FW: Backup VLDB's

  • From: "Mark W. Farnham" <mwf@xxxxxxxx>
  • To: <oracle-l@xxxxxxxxxxxxx>
  • Date: Tue, 10 Nov 2009 07:12:41 -0500

snipped to fit

 

  _____  

From: Mark W. Farnham [mailto:mwf@xxxxxxxx] 
Sent: Tuesday, November 10, 2009 6:39 AM
To: 'hrishys@xxxxxxxxxxx'; 'gnahmed.c@xxxxxxxxxx'; 'oracle-l@xxxxxxxxxxxxx'
Subject: RE: Backup VLDB's

 

Usually when I hear or read the question "what is the fastest way to back
up?" the person should be asking "what is the fastest way to recover?" As
long as the process of taking a backup does not affect throughput and
response time in a material way, how long it takes is only relevant in the
context of "how fast can I deliver the required repair." Again, though, that
is recovery, not backup. How often, and under what circumstances, will a
full reload be required? How often, and under what circumstances will a
block repair be required? Think that through and design a recovery process
that meets your operational requirements and can be explained to and
understood by management. Triage of what to recover in what order is also
germaine in the context of restoring operations. There is a reason why
Oracle invested the effort to make it possible to get a database back up and
running with a few key tablespaces and files. Priority of recovery of the
tablespaces after that should take into account which operational features
cost a company the most per hour to have down. If your backup design cannot
restore files with priority without scanning lower priority files, that
represents a limitation to restoration flow. Whether that limitation is
important in dollars and cents is another issue. And of course your overall
system architecture with the possibility of standby databases held at
varying archive application delays may make the need for a full restore
vanishingly small even in the face of application error or sabotage. But
thinking these things through and deciding what your recovery requirements
are should be upstream of working on the sheer speed of the backup. Of
course it should not be overlooked that in some cases the fast technology
based backup means that a fast restore is simply the reverse operation.

 

Hrishy's comments are on target. Money equals speed. Or maybe money plus
brains equals speed.

 

Oracle's recovery capabilities are an enduring technological marvel in
conception, architecture, programming, and execution.

 

mwf

 

<snip>

Other related posts: