and each archive device contains data pages from multiple ones of said plurality of storage devices.
In an on-line transaction processing system comprising a database server storing an on-line database, said on-line database comprising a plurality of data records, said database server processing in real-time transactions which modify content of some of said data records, a two-phase method for creating a backup copy of said on-line database, the method comprising:
- (a) while continuing to process transactions in real-time which modify content of some of said data records, copying in a first phase all said data records to a backup server regardless of whether a particular data record is subject to modification by a transaction which has not completed processing;
- (b) concurrently with step (a), storing log information describing which of said data records were subject to modification during said first phase;
- and
- (c) based on said log information stored in step (b), copying in a second phase those data records which were subject to modification during said first phase to said backup server so that said data records copied in said second phase account for data records that were changed during said first phase.
In a database server system, a method for backing up to a plurality of archive devices a database comprising a plurality of data pages stored on a plurality of storage devices, said database processing in real-time transactions which modify content of some of said data pages, the method comprising:
- (a) for each storage device, dividing data pages stored on said each storage device into distinct groups;
- (b) assigning one group from each storage device to each archive device, until each group of each storage device is assigned to a particular archive device;
- (c) concurrently for each storage device and while continuing to process transactions in real-time which modify content of some of said data pages, copying successive groups from said each storage device to the respective assigned archive devices regardless of whether a particular data page is subject to modification by a transaction which has not completed processing;
- and
- (d) repeating step (c) for any data page subject to modification by a transaction which has not completed processing during initial operation of step (c).
Dependencies are inferred from the claim text.