Re: Partitions/Drives for SBS2003??

Tech-Archive recommends: Fix windows errors by optimizing your registry

From: root (postmaster_at_buchanangc.com)
Date: 04/25/04 

Date: Sun, 25 Apr 2004 14:24:42 -0700

"root" <postmaster@buchanangc.com> wrote in message
news:%23XvrUqkKEHA.2576@TK2MSFTNGP12.phx.gbl...
>
> "Jeff L" <newsgroupsremoveandunderscore_jeff@availabletech.net> wrote in
> message news:%23hj4E8jKEHA.1264@TK2MSFTNGP12.phx.gbl...
> > 1) Consider the impact of Indexing.
> >
> > Indices perform better if they are small. By default the more data you
put
> > in one drive the slower your index performance will be.
>
> I wouldn't expect that to be a significant issue with sufficient RAM and
> caching.
>
> > 2) Consider Fragmentation:
>
> Running defrag frequently(nightly) should handle that.
>
> > I actually add a third partition if the client is using software RAID
and
> > two drives.
>
> What is significant about SW vs HW RAID here?
>
> > A raid Zero Drive for Paging. I call it the P: drive.
>
> Two extra HDs just for a RAID 0 swap file hosting doesn't seem reasonable
> except in a large multiprocessor configuration. Running RAID 0 on
physical
> drives that hold things other than RAID 0 arrays kills some of the
> performance benefits of RAID 0.
>
> >I have
> > never done this with one drive (because I would not install SBS without
> > redundant drives)
>
> Right, but that is RAID 1(mirrored) and not the anti-redundant but very
> fast RAID 0.
>
> > but it could be a good way to isolate the space available
> > for paging and reduce the impact of Disk Fragmentation on system paging
> > performance.
>
> Setting the swap file size appropriately handles that issue.
>
> > The P: Drive is always 2.5 times the size of system memory and
> > I keep a smallest recommended Paging file on the OS drive. In practice
you
> > do not want to use the entire capacity of the P: drive for paging. You
> want
> > some free space.
>
> Unless one partitions the the RAID 1 array two drive set to locate the
"P:"
> drive physically first on the disk's surface, forcing the swap file to
> another partition will likely slow things down as the swap file will be
> located at a less optimal place on the HD.

Yes, see: "Do not place multiple paging files on different partitions on the
same physical disk drive."
http://support.microsoft.com/default.aspx?scid=kb;EN-US;308417

> Having the swap file on a SW
> RAID 1 array is actually suboptimal performance wise as a single drive
would
> perform slightly better for swap file hosting. The requirement for RAID
> 1(or other redundant array type) on the C: boot/system drive is a
> transcendent requirement.
>
> So it seems that I've talked myself back to your solution:
> Create a partition first on the each of the two disks' surfaces where the
> disk is fastest of an appropriate size for the half of that needed for the
> swap file. Join these two partitions into a SW RAID 0 array for the swap
> file. Having a RAID 0 array on the same physical drives as a RAID 1 array
> is not what one would normally do performance wise. HOWEVER putting a the
> swap file on a SW RAID 1 array is also suboptimal. I perceive the lesser
of
> the evils to be the separate RAID 0 array/partition for the swap file and
it
> saves a little HD space. That suboptimal RAID 0 swap file array located
> 'first' should perform better than the swap file on the RAID 1 C:.

Putting the page file and temp dir on a SW RAID 0 array located at the
physical beginning of a HD prior to the SW RAID 1 sysvol would provide
significant performance advantages especially if one is running Photoshop.

In many server environments the more important feature of having RAID 1 is
that the server just keeps on running with no loss of performance nor data
if a single drive fails. Adding a RAID 0 array to that guarantees an
instant STOP if either drive fails. So just say no to RAID 0 for any of the
purposes described above on a server.

Have enough RAM and leave the page file on C: with a RAM sized initial size
is the only reasonable thing to do.

If one is concerned about page file performance issues then watch it for a
day or three as described here:
http://www.theeldergeek.com/paging_file_performance_monitoring.htm

If the page file is getting hit hard enough to significantly affect
performance then add enough RAM until it isn't. If your configuration grows
to exceed the feasibility of this then get bigger iron and a good consultant
who understands HW and OS performance and tuning issues.

> Now will SBS2003 actually allow one to setup two HDs which host both a SW
> RAID 0 array and a SW RAID 1 array?
>
> If one has sufficient RAM to prevent significant swap thrashing then is
swap
> file I/O performance a significant issue to even consider? Are there
system
> operation points where a swap file of RAM size or bigger can be in regular
> use such that swap file I/O performance is significant but overall systems
> performance remains acceptable?? OR is the swap file in rather infrequent
> use and any entry to significant swap file usage state automatically means
> one's in thrash mode with degraded overall performance and therefore one
> needs more RAM?? AKA is swap I/O performance ever really a significant
> issue in responsive modest sized SBS2003 servers??
>
> What are the theoretics behind the selection of the swap file size and
> therefore the P: partition size?
>
> > I would like to see more support in the OS for non C: Paging files.
Namely
> > debuging infomartion should not be based on the size of your c: drive
> > pagefile.
>
> Does there have to be any(some minimum amount) of a swap file on the C:
> drive?
> OR can one configure SBS2003 to only have the P: swap file.
>
> > Jeff Loucks
> > Available Technology
> > Solutions for Professionals
> >
> > "root" <postmaster@buchanangc.com> wrote in message
> > news:OocWCkaKEHA.620@TK2MSFTNGP10.phx.gbl...
> > > Is there anywhere/anyone who has worked out good reasons of how to
> > partition
> > > a drive for SBS2003 for just a single disk(or RAID 1 set)?
> > >
> > > I'm not interested is "here's how I do it"-s. I'd like to see a real
> > > thoughtful summary of reasons/decisions for such.
> > >
> > > Fewer drives means the free space is pooled and therefore fewer is
> optimal
> > > from this standpoint. What other important considerations go into the
> > > partitioning issue?
>
>

Relevant Pages

  • Re: Problems with software RAID on SATA
    ... Connected to this are two 320GB drives ... >>which I want to turn into a RAID1 array. ... >>I'm almost certain it's a problem with initting the RAID arrays at boot. ...
    (Debian-User)
  • Re: RAID newbie...can I have several partitions on a RAID 1 array?
    ... You haven't expounded upon why you think you need raid. ... better backup device rather than buy 2 cheap RAID HBAs. ... RAID array then I would have to replace the mobo with the same one or at ... Lets say, for example, you buy 2 identical model drives, from ...
    (comp.sys.ibm.pc.hardware.storage)
  • Re: [PATCH 000 of 5] md: Introduction
    ... "why linux raid isn't Raid really, why it can be worse than plain disk") ... After this, the array ... error is in the filesystem, due to the complex layout of raid5. ... hundreds or 1000s of drives, you've quite high probability that some of them will fail sometimes, or will develop a bad sector etc). ...
    (Linux-Kernel)
  • How I built a 2.8TB RAID storage array
    ... My 2.8TB RAID 5 array is finally up and running. ... Nine 400GB PATA drives; eight for use, ... Two Highpoint RocketRAID 454 cards. ...
    (comp.os.linux.hardware)
  • Re: Partitions/Drives for SBS2003??
    ... for RAID 0 vs RAID 1. ... I set size the equal to the physical memory. ... The swap file on C: is either being used or it isn't. ... > Regarding the significance of SW vs. HW RAID and two drives. ...
    (microsoft.public.windows.server.sbs)