13.12.2012 Views

MAA - Oracle 10gR2 Redo Transport and Network Best Practices

MAA - Oracle 10gR2 Redo Transport and Network Best Practices

MAA - Oracle 10gR2 Redo Transport and Network Best Practices

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Maximum Availability Architecture<br />

of redo volume is to collect Automatic Workload Repository (AWR) reports on the<br />

primary database during normal <strong>and</strong> peak workloads, <strong>and</strong> determine the number of<br />

bytes per second of redo data the production database is producing. You can then<br />

compare the speed at which redo is being generated with the Active Apply Rate<br />

columns in the V$RECOVERY_PROGRESS view to determine if the st<strong>and</strong>by<br />

database is able to maintain the pace.<br />

It is not necessary to tune redo apply If the st<strong>and</strong>by database is able to keep pace<br />

with the primary database during peak periods. If your examination of the<br />

performance statistics described above indicate that redo apply cannot keep pace,<br />

<strong>and</strong> that the resulting apply lag is not the result of an archive log gap, then proceed<br />

with tuning as described below.<br />

<strong>Best</strong> <strong>Practices</strong> for Tuning <strong>Redo</strong> Apply<br />

Before tuning the <strong>Redo</strong> Apply process, it is important to underst<strong>and</strong> the various<br />

wait events specific to <strong>Redo</strong> Apply.<br />

Parallel Recovery Wait Events<br />

The vast majority of wait events related to parallel recovery coordinators <strong>and</strong> slaves<br />

apply to the coordinator. Slaves are either applying changes (clocking on CPU) or<br />

waiting for changes to be passed from the coordinator. The key database wait<br />

events on a physical st<strong>and</strong>by database are:<br />

• Parallel Recovery Coordinator Wait Events<br />

• Parallel Recovery Slave Wait Events<br />

Parallel Recovery Coordinator Wait Events<br />

• Log file sequential read - The parallel recovery coordinator is waiting on I/O<br />

from the online redo log or the archived redo log<br />

• Parallel recovery read buffer free - All read buffers are being used by slaves,<br />

<strong>and</strong> usually indicates that the recovery slaves lag behind the coordinator.<br />

• Parallel recovery change buffer free - The parallel recovery coordinator is<br />

waiting for a buffer to be released by one of the recovery slaves. Again, this<br />

is a sign the recovery slaves are behind the coordinator.<br />

• Datafile init write - The parallel recovery coordinator is waiting for a file<br />

resize to finish, as would occur with file auto extend.<br />

• Parallel recovery control message reply - The coordinator has sent a<br />

synchronous control messages to all slaves, <strong>and</strong> is waiting for all slaves to<br />

reply.<br />

<strong>Oracle</strong> Active Data Guard: <strong>Oracle</strong> Data Guard 11g Page 30

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

Saved successfully!

Ooh no, something went wrong!