Standby Redo Logs

Of course, Standby Redo Logs are obligatory if you are using a protection level of Maximum Protection or Maximum Availability, but I think any dba who is even using a dataguard configuration with a protection level of maximum performance would be nuts to not use Standby Redo Logs (SRLs). If it’s all about protecting the data, then basically you will lose less data in the event of having to perform a failover if your configuration has SRLs. When you throw into the mix the ability to send the redo via LGWR you can set up your dataguard environment to provide a high degree of data protection with an absolutely minimal impact on your primary database. When you are using SRLs, the LGWR process on the primary talks to the RFS process on the standby which is responsible for writing the redo information to disk, and it’s therefore the RFS process which writes the information to the SRLs.

Sure, if you need an absolute guarantee of NO data loss then maximum performance may not be the acceptable mode for your organisation, but if you can live with a small window of data loss then maximum performance combined with LGWR transmission and SRLs may be the sweet spot for you. On a 10.2 system running LGWR ASYNC with a Round Trip Time (RTT) of around 15ms between primary and standby sites, I typically see a redo transport lag of a handful of seconds. You also can only use the real time apply mode if you are using SRLs, and this may (or may not) be an added benefit for you.

The syntax for creating your SRLs is very similar to that for creating the online redo logs:

sql> alter database add standby logfile thread 1 group 42 ‘/path/to/logfile’ size 512M;

To delete them:

sql> alter database drop standby logfile group 42;

You can find out information on how the SRLs are configured on your system by looking at the V$logfile or the V$standby_log views.

Advertisements

4 thoughts on “Standby Redo Logs

  1. hi neil,

    i read this blog entry with great interest but is it possible that you made a typo ?

    “Sure, if you need an absolute guarantee of NO data loss then maximum performance may not be the acceptable mode for your organisation, but if you can live with a small window of data loss then maximum protection combined with LGWR transmission and SRLs may be the sweet spot for you.”

    i think you mean maximum performance instead of maximum protection, since maximum protection uses the lgwr sync…

    krgds,

    Philippe

  2. Hi Philippe,

    First of, thank you for reading!

    You are absolutely correct! Thank you for the correction.

    cheers,

    jason.

  3. Hi,

    I’m a new to Data Guard and I’m setting DG up between two Sun SPARC Solaris servers with Oracle EE 10gR2 (10.2.0.3). Do you have a reliable method to test RTT between the primary and standby servers so I can determine how much latency there will be? Thanks.

    -Jeff

    • Hi Jeff,

      I’d have a look at ping. I’m also looking to do a blog entry at some point on the wait event LNS WAIT on SENREQ which encompasses among other things the rtt. If you are 11g, you may also be look at the view V$REDO_DEST_RESP_HISTOGRAM.

      regards,

      jason.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s