Re: Extremely Frustrated with EMC SAN
From: Lee Mah (lytmah_at_telusplanet.net)
Date: 03/23/04
- Previous message: Tim Sneddon: "Re: Server hang - compute bound"
- In reply to: David J. Dachtera: "Re: Extremely Frustrated with EMC SAN"
- Next in thread: Rob Young: "Re: Extremely Frustrated with EMC SAN"
- Reply: Rob Young: "Re: Extremely Frustrated with EMC SAN"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Tue, 23 Mar 2004 04:30:50 GMT
Last year, over several months, I ran hundreds of extensive test jobs
against
ten shadow sets comprised of EMC-EMC, EMC-HSJ, and HSJ-HSJ disk
members in a five-node cluster (four ES45's, one DS10, three separate sites)
using HBVS. The cluster was at VMS 7.3 plus patches.
I encountered no problems or any differences in behaviour or
functionality of
the three different types of shadows, other than one which I'll discuss
later.
Of course, the EMC-EMC shadow sets showed better results in terms of
throughput. DCL commands like ANALYZE/DISK, CONVERT, COPY...
worked normally. As individual DGA disks, the EMC disks behaved no
differently from the Storageworks HSJ DUA disks.
The positive aspect of going to EMC is that my life would be much easier
in regards to disk maintenance and tape backups. I would no longer have
to look after these onerous duties.
EMC provided me with two customer reference contacts. One was running
SRDF, not HBVS. The second contact still hasn't replied to my list of
questions,
so I don't know whether they're using HBVS or not.
So, with these two sites, plus David's count of one, plus the site whose
VMS administrator posted a complaint two years ago and got burned
for doing so, we have four known VMS sites using EMC disks.
Now, the problem with EMC shadow sets. If a node crashes, each shadow
set comprised of EMC-EMC members went into mount verification and
took over 12 hours to do full shadow merges. The EMC-HSJ shadows took
from 1.5 to 5 hours to full merge. The HSJ-HSJ shadows went into minimerge
and recovered to normalcy within six seconds. Of course, the EMC-EMC
shadow problem also applies to HP SAN disks. HP is supposed to have
a retro-fit available for this problem some time in the future.
Also, a note of warning if you want to do shadowing of dissimilar-size
members
under VMS 7.3-2. There is a restriction in the mounting process. I
wouldn't
recommend going for dissimilar shadowing until this restriction is resolved.
David J. Dachtera wrote:
>PhilThayer wrote:
>
>
>>I have been fighting to get this *?&%$*& EMC SAN working with the
>>AlphaServer ES40 and it's pushing me to pull my hair out.....
>>
>>I have gotten the ES40 to a point where I can see the disks from the
>>console prompt when I do a SHOW DEVICE. I have installed all the
>>current patches on the ES40 for VMS V7.3. I can sit on the EMC
>>monitoring utility and see the transmit/receive counts increment when
>>I do SYSMAN IO AUTOCONFIGURE.
>>
>>Everything seems to be setup correctly so, WHY DO I NOT HAVE ANY
>>DRIVES GETTING CONFIGURED IN VMS?
>>
>>I wish this company had just bought a StorageWorks SAN instead of this
>>*&%$#?%$%$ EMC SAN. Unfortunately the decision on this was made
>>before I arrived here.
>>
>>
>
>Well, in a recent conference call with hp, I was told - more or less
>officially - that EMC is not certified with/by OpenVMS and likely never
>will be. EMC says they support OpenVMS, but then Dubya said... well,
>let's keep politics out of it, I think the point is made.
>
>That said, I am aware of at least one VMS site running on EMC (it's an
>IDX machine).
>
>I believe V7.3-1 with patches may be a minimum, but I'm probably wrong
>there.
>
>
>
- Previous message: Tim Sneddon: "Re: Server hang - compute bound"
- In reply to: David J. Dachtera: "Re: Extremely Frustrated with EMC SAN"
- Next in thread: Rob Young: "Re: Extremely Frustrated with EMC SAN"
- Reply: Rob Young: "Re: Extremely Frustrated with EMC SAN"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|