Following on from Cary's response, both internal and external testing
has shown the performance improvements of storing objects in tablespaces
with different block sizes to be so miniscule as to be not worth the
effort (or putting it another way, there are lots of better ways to
spend your performa
Based on Cary's paper regarding when to use an index, would there not be value
in having index tablespaces with a smaller block size vs tables using a larger
block size?
American Express made the following
annotations on 07/16/2003 10:11:40 AM
---
Cary,
On the same line, I want to propose a different thought - smaller block
sizes for index tablespaces to reduce the chance that a single block is
contended for by two different sessions, which indices the wait event
"buffer busy waits". Making them smaller, a typical index block will hold
less
Tracy,
I would have expected you to say the opposite: big blocks for index segments
(to reduce B*-tree height), and small blocks for table data (to improve
block selectivity).
It's a pretty expensive thing to implement though (assuming you're already
"up," the downtime to rebuild a tablespace cou
why would an index necessarily need to use a smaller tablespace? is this
article on hotsos? Which one is it?
- Original Message -
To: "Multiple recipients of list ORACLE-L" <[EMAIL PROTECTED]>
Sent: Tuesday, July 15, 2003 6:14 PM
>
> Based on Cary's paper regarding when to use an index, w
Based on Cary's paper regarding when to use an index, would there not be value
in having index tablespaces with a smaller block size vs tables using a larger
block size?
AM PST
Please respond to [EMAIL PROTECTED]
Sent by:[EMAIL PROTECTED]
To:"Multiple recipients of list O
