24.12.2012 Views

Backing Up Oracle - Computing at Cornell

Backing Up Oracle - Computing at Cornell

Backing Up Oracle - Computing at Cornell

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Figure 4. Log files<br />

D<strong>at</strong>abase<br />

Some RDBMSs support the use of log files to perform forward recovery.<br />

Forward recovery takes advantage of the fact th<strong>at</strong> log files hold details of all<br />

changes th<strong>at</strong> have been made to the d<strong>at</strong>abase, and therefore you do not<br />

necessarily need to undo changes but instead can reapply changes. With<br />

forward recovery, the recovery process can:<br />

Restore a d<strong>at</strong>abase to the st<strong>at</strong>e it was in <strong>at</strong> the time the last backup was<br />

taken<br />

Use the log files to reapply the changes th<strong>at</strong> had been made since the last<br />

backup was taken<br />

Back out (undo) any partially completed changes.<br />

A standard RDBMS concept rel<strong>at</strong>ed to log files is the checkpoint process. All<br />

RDBMSs use buffers in memory to hold changes to the d<strong>at</strong>abase and log<br />

files. The purpose of buffers is to improve the oper<strong>at</strong>ional performance of the<br />

RDBMS. However, the use of buffers means th<strong>at</strong> most changes to d<strong>at</strong>abases<br />

and log files do not get written to disk until some time after the RDBMS has<br />

indic<strong>at</strong>ed to the user applic<strong>at</strong>ion th<strong>at</strong> the upd<strong>at</strong>e has been made successfully.<br />

Checkpoints ensure th<strong>at</strong> all d<strong>at</strong>abase and log file changes held in the<br />

RDBMS' buffers are flushed out to disk. This shortens the time it takes to<br />

recover a d<strong>at</strong>abase after a system crash because the number of redundant<br />

log records processed during the recovery is reduced. All RDBMSs support<br />

checkpoints and issue them autom<strong>at</strong>ically <strong>at</strong> intervals.<br />

It is recommended th<strong>at</strong> log files be mirrored or duplexed.<br />

14 <strong>Backing</strong> <strong>Up</strong> <strong>Oracle</strong> using Tivoli Storage Management<br />

records changes<br />

to DB<br />

Log Files

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!