Frequency of joins between two tables in a database.

Hi All,
Is there anyway to tell the frequency of joins between two tables in a
database?. I need help on this one.
Thanks.
Message posted via http://www.webservertalk.comI don't know of any way to profile this, but what you could do
reasonably quickly is build up a quick structure mapping queries to the
tables they join, then run a profiler, log all the queries run into a
table, and thus be able to interrogate this. However that will assume
that your queries always join the same tables.|||Hi
Can you say what you need this for? If you are looking at de-normalisation
then you may want to base that on actual performance.
If you script your code you may get some idea of how often a table is joined
by using test searches.
John
"Naana via webservertalk.com" <u14055@.uwe> wrote in message
news:5ef609f8fb256@.uwe...
> Hi All,
> Is there anyway to tell the frequency of joins between two tables in a
> database?. I need help on this one.
> Thanks.
> --
> Message posted via http://www.webservertalk.com|||The reason to get this information is, I'm trying to split a large DB( 700gb
)
to group the join tables on one drive for performance purposes.
Thanks.
John Bell wrote:
>Hi
>Can you say what you need this for? If you are looking at de-normalisation
>then you may want to base that on actual performance.
>If you script your code you may get some idea of how often a table is joine
d
>by using test searches.
>John
>
Message posted via http://www.webservertalk.com|||I could be wrong, but won't it be better to have the joined tables on
different drives, so multiple IO paths can be used at the same time? Or are
we talking about moving certain data to faster drives and leaving other data
on slower drives?
"Naana via webservertalk.com" <u14055@.uwe> wrote in message
news:5ef800aece75b@.uwe...
> The reason to get this information is, I'm trying to split a large DB(
700gb)
> to group the join tables on one drive for performance purposes.
> Thanks.
> John Bell wrote:
de-normalisation
joined
> --
> Message posted via http://www.webservertalk.com|||Hi
You can look profile and look at high reads/writes as well as duration, that
will hopefully give some indication where you can split things up. You may
also want to split indexes onto different spindles, possibly looking at ITW
to show you some indication of what indexes are frequently used.
Also you can try creating multiple files on different drives that belong to
the same filegroup. The other option is to add extra discs to an existing
array so splitting I/O over more spindles.
John
"Naana via webservertalk.com" <u14055@.uwe> wrote in message
news:5ef800aece75b@.uwe...
> The reason to get this information is, I'm trying to split a large DB(
> 700gb)
> to group the join tables on one drive for performance purposes.
> Thanks.
> John Bell wrote:
> --
> Message posted via http://www.webservertalk.com|||And... don't forget move tempdb onto it's own spindles, possibly splitting
it into multiple files as well as moving the system databases into their own
discs.
John
"Naana via webservertalk.com" <u14055@.uwe> wrote in message
news:5ef800aece75b@.uwe...
> The reason to get this information is, I'm trying to split a large DB(
> 700gb)
> to group the join tables on one drive for performance purposes.
> Thanks.
> John Bell wrote:
> --
> Message posted via http://www.webservertalk.com|||Hi Jim
I had mis-read this... yes you would want to split I/O. I prefer a more
quantative approach i.e.target what is bad!
John
"Jim Underwood" <james.underwoodATfallonclinic.com> wrote in message
news:uz60VkwYGHA.3684@.TK2MSFTNGP05.phx.gbl...
>I could be wrong, but won't it be better to have the joined tables on
> different drives, so multiple IO paths can be used at the same time? Or
> are
> we talking about moving certain data to faster drives and leaving other
> data
> on slower drives?
>
> "Naana via webservertalk.com" <u14055@.uwe> wrote in message
> news:5ef800aece75b@.uwe...
> 700gb)
> de-normalisation
> joined
>|||We are thinking of moving certain large tables that are frequently access to
faster drives and leaving other data on slower drives.
Jim Underwood wrote:
>I could be wrong, but won't it be better to have the joined tables on
>different drives, so multiple IO paths can be used at the same time? Or ar
e
>we talking about moving certain data to faster drives and leaving other dat
a
>on slower drives?
>
>[quoted text clipped - 17 lines]
Message posted via webservertalk.com
http://www.webservertalk.com/Uwe/Forum...amming/200604/1