Thanks for the feedback! 

e-

-----Original Message-----
From: Jon Hill [mailto:[EMAIL PROTECTED]] 
Sent: Sunday, December 22, 2002 7:35 AM
To: Exchange Discussions
Subject: RE: the IBM Shark

I'm way late to this thread but have some experience with HP's XP-512 that
might be relevant.

As you probably know, the XP series and HDS 99xx series are basically the
same machine - HP & HDS develop the hardware together and put their own
software on board (actually, HP used to work with HDS software but over time
developed their own).  I can't recall which HDS array maps to the XP-512 but
I'm sure your reseller can tell you.

Anyway, we have a pair of XP-512s that we're using for DR.  One array goes
in our Manhattan office; the other in a DR site in Exton, Pennsylvania.  The
arrays are connected via a 40Mbps IP WAN provisioned by Yipes.

On the local array, we have an HP 3000, two HP 9000s (a D and an L-1000) and
22 Intel-based servers (mostly NetServer LP2000r, plus one LPr and one
Compaq DL-380).  All Intel servers and the L-1000 boot from the SAN.  Each
Intel box uses a pair of Emulex LP8000 HBAs to communicate with the array.
HP's multipathing software, Auto Path, manages path failover.  The remote
array is almost identical, with the major difference being that most of the
Intel servers are powered down.  FWIW, we replicate track-level changes to
the remote array via HP's Continuous Access Extension software (HDS has
something similar):  think EMC's SRDF but asynchronous.  

All comm between servers and array is via FC over Brocade 2800s (6 switches
per site operating as two 3-switch SANs), except for the HP 3000, which
can't handle FC.  That server gets FC-SCSI routers to communicate with the
Brocades.  All FC traffic is switched FC, except again for the HP 3000 which
is switched FCAL (there's a proper term for this, but essentially the
Brocade ports are operating in FCAL mode).

One of the servers runs E2K on W2K.  I won't say that our experience has
been 100% perfect, but the problems have had nothing to do with the arrays
themselves.  We are using an HP product called XP Cluster Extension for MSCS
that tricks MS Cluster Services on two nodes (one here, one in Exton) into
thinking that they're sharing a hard disk.  Some of the E2K problems have
arisen due to comm interruptions between the two arrays (some of which were
our fault), and some were probably simply the result of ignoring Ed and
running E2K over a cluster.  

However, the XP-512 with Emulex HBAs and Brocade switches has handled our
servers extraordinarily well.  The day after the switch people started
asking me what we'd done to improve performance so dramatically.

I didn't do much research into the Shark (it might have been limited to
ESCON at the time I did my research), but we had two reasons to pick the XP
over EMC's Symmetrix:  Continuous Access can handle asynchronous replication
over IP, while SRDF is synchronous-only; and the XP is better built.  When
EMC reps tell you how reliable their arrays are, ask them what happens when
you pull a live cache card out.  The XP can handle it because its cache
cards are arranged in a RAID5-like configuration; the Symm crashes.  We
chose the XP over the equivalent HDS because we felt that HP had a bigger
support presence in the U.S., we preferred their software to HDS's, and we
already had a lot of HP hardware in our shop.  


-----Original Message-----
From: Hansen, Eric [mailto:[EMAIL PROTECTED]]
Sent: Monday, December 16, 2002 10:02 AM
To: Exchange Discussions
Subject: RE: the IBM Shark


They came in under the 9900v series machines, and only slightly under the 15
year old 9900 series and WAY below the emc CX6000.  We aren't looking at the
Compaq San cause we have one that runs on our legacy mainframe(open vms) and
the admins of that system hate it.  The price wars have been pretty
extensive, but its nice to be able to turn vendors on each other to get the
best price for the company.  Ive been out to lunch to more sales meetings
this last month then I was all last year.  :)

e-

-----Original Message-----
From: Andy Haigh [mailto:[EMAIL PROTECTED]] 
Sent: Friday, December 13, 2002 2:02 PM
To: Exchange Discussions
Subject: RE: the IBM Shark

Be wary of the upfront pricing. Currently IBM is offering the kit at
very low prices to win back market share from the likes of EMC, HDS and
Compaq. You must nail them down to the costs of adding new storage
arrays, the cost of additional drives and any tiering of licence prices
as your storage grows, also any services you will require afterwards.

Again, if you haven't already done so, I would advise to have a look the
HDS 9570V or even the EMC CX600. If you are currently looking at the
9900V these units will give them a run for their money and be cheaper.

