Re: issues with oracle partitioning and SAN striping

IMHO that is a very bogus speculation. We have several large tables range partitioned by date, with different years in year-specific tablespaces, whose datafiles are on different mount points - all on SAN. It's very unlikely, though possible, that some of those end up on the same SAN spindle. That would be OK, especially if they were different month and/or year-specific segments, which are accessed with different frequency.

The huge advangages of partitioning WRT SQL performance, management of data, indexing, etc. far, far outweigh any miniscule possibility of
"collisions" with SAN striping. Your team member definitely needs further
enlightenment about partitioning.

Jack C. Applewhite - Database Administrator Austin (Texas) Independent School District 512.414.9715 (wk) / 512.935.5929 (pager)

 I feel so unnecessary. -- Rufus Thomas

               ( "Do the Funky Chicken")

ryan_gaffuri_at_comcast.net
Sent by: oracle-l-bounce_at_freelists.org
11/06/2006 09:02 AM
Please respond to
ryan_gaffuri_at_comcast.net

To
oracle-l_at_freelists.org
cc

Subject
issues with oracle partitioning and SAN striping

Someone on our team raised concerns about using both oracle table level partitioning and SAN striping. As I have said before I am more on the developer side of things. Has anyone had problems with this?  

"SAN striping of data file and Oracle partitions) techniques we are using
will sure collide with each other creating worse performance than using either one of the two. Since SAN is not aware of how oracle is handling the partitions, data from two hash or range partitions can end up on the same spindle thereby making it serial I/O."    

--
http://www.freelists.org/webpage/oracle-l