Re: Parallelism algorithm
From: Ami Levin (XXX_NOSPAM___XXX__amlevin_at_mercury.com)
Date: 01/02/05
- Next message: Steve Kass: "Re: Simple Bulk Insert Problem"
- Previous message: David Portas: "Re: trigger won't fire when update table (with commit) used within stored procedure"
- In reply to: Uri Dimant: "Re: Parallelism algorithm"
- Next in thread: Uri Dimant: "Re: Parallelism algorithm"
- Reply: Uri Dimant: "Re: Parallelism algorithm"
- Messages sorted by: [ date ] [ thread ]
Date: Sun, 2 Jan 2005 16:02:42 +0200
Hi Uri,
Thank you for your reply.
That does not answer my question.
What you are actually saying is "don't bother, who cares", well - I do.
I wouldn't have posted this question if it wasn't important (for me).
No offense meant, but I have to say I disagree with your attitude of "Nobody
knows and who even cares".
I think this attitude is both wrong and dangerous.
I see this as a sure way to mediocrity in your profession and with it you
are bound to design less than optimal solutions.
I try to take my profession very seriously and for me, a thorough
understanding of these "small details" is very important. It can make all
the difference.
Celko mentioned in this forum the case of the hospital's DB where such small
"who cares" details could mean the difference between life and death.
Thank you again and happy new year.
Ami
"Uri Dimant" <urid@iscar.co.il> wrote in message
news:ensmg2M8EHA.1408@TK2MSFTNGP10.phx.gbl...
> Ami,shalom
> Look, I avoid changing the server setting and I have a few queries when
we
> use MADOP hint. Nobody knows how does SQL Server work exactly internally
( i
> mean actually seeing the algoritm). If you put the hint and it does the
job
> , i'd not bother about an algoritm so I'm pretty certain that it is not
the
> same.
>
>
> "Ami Levin" <XXX_NOSPAM___XXX__amlevin@mercury.com> wrote in message
> news:OxvD3TM8EHA.3336@TK2MSFTNGP11.phx.gbl...
> > Hi all and happy new year,
> >
> > I have an architectural question regarding SQL Server's parallelism
> > algorithm:
> >
> > Let's say I have a quad CPU machine and affinity mask is set to use all
> four
> > CPUs.
> > If I change the server's setting "Max degree of parallelism" to 3, will
> the
> > parallel plans always use the same 3 CPUs, let's say 0,1 and 2 or will
the
> > optimizer balance the load among the 4 available CPUs using 0,1 and 2
for
> > one query, 1,2 and 3 for another query etc.?
> > Will the algorithm be the same when using MAXDOP hint instead of
changing
> > the server setting?
> >
> > TIA
> >
> > Ami
> >
> >
>
>
- Next message: Steve Kass: "Re: Simple Bulk Insert Problem"
- Previous message: David Portas: "Re: trigger won't fire when update table (with commit) used within stored procedure"
- In reply to: Uri Dimant: "Re: Parallelism algorithm"
- Next in thread: Uri Dimant: "Re: Parallelism algorithm"
- Reply: Uri Dimant: "Re: Parallelism algorithm"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
