Flashback Database, Table, Query, Transaction and Drop provides an extremely useful set of tools to the DBA. Once understood they are easily configured, implemented and used, in case of need.
They are not only useful for recovery purposes but sometimes for informational purposes as well, a question like: which was the value for some record the last 6 hours? can be easily answered using flashback records.
Metalink Note 435998.1 : What Do All 10g Flashback Features Rely on and what are their Limitations ? contain additional conceptual information and links to some of the sources I did use to prepare the document I'm attaching here:
« Data Guard, Broker, Observer and Switchover, Step by Step | Main | 10g Dataguard, Flashback and New Features Q&A »
Oracle Flashback Technologies
TrackBack
TrackBack URL for this entry:
http://blogs.oracle.com/mte1521/mt-tb.cgi/4305
Search All Blogs
Google Search
About This Entry
This page contains a single entry from the blog posted on October 8, 2007 11:08 AM.
The previous post in this blog was Data Guard, Broker, Observer and Switchover, Step by Step.
The next post in this blog is 10g Dataguard, Flashback and New Features Q&A.
Many more can be found on the main index page or by looking through the archives.
Comments (2)
The only issue I had when evaluating Flashback, was the additional I/O generated. When we set up our 10g system, using a NetApp Filer as the central storage on a 2-Node RAC on Solaris 10 , inserts were taking twice the time with flashback on. As we wanted to maxismise throughput, we did not use Flashback.
I'd appreciate any thoughts on this.
Posted by Bob McAusland | October 11, 2007 9:22 AM
Posted on October 11, 2007 09:22
Flashback need to write flashback logs to some destination that you need to define, nevertheless the flashback logs are not as critical they need to be located on disk performing as well as the database datafiles disks.
You can also enable trace with event 10046 level 12 to see which is the difference in terms of waits, it should be some overhead because you are doing more IO but 50% seems to be too much
Posted by Alejandro Vargas | October 29, 2007 4:14 PM
Posted on October 29, 2007 16:14