Andy

-----Original Message-----
From: Hansen, Eric [mailto:[EMAIL PROTECTED]] 
Sent: Saturday, 14 December 2002 1:35 AM
To: Exchange Discussions
Subject: RE: the IBM Shark


"Other technologies" being what?, cause the only other one I have seen
in the backstore of a SAN is Fiber Channel Arbitrated Loop, SSA is not
arbitrated.  TO me this is a pro for multipathing across single ports.
Although I have heard that IBM plans to go to FCAL when its throughput
gets high enough that the arbitration no longer is a issue.


So far it looks like it may be IBM cause of price, they are really
coming in strong.  And all our figures indicate that the performance
between IBM and HDS is a wash.

-----Original Message-----
From: Exchange (Swynk) [mailto:[EMAIL PROTECTED]] 
Sent: Friday, December 13, 2002 7:25 AM
To: Exchange Discussions
Subject: RE: the IBM Shark

But token ring does blow chunks.  :)

I'm glad to hear you've had a successful implementation; ours was far
from successful and has left a bad taste in our collective mouth about
this product.  I think the problem, more than anything, was that a
number of different IBM people from different teams and areas of
expertise came out here, and no one was able to provide a solution to
our problems.  The mantra of "it should work, I don't know why it
doesn't" lost its humor after about the 27th time.  Perhaps we just got
a bad batch of guys, who knows.  

Cost aside, the maximum potential throughput of SSA does not come close
to that of some other technologies ... And while "blow chunks" may not
have been the best choice of words, I will stand by my lowly opinion of
the technology.  (Take that with a grain of salt of course, for my
opinions rarely hold any value! :)

Jon

> -----Original Message-----
> From: Thompson, Elizabeth [mailto:[EMAIL PROTECTED]]
> Posted At: Thursday, December 12, 2002 4:02 PM
> Posted To: Exchange (Swynk)
> Conversation: the IBM Shark
> Subject: RE: the IBM Shark
> 
> 
> Hi this is Ben Thompson.  I am the Sr. Network Engineer at
> the college that
> Liz is the Exchange Administrator.  (I just also happen to dual as her
> husband.)  We have had a Shark for several years now.  The 
> performance is
> great.  You may want to look at NT related buffering issues 
> when dealing
> with its performance, we did not have to modify anything to 
> get it working
> though.  We are a Compaq shop and the Compaq SANS just was 
> not up to spec
> for Novell, NT/2000, and AIX/Linux.  (Novell is not even 
> supported with NT
> on the same unit at the same time.)  Exchange has been 
> working wonderfully,
> as well as Novell, Linux, DOS and 2000. (We are Exchange 5.5) 
>  We have been
> booting from the SANS since day one, something IBM does not 
> like but will
> support.  We also run our production Oracle database on the 
> Shark.  If that
> performance lagged for any reason I would have 3 major sites 
> and 11 remote
> sites down my throat in a heartbeat.  If anyone would care to 
> look at our
> installation, feel free to e-mail me at [EMAIL PROTECTED] 
> and I can set
> it up.  
> 
> As to SSA blowing chunks, most people thought the same of
> Token-Ring.  Just
> because something is more complicated to understand and 
> expensive does not
> make the architecture "blow chunks", just cost prohibitive.
> 
> Benjamin N. Thompson
> Senior Network Engineer/Manager
> CCBC - Catonsville Campus
> 
> -----Original Message-----
> From: Exchange (Swynk) [mailto:[EMAIL PROTECTED]]
> Sent: Tuesday, December 10, 2002 7:58 AM
> To: Exchange Discussions
> Subject: RE: the IBM Shark
> 
> 
> It's definitely related to the architecture -- SSA blows
> chunks.  We've
> had several IBM guys out here to apply their "expertise" 
> (read: blindly
> poke around) ..... plus, paying $30k x 2 for just a couple 
> hundred gb is
> highway robbery!
> 
> Can you tell I hate IBM?  :)
> 
> > -----Original Message-----
> > From: Roger Seielstad [mailto:[EMAIL PROTECTED]]
> > Posted At: Tuesday, December 10, 2002 6:58 AM
> > Posted To: Exchange (Swynk)
> > Conversation: the IBM Shark
> > Subject: RE: the IBM Shark
> > 
> > 
> > While I know the Compaq stuff is some of the best out there,
> > I'd be very
> > interested to see if the performance issues you're seeing 
> aren't more
> > directly related to poor drive/array/LUN partitioning rather
> > than issues
> > with specific architecture - after all, once it leaves the 
> > HBA, FC is FC.
> > 
> > ------------------------------------------------------
> > Roger D. Seielstad - MCSE
> > Sr. Systems Administrator
> > Inovis - Formerly Harbinger and Extricity
> > Atlanta, GA
> > 
> > 
> > > -----Original Message-----
> > > From: Exchange (Swynk) [mailto:[EMAIL PROTECTED]]
> > > Sent: Monday, December 09, 2002 1:11 PM
> > > To: Exchange Discussions
> > > Subject: RE: the IBM Shark
> > > 
> > > 
> > > We have a Shark here and found that it is CRAP when it
> comes to I/O
> > > intensive Win32 applications.  Someone here got the bright
> > > idea to have
> > > an enterprise-wide SAN solution, instead of looking at it from the
> > > perspective of how each platform actually works .... the 
> Shark works
> > > great for legacy (i.e. IBM) systems, and works marginally
> > well for NT
> > > file servers, but try sticking a large SQL database on
> > there and watch
> > > what happens.  Of all the SANs out there (at least 18
> > months ago when
> > > ours was purchased), the Shark was one of the most expensive,
> > > and one of
> > > the slowest.  It may not be the same with newer Sharks, but 
> > ours is a
> > > slow-as-hell drive technology that choked whenever we tested SQL 
> > > databases and Exchange 5.5 on it.
> > > 
> > > We have found that Compaq's SAN solution works well for our
> > > environment
> > > -- it's almost half the price of comparable storage on the 
> > Shark, and
> > > much much faster.  Since we're an all-Compaq shop for our
> > > Win32 systems,
> > > that's what we're moving to now.
> > > 
> > > 
> > > > -----Original Message-----
> > > > From: Hansen, Eric [mailto:[EMAIL PROTECTED]]
> > > > Posted At: Friday, December 06, 2002 10:29 AM
> > > > Posted To: Exchange (Swynk)
> > > > Conversation: the IBM Shark
> > > > Subject: OT: the IBM Shark
> > > > 
> > > > 
> > > > Is anyone here happen to be running a IBM shark or possibly a
> > > > Hitachi 9900
> > > > series SAN?  We are looking at both of these and I have heard 
> > > > rumors that
> > > > the shark has a performance boundary of 3.36 TB.  Just curious.
> > > > 
> > > > e-
> 
> 
> _________________________________________________________________
> List posting FAQ:       http://www.swinc.com/resource/exch_faq.htm
> Archives:               http://www.swynk.com/sitesearch/search.asp
> To unsubscribe:         mailto:[EMAIL PROTECTED]
> Exchange List admin:    [EMAIL PROTECTED]
> 

