Re: Performance Newsgroup
From: Roberto Ruiz (robert5ok_at_yahoo.com)
Date: 09/19/04
- Next message: David H. Lipman: "Re: Earn $$$$$$$ FAST! Piece of cake. -xb:0"
- Previous message: Pegasus \(MVP\): "Re: boot problem"
- In reply to: Dan Seur: "Re: Performance Newsgroup"
- Next in thread: Dan Seur: "Re: Performance Newsgroup"
- Reply: Dan Seur: "Re: Performance Newsgroup"
- Messages sorted by: [ date ] [ thread ]
Date: Sun, 19 Sep 2004 11:55:19 -0500
Dan, thanx for you answer. Actually I'm trying to start studying the
subject, I've been about doing it for some time, but now I have a specific
problem: I wanted to redirect and centralize the My Documents and Desktop
folders for all users and bought two new computers equiped with IDE RAID
Controllers (Promise Fastrak SX4000) to act as file servers, in an attempt
to attain a quite high capacity and performance without embarking in an
expensive SCSI RAID solution, I mounted them with 4 hard drives in a RAID 5
configuration and the result is that the servers are performing quite slow
and I have no point of comparison because the whole thing is new (the
hardware and the distribution of information). As a matter of fact, users
(who actually don't know where the documents are) have not complained of
poor performance, but personally I don't feel it good enough, furthemore,
when I log on locally on the servers the responsiveness is miserable. I
think that this could be because the server is "concentrated" in her file
server work. The point is that I'd like to find the way to know if this
behavior is normal according to the load that the server is handling, if the
memory is being taken advantage of, if the RAIDs have actually improved the
disk subsystem's performance, if it would be better to implement another
RAID configuration or if the Fastrak SX4000 sucks and I better implement a
software RAID or something else.
I have read the Windows documentation on this performance subjects and it's
not quite clear, the figures of memory utilization in Task Manager seem
extrange to me and the counters I have measured in Perfomance Monitor
frecuently seem contradictory and they donīt tell me anything about simple
performance facts of the hardware like pure read and write speeds.
This is a comment anyway, I'm just looking for more information and some
environment to share and discuss my tests and observations.
Thanx,
Robert
"Dan Seur" <click@casta.net> wrote in message
news:uE7VjoanEHA.3464@tk2msftngp13.phx.gbl...
> Roberto - indeed I didn't understand. I thought you were embarking on
> some scholarly pursuit.
>
> Is there a specific "irresponsiveness" problem you are trying to
> diagnose? There may be a simple solution, if such a problem exists and
> you describe it clearly. Gradual hard drive failure, for example, can
> lock up a system for increasingly long times (eventually even hours) in
> certain circumstances.
>
> I'm unable to help with data reduction/analysis for various performance
> monitor output streams.
>
> Roberto Ruiz wrote:
>
> > Hi again,
> > Maybe I didnīt make my point clear enough. It's true that some tricks
may do
> > the job sometimes when trying to tweak performance, but I was actually
> > talking about studying performance in the most orthodox way, that is:
using
> > Performance logs and counters to monitor the performance of a machine
and
> > extracting some useful information from those endless graphics and
numbers.
> > For example:
> > - A computer is extremely irresponsive and when I take a look at the
task
> > manager I see that the processor is practically idle, there is plenty of
> > free memory so there canīt or shouldnīt be any important paging activity
so
> > I say what the hell is going on? Maybe there is a lot of disk reading,
maybe
> > it's browsing the network looking for I donīt know what ....
> > The numbers given by the performance monitor are dificult to interpret
and
> > seem contradictory at times, even when you know something about computer
> > hardware and OS architecture ..... thatīs the kind of information Iīm
> > looking for ....
> > Thanx anyway,
> > Robert
> >
> >
> > "Dan Seur" <click@casta.net> wrote in message
> > news:e2sGD$RnEHA.1992@TK2MSFTNGP10.phx.gbl...
> >
> >>You might Google for W2k "performance tweaks" and similar search args.
> >>You're right, it's a complex arena - but it is by no means unexplored.
> >>The gaming community, and to some extent the server community, who try
> >>for every ounce of horsepower, have batteries of tricks. As always, you
> >>travel these roads at your own risk. It's often the case that optimal
> >>performance, predictability, and stability intersect at different points
> >>for different purposes.
> >>
> >>Vanilla W2k benefits from years of experience with fairly complex
> >>algorithms across the spectrum of many manufacturers' and universities'
> >>research and design teams, but in a given case a registry tweak or two
> >>may improve certain characteristics. Mostly, it's math; down and dirty
> >>queueing theory, statistics, etc.
> >>
> >>
> >>
> >>Roberto Ruiz wrote:
> >>
> >>
> >>>Hi all,
> >>>I'm trying to start digging seriously into performance issues,
something
> >>>that I believe is really a complex and quite unexplored arena and I
came
> >
> > to
> >
> >>>the newsgroups looking for something like a performance newsgroup, but
> >
> > I've
> >
> >>>found there's nothing like this here. I'd appreciate any suggestions on
> >>>resources about this subject (Newsgroups, Websites, etc)
> >>>
> >>>Thanx in advance,
> >>>Robert
> >>>
> >>>Roberto Ruiz
> >>>Brainbench MVP for Windows 2000 Desktop Admin
> >>>http://www.brainbench.com
> >>>
> >>>
> >>
> >
> >
>
- Next message: David H. Lipman: "Re: Earn $$$$$$$ FAST! Piece of cake. -xb:0"
- Previous message: Pegasus \(MVP\): "Re: boot problem"
- In reply to: Dan Seur: "Re: Performance Newsgroup"
- Next in thread: Dan Seur: "Re: Performance Newsgroup"
- Reply: Dan Seur: "Re: Performance Newsgroup"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
