AF:
NF:0
PS:10
SRH:1
SFN:
DSR:
MID:<20081216103136.04edc8b5@ripper.onstor.net>
CFG:
PT:0
S:andy.sharp@onstor.com
RQ:
SSV:exch1.onstor.net
NSV:
SSH:
R:<michael.baugh@onstor.com>,<brian.stark@onstor.com>,<narayan.venkat@onstor.com>,<patrick.haverty@onstor.com>,<dl-LeopardCoreTeam@onstor.com>
MAID:1
X-Sylpheed-Privacy-System:
X-Sylpheed-Sign:0
SCF:#mh/Mailbox/sent
RMID:#imap/andys@onstor.net@exch1.onstor.net/INBOX	0	2779531E7C760D4491C96305019FEEB51762E4F090@exch1.onstor.net
X-Sylpheed-End-Special-Headers: 1
Date: Tue, 16 Dec 2008 10:31:48 -0800
From: Andrew Sharp <andy.sharp@onstor.com>
To: Michael Baugh <michael.baugh@onstor.com>
Cc: Brian Stark <brian.stark@onstor.com>, Narayan Venkat
 <narayan.venkat@onstor.com>, Patrick Haverty <patrick.haverty@onstor.com>,
 dl-Leopard Core Team <dl-LeopardCoreTeam@onstor.com>
Subject: Re: SAS HBA - Which one?
Message-ID: <20081216103148.5c00d7bf@ripper.onstor.net>
In-Reply-To: <2779531E7C760D4491C96305019FEEB51762E4F090@exch1.onstor.net>
References: <2779531E7C760D4491C96305019FEEB51762E4F081@exch1.onstor.net>
	<2779531E7C760D4491C96305019FEEB51762E4F086@exch1.onstor.net>
	<2779531E7C760D4491C96305019FEEB51762E4F090@exch1.onstor.net>
Organization: Onstor
X-Mailer: Sylpheed-Claws 2.6.0 (GTK+ 2.8.20; x86_64-pc-linux-gnu)
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit

Your completely welcome!

On Tue, 16 Dec 2008 10:26:43 -0800 Michael Baugh
<michael.baugh@onstor.com> wrote:

