Oracle Redo is oftentimes misunderstood, but most developers and DBAs assume it to mean that in the case of instance failure the database can 'look' here in an attempt to 'redo' operations in a recovery situation.
And they are essentially right. The redo logs are a log of all the changes made to the data. The logs also contain undo information so the data can be 'unwound' to the starting position. What they do not contain is data changes for which we have asked that there be no logging i.e. tables created with NOLOGGING or truncate operations.
But how do we measure our REDO? You can query v$my_stat, see how much redo you generated so far, update or insert a couple of rows, requery and subtract. Nice and simple.
Remember doing row-by-row operations with commits generates a lot more redo than a one-hit SQL statement before commit.
The amount of redo generated for each operation for the same SQL statement varies from release to release.
No comments:
Post a Comment