Another note on "Best Practices" for DataGuard, Oracle Services and RAC

From: Charles Schultz <sacrophyte_at_gmail.com>
Date: Fri, 26 Jun 2009 10:42:35 -0500
Message-ID: <7b8774110906260842j7f997761ld3fbabea8db8cf89_at_mail.gmail.com>



Still reading documentation. =)
http://www.oracle.com/technology/deploy/availability/pdf/MAA_WP_10gR2_DataGuardNetworkBestPractices.pdf

<http://www.oracle.com/technology/deploy/availability/pdf/MAA_WP_10gR2_DataGuardNetworkBestPractices.pdf>Pages 7 and 8 suggest setting the SDU to a larger value to accommodate redo log packets. However, the example given is for a SID called "sales". In other areas of MAA documentation, "sales" is the service name that applications connect to. Do we really want to set the SDU for normal application traffic? My understanding is that we would most likely want a separate service_name (and thus a independent path) from the primary to the standby, in which we can set the SDU as large as needed without impacting things like OLTP packets that are typically smaller. Additionally, since this is DataGuard, we will want a reverse-path for the same reason.

Additionally, I am not a network guy by any means; if a system is currently underutilizing existing resources, optimizing esoteric settings like the SDU and the send/receive buffers seems a bit overkill to me. So, in light of the context, would it be considered "Best Practice" to set these parameters regardless?

I still want to know who defines what the heck "Best Practice" means anyway. =) I wonder what would happen if I put that into all my documentation from here on out......

-- 
Charles Schultz

--
http://www.freelists.org/webpage/oracle-l
Received on Fri Jun 26 2009 - 10:42:35 CDT

Original text of this message