Re: Booting from SAN or local?

I'll add this point also. Make sure you're using the correct (read:
STORPORT) drivers for your HBAs and ensure they're configured properly
(firmware, settings, etc.). Work with your storage vendor if you wish /
must.

Since the system disks are SAN-based, everything now depends on the
performance of your Fibre Channel zoning, HBAs, drivers, physical layer
(cabling), etc....

--
Ryan Sokolowski
MVP - Windows Server - Clustering
MCSE, CCNA, CCDA, BCFP
Avanade
http://www.Avanade.com

"A troubleshooter's best tool is the Event Viewer and understanding the
events and messages contained therein."

This posting is provided "AS IS" with no warranties, and confers no rights.

"Russ Kaufmann [MVP]" <russ@xxxxxxxxxxxxxxx> wrote in message
news:%23sh18S6vFHA.1988@xxxxxxxxxxxxxxxxxxxxxxx
> "Matt Povey" <Matt Povey@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
> news:384D6D50-CF92-48B8-A2F8-48C61AC8EA33@xxxxxxxxxxxxxxxx
>> 1) Rapid deployment \ blade environments - SAN boot volumes can be
>> re-assigned \ copied dynamically within the SAN fabric aiding fast
>> deployment
>> and recovery from localised hardware failures
>
> That is a nice bit of functionality, but does not result in performance
> increases unless the blades are using slow drives.
>
>> 2) DR. By synchronously replicating a boot volume across sites, you can
>> massively simplify your DR process though the use of cold standby server.
>
> An excellent point if you have geographically disperse SANs that are
> configured to synch. I agree that it is a great benefit.
>
>>> My question to the group is: What is the preferred method of booting.
>>> My
>>> opinion is it would be local with a mirrored disk seeing as booting up
>>> off
>>> the SAN is really slow.
>>
>> You really shouldn't be seeing any performance problems booting from SAN.
>> If
>> performance from your data volumes is adequate but the boot disk is poor
>> I'd
>> look at the BIOS currently installed on your HBA. Is your environment
>> qualified end to end by your SAN vendor?
>
> If you look at this regarding strict I/O performance and data transfer,
> which is capable of delivering uncached data quicker? I would counter that
> the SAN throughput, even using multipathing, will be slower than using
> local disk connecting using a local bus.
>
>> You really shouldn't have your page file on SAN attached disk although
>> this
>> is really for availability\reliability reasons rather than performance
>> (although it's a waste of SAN resources to have the page file there).
>
> Again, this is also a performance issue. Page file swaps will be much
> quicker across a local bus than across the SAN fabric.
>
>>> Does having the system disk on the SAN slow things down in a SQL and
>>> file
>>> sharing cluster environment?
>>
>> It certainly shouldn't do. Once booted, a Windows server with sufficient
>> RAM
>> doesn't really touch the boot disk. Even if it did, access to SAN
>> attached
>> disk should be as fast and low latency as DAS. Note that sync rep, SAN
>> based
>> mirrors , snaphots etc. can have an effect on performance.
>
> SANs utilize extremely high amounts of RAM for caching, so in most "read"
> cases, cache hits more than make up for the lesser transfer rates. In most
> cases, with file and SQL access, you will see other bottlenecks to high
> performance.
>
>>> Any ideas would be great as I think doing the disks local would have
>>> performance advantage but I would need to produce something to
>>> management.
>>
>> There's no reason to think that you'll get a useful performance advantage
>> from DAS (depending on the configuration of your SAN of course) and
>> you'll
>> likely lose significant functionality by switching to local disk.
>
> I disagree with Matt on this one. DAS is capable of much greater transfer
> rates than SAN attached drives.
>
>
> --
> Russ Kaufmann
>
> MVP - Windows Server - Clustering
> http://www.clusterhelp.com - Cluster Website
> http://msmvps.com/clusterhelp - New Blog
> http://spaces.msn.com/members/russkaufmann - Old Blog
>


.

Relevant Pages

  • Re: How to verify/fix High Disk Read Latencies in Exch2003 ?
    ... Exchange production servers are, the SAN is an EMC CX600. ... >>> current log file on disk and continues until data in the log buffers ... Comingling occurs whe two or more LUNs reside ...
    (microsoft.public.exchange.admin)
  • Ghost Drive Letters on 6 node Exchagne cluster
    ... test both QLogic 2300 HBAs with an IBM FasT SAN, and Emulex 9802 HBAs with a ... EMC Symmetric SAN. ... regarding failover, when a disk resource is failed from node 1 to node 2, in ...
    (microsoft.public.windows.server.clustering)
  • Re: HP EVA4000 / IBM DS4300 / EMC CX3-20/40
    ... disk array with the virtual raidsets on top. ... So, the system admin, and the DBAs had to create and manage lots of ... separate LUNs and *manually* manage the performance among them to ... applications on the SAN. ...
    (comp.arch.storage)
  • Re: How to verify/fix High Disk Read Latencies in Exch2003 ?
    ... Do they all map LUNs to your SAN? ... > wondering how the disk could be highly utilized if I have no users ... >> following formulats for the RAID type of each LUN: ...
    (microsoft.public.exchange.admin)
  • Re: LAN Free Backup TSM/SAN/ESS
    ... we have on our SAN is current using LAN-free backup, ... Intel) on the SAN have dual HBAs (using SDD for the disk) with access to ... wording I have read on the disclaimer about disk and tape sharing an HBA ...
    (comp.unix.aix)
Quantcast