RelationalDBDesign RelationalDBDesign

Manual | Automatic Archiving   «Prev  Next»
Lesson 3Normal ARCH processing
ObjectiveDiscuss the flow of events in normal ARCH archive processing.

Normal ARCH Processing

Let us review the normal flow of events while ARCH copies the online redo log files to the archive destination. The normal processing of ARCH starts with the opening of the log file members. If none of the log file members can be opened, ARCH signals an error. Otherwise, all the members that can be opened are opened. For instance, if you have three members in a redo log file group and one is bad, ARCH opens the other two and proceeds with the archive process.
ARCH then reads the headers of the redo log file members and validates the information with the control file information. ARCH then reads a group of data blocks, verifies the blocks and copies them to the archive destination. In earlier versions of Oracle, the data was not verified and you did not know that you had a bad block until you did a restore, not the best time to be notified. ARCH does not check the contents of the data blocks.
ARCH rotates reading blocks from each log file. If a log file member is read successfully, the log file is noted as good. If the next log file member read fails, ARCH tries to read the corresponding blocks from the log file that was noted as good. If a log file member fails three times, it is considered bad and is not used for subsequent operations. If you have log file members on different drives, your disk IO is distributed among the drives. Let us review this process pictorially with a system containing two redo log groups and two members per group.
Redo Log Groups 1
1) Redo Log Groups 1
Redo Log Groups 2
2) Redo Log Groups 2
Redo Log Groups 3
3) Redo Log Groups 3
Redo Log Groups 4
4) Redo Log Groups 4
Redo Log Groups 5
5) Redo Log Groups 5

The next lesson demonstrates how to manually archive online redo log files.