| Author |
which would be faster ?
|
|
| Hassan 2005-07-28, 11:26 am |
| I have 2 tables that i want to do an insert select from
from QA Window 1
insert <newtable>
select * from <oldtable>
insert <newtable1>
select * from <oldtable1>
----------------------------
QA window 2
insert <newtable>
select * from <oldtable>
----------------------
QA window 3
insert <newtable1>
select * from <oldtable1>
----------------------------
Will QA Window 1 where I have the 2 inserts running serially be slower than
when I run QA window 2 and 3 in parallel, assuming they are all on the same
disk or will it be the same ? Just seeking answers based on experience and
theory
| |
| David Gugick 2005-07-28, 11:26 am |
| Hassan wrote:
> Will QA Window 1 where I have the 2 inserts running serially be
> slower than when I run QA window 2 and 3 in parallel
Probably, but no way to tell without testing. You'll increase the disk
IO executing in parallel.
--
David Gugick
Quest Software
www.imceda.com
www.quest.com
| |
| Hassan 2005-07-28, 1:25 pm |
| So 2 is faster ?
"David Gugick" <david.gugick-nospam@quest.com> wrote in message
news:%23LR7Z64kFHA.3960@TK2MSFTNGP12.phx.gbl...
> Hassan wrote:
>
> Probably, but no way to tell without testing. You'll increase the disk
> IO executing in parallel.
>
>
>
>
> --
> David Gugick
> Quest Software
> www.imceda.com
> www.quest.com
>
| |
| Chandra 2005-07-28, 1:25 pm |
| hi Hassan
just see the execution plan. u might get any idea. the result also depends
on the data that u are using
--
best Regards,
Chandra
http://chanduas.blogspot.com/
http://groups.msn.com/SQLResource/
---------------------------------------
"David Gugick" wrote:
> Hassan wrote:
>
> Probably, but no way to tell without testing. You'll increase the disk
> IO executing in parallel.
>
>
>
>
> --
> David Gugick
> Quest Software
> www.imceda.com
> www.quest.com
>
>
| |
| David Gugick 2005-07-28, 1:25 pm |
| Hassan wrote:
> So 2 is faster ?
I'll repeat myself :-)
Probably, but no way to tell without testing.
--
David Gugick
Quest Software
www.imceda.com
www.quest.com
| |
| Gert-Jan Strik 2005-07-28, 8:23 pm |
| I did a little experiment for this scenario with 150 MB in "oldtable"
and 150 MB in "oldtable1". There was no significant performance
difference. All was done on the same one SCSI disk (both sources and
both destinations).
Then I did the same experiment, but this time "oldtable" and "newtable"
were on a different disk from "oldtable1" and "newtable1". It will come
as no surprise that in this case the seperate processes were
considerably faster.
HTH,
Gert-Jan
Hassan wrote:
>
> I have 2 tables that i want to do an insert select from
>
> from QA Window 1
>
> insert <newtable>
> select * from <oldtable>
>
> insert <newtable1>
> select * from <oldtable1>
>
> ----------------------------
> QA window 2
>
> insert <newtable>
> select * from <oldtable>
>
> ----------------------
>
> QA window 3
>
> insert <newtable1>
> select * from <oldtable1>
>
> ----------------------------
>
> Will QA Window 1 where I have the 2 inserts running serially be slower than
> when I run QA window 2 and 3 in parallel, assuming they are all on the same
> disk or will it be the same ? Just seeking answers based on experience and
> theory
|
|
|
|