_________________________________________________________________
List posting FAQ:       http://www.swinc.com/resource/exch_faq.htm
Archives:               http://www.swynk.com/sitesearch/search.asp
To unsubscribe:         mailto:[EMAIL PROTECTED]
Exchange List admin:    [EMAIL PROTECTED]

_________________________________________________________________
List posting FAQ:       http://www.swinc.com/resource/exch_faq.htm
Archives:               http://www.swynk.com/sitesearch/search.asp
To unsubscribe:         mailto:[EMAIL PROTECTED]
Exchange List admin:    [EMAIL PROTECTED]

_________________________________________________________________
List posting FAQ:       http://www.swinc.com/resource/exch_faq.htm
Archives:               http://www.swynk.com/sitesearch/search.asp
To unsubscribe:         mailto:[EMAIL PROTECTED]
Exchange List admin:    [EMAIL PROTECTED]

_________________________________________________________________
List posting FAQ:       http://www.swinc.com/resource/exch_faq.htm
Archives:               http://www.swynk.com/sitesearch/search.asp
To unsubscribe:         mailto:[EMAIL PROTECTED]
Exchange List admin:    [EMAIL PROTECTED]

_________________________________________________________________
List posting FAQ:       http://www.swinc.com/resource/exch_faq.htm
Archives:               http://www.swynk.com/sitesearch/search.asp
To unsubscribe:         mailto:[EMAIL PROTECTED]
Exchange List admin:    [EMAIL PROTECTED]

_________________________________________________________________
List posting FAQ:       http://www.swinc.com/resource/exch_faq.htm
Archives:               http://www.swynk.com/sitesearch/search.asp
To unsubscribe:         mailto:[EMAIL PROTECTED]
Exchange List admin:    [EMAIL PROTECTED]

Reply via email to