Resuming of interrupted database backups (17-Aug-2009)
Disclosed is a method for resuming an interrupted database backup at the point of failure. The method provides a backup interface utility which buffers the data to be backed up until they have been successfully transferred to a backup storage. In the case of a backup failure, e.g., when the TCP/IP connection is lost, the utility tries to resume the backup periodically based on the content of the buffer. The method enables to finish the backup successfully without restarting it. The advantages of the provided solution are: 1. Resuming of an interrupted database backup at the point of failure; 2. Allowing for dependencies on resources, e.g. CPU, network, storage manager; and 3. Increasing reliability of (scheduled) database backups.
Resuming of interrupted database backups
Database management systems (DBMS) provide mechanisms for creating backup images of databases mainly to protect its operational data against data loss. The backup images are usually stored, for example, on hard disks, on magnetic tapes, or on other storage devices. Typically, database backups are done at least once a
. These backups are scheduled to start automatically
-called backup window, that is, a given time slot. To execute the backups, required resources, e.g. Tivoli Storage Manager (TSM), must be reserved in advance.
Since a typical database backup involves a large amount of data, only one backup run per backup window can be started. This condition works fine if the backup finishes successfully. If the database backup fails for some reasons, e.g., due to a TCP/IP connection failure, a second try within the same backup window is almost impossible because of insufficient resources. To complete the failed backup, the DBMS backup process would have to read all database data again from the beginning, no matter whether some of the data had already been transferred successfully before the backup failure.
Today, backup utilities do not provide methods for resuming of interrupted database backups. The following flow chart illustrates a workflow of a prior art database backup (
without resumption of an interrupted backup
week right up to once a day
within a predefined so
).
1
For example, if a database backup with a size of 10 terabytes has successfully transferred about 99% of the data to the storage manager and the last data package to be transferred fails for some reason, the data already sent to the storage manager will be rolled back. The only way to complete the backup process successfully is to restart the complete database backup,
which may have a size of
. The allocation size of the backup
up to 10 terabytes of data.
A backup utility that is able to support resumable backups utilizes a backup buffer,
which is allocated and controlled by this utility
2
buffer is a multiple of...
Find More Documents Like This
Download this Disclosure
Click here if you want to sign-in now.
