Recovery
Having spent so much time on how dbwr and lgwr work, it’s about time we looked at one of the most important tasks the DBA has to succeed at—recovery. When everything breaks, the DBA has to be able to recover the database. I’ve already mentioned instance recovery (i.e., what happens if the instance crashes): in principle we just start it again and Oracle simply replays all the entries from the redo log to bring the database up to the moment it crashed. The clever bit is how Oracle minimizes the amount of work it needs to do—and this is where the media recovery checkpoint, incremental checkpoint, and block written records come into play.
On a media recovery checkpoint, Oracle updates every data file header to hold the SCN at which the ...
Get Oracle Core: Essential Internals for DBAs and Developers now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
