What additional capabilities do you feel that being
able to force a table to stay in memory brings to the product If a
given table's pages are "hot" (read from/written to a lot), they'll stay
in the buffer cache anyway. And I'm assuming you wouldn't want
to pin a table that's NOT used very often--doing so would essentially waste
the memory, starving other tables out of RAM that are potentially used a lot
more often and can make better use of repeated cache hits. So I personally
don't see a use case for PINTABLE... I'd be very interested in
what scenarios you have in mind.
This
post has been edited either by the author or a moderator in the Microsoft
Forums: http://forums.microsoft.com I am shocked to read this
document...
The scenario you describe would not be helped by
pinning. Pinning does not tell SQL Server to load the table into memory;
it only tells it not to clear that memory out of the buffer cache once it's
there. If your app is only using 2 GB of RAM, pinning would not increase
that amount, or even give you any additional buffer page lifetime -- since you
have no memory pressure, those pages are not getting expired
anyway.
We have 8 GB of memory on our server and barely use 2GB on a regular basis.
We would like to see if pinning some reference tables in memory (regardless of
how often they are used) increases our performance. Our data is stored on a
SAN and sometimes IO is not all that was promised.
We would rather have the option to use this feature
rather than it being discontinued. Otherwise, we are left relying 100% on SQL
server to make the decision for us - sure, most of the time it makes the right
choice but there is always that strange case where you might need this
functionality.
We have 8 GB of memory on our server and barely use 2GB on a regular basis. We would like to see if pinning some reference tables in memory (regardless of how often they are used) increases our performance. Our data is stored on a SAN and sometimes IO is not all that was promised.
We would rather have the option to use this feature rather than it being discontinued. Otherwise, we are left relying 100% on SQL server to make the decision for us - sure, most of the time it makes the right choice but there is always that strange case where you might need this functionality.
Why 2005 is missing PIN TABLE?
Morris LN
--
Adam Machanic
Pro SQL Server 2005, available now
http://www..apress.com/book/bookDisplay.html bID=457
--
Shibu Bhattarai
--
Adam Machanic
Pro SQL Server 2005, available now
http://www..apress.com/book/bookDisplay.html bID=457
--
shade29450
We have 8 GB of memory on our server and barely use 2GB on a regular basis. We would like to see if pinning some reference tables in memory (regardless of how often they are used) increases our performance. Our data is stored on a SAN and sometimes IO is not all that was promised.
We would rather have the option to use this feature rather than it being discontinued. Otherwise, we are left relying 100% on SQL server to make the decision for us - sure, most of the time it makes the right choice but there is always that strange case where you might need this functionality.