> Understood, thanks Brain for the clarification
> 
>  
> 
> 
> Mike Baugh
> Director North America Client Relations
> 
> ONStor, Inc.
> office: 408-376-3160
> mobile: 408-431-6803
> 
> 
> Michael.Baugh@onstor.com
> http://www.onstor.com
> 
> 
> -----Original Message-----
> From: Brian Stark 
> Sent: Tuesday, December 16, 2008 1:19 PM
> To: Michael Baugh; Narayan Venkat; Andy Sharp; Patrick Haverty
> Cc: dl-Leopard Core Team
> Subject: RE: SAS HBA - Which one?
> 
> No, not technically.  Every component would replicated across the
> cluster and redundancy is built into the cluster, so there is no
> SPOF.  
> 
> For example, the HBA in one cluster partner would provide redundancy
> in the event that the HBA in the other cluster partner failed.  This
> is the same scenario for the other chips in each system, like the
> processors, memory, Northbridge, Southbridge, etc.
> 
> 
> Brian
> 
> 
> -----Original Message-----
> From: Michael Baugh 
> Sent: Tuesday, December 16, 2008 10:16 AM
> To: Brian Stark; Narayan Venkat; Andy Sharp; Patrick Haverty
> Cc: dl-Leopard Core Team
> Subject: RE: SAS HBA - Which one?
> 
> Just some I am clear, the HBA's are still a SPOF?
> 
>  
> 
> 
> Mike Baugh
> Director North America Client Relations
> 
> ONStor, Inc.
> office: 408-376-3160
> mobile: 408-431-6803
> 
> 
> Michael.Baugh@onstor.com
> http://www.onstor.com
> 
> -----Original Message-----
> From: Brian Stark 
> Sent: Tuesday, December 16, 2008 1:10 PM
> To: Narayan Venkat; Andy Sharp; Patrick Haverty
> Cc: dl-Leopard Core Team
> Subject: RE: SAS HBA - Which one?
> 
> Spoken like a true hardware designer, chips don't typically fail.  I
> think we protect against cable failure and also rely on the dual JBOD
> controllers for redundancy.  Since each node has 2 connections to the
> JBOD in an HA config, we have redundant paths and controllers to
> provide HA.
> 
> 
> 
> -----Original Message-----
> From: Narayan Venkat 
> Sent: Tuesday, December 16, 2008 9:58 AM
> To: Brian Stark; Andy Sharp; Patrick Haverty
> Cc: dl-Leopard Core Team
> Subject: RE: SAS HBA - Which one?
> 
> Fair enough!  I agree with your statement if we believe that HBA
> redundancy is not required.
> 
> Narayan Venkat
> Vice President, Marketing
> ONStor Inc. (www.onstor.com)
> Tel: (408) 963-2404
> Cell: (408) 221-4297
> 
> -----Original Message-----
> From: Brian Stark 
> Sent: Tuesday, December 16, 2008 9:54 AM
> To: Narayan Venkat; Andy Sharp; Patrick Haverty
> Cc: dl-Leopard Core Team
> Subject: RE: SAS HBA - Which one?
> 
> Narayan,
> 
> I'm not sure we need 2 HBAs per system.  This would protect against
> the controller as a single point of failure, but we always have the
> second node for failover.
> 
> With a single HBA and 8 ports per (as with the LSI), we get
> redundancy in cabling from each controller to the JBOD shelves.
> Assuming the JBOD has dual controllers, then each HA pair would have
> 2 cables connected to the JBOD, which provides further redundancy.
> 
> With this in mind, I would not advocate having 2 HBAs per system.
> 
> 
> Brian
> 
> 
> -----Original Message-----
> From: Narayan Venkat 
> Sent: Tuesday, December 16, 2008 9:34 AM
> To: Brian Stark; Andy Sharp; Patrick Haverty
> Cc: dl-Leopard Core Team
> Subject: RE: SAS HBA - Which one?
> 
> Agree with Pat that we want multiple paths.  For HA configurations we
> will need 2 HBAs per system for no SPOF.  
> 
> If everybody is okay with 3801E, then I will enquire about OEM
> pricing from LSI.
> 
> Thanks.
> 
> Narayan Venkat
> Vice President, Marketing
> ONStor Inc. (www.onstor.com)
> Tel: (408) 963-2404
> Cell: (408) 221-4297
> -----Original Message-----
> From: Brian Stark 
> Sent: Tuesday, December 16, 2008 9:28 AM
> To: Andy Sharp; Patrick Haverty
> Cc: Narayan Venkat; dl-Leopard Core Team
> Subject: Re: SAS HBA - Which one?
> 
> I agree with Pat.  We should leverage what we've tested, and we know
> it will support all configs that we plan to offer.  I'd rather not
> get into multiple part numbers and configs depending on what the
> customer orders.  Plus, we should get very favorable pricing direct
> from LSI.
> 
> 
> Brian
> 
> 
> 
> 
> ----- Original Message -----
> From: Andy Sharp
> To: Patrick Haverty
> Cc: Narayan Venkat; dl-Leopard Core Team
> Sent: Tue Dec 16 09:14:36 2008
> Subject: Re: SAS HBA - Which one?
> 
> On Mon, 15 Dec 2008 20:15:05 -0800 Patrick Haverty
> <patrick.haverty@onstor.com> wrote:
> 
> > I think if we are going to offer dual paths, either redundant
> > connection or connection to another appliance's array, then the only
> > choice of the three would be the LSI 3801E.
> 
> Why?  What's wrong with the adaptec?
> 
> 
> > ________________________________
> > From: Narayan Venkat
> > Sent: Monday, December 15, 2008 7:13 PM
> > To: dl-Leopard Core Team
> > Subject: SAS HBA - Which one?
> > 
> > Leopard team,
> > 
> > We have two or possibly 3 options for the SAS HBAs.  Our choice of
> > HBAs will come down to robustness and price.
> > 
> > Since our initial focus is the entry-level price band it would be
> > sufficient for a SAS HBA to support up to 48 drives.  Most SAS HBAs
> > that we find in the market today support up to 122/128 drives or
> > even up to 244/256 drives.
> > 
> > The following HBAs work well with OpenSolaris.
> > 
> > a)       LSI 3801E (this is what we have been testing with.  It has
> > 8 external ports using SFF8088 connectors and supports up to 244
> > drives/end points) b)       LSI 3442E-R (It has 4 internal and 4
> > external ports.  It uses SFF8470 connectors and supports up to a 122
> > drives/end points) c)       Adaptec 2045 (It has 4 external ports.
> > It uses SFF8088 connectors and supports up to 128 drives/end points)
> > 
> > All of these HBA are known to be robust and have demonstrated good
> > adoption in the market.  Note that the LSI 3801E and LSI 3442E-R are
> > basically the same HBA (based on LSI 1068 chipset)
> > 
> > Given that we have to be competitive with the NTAP StoreVault line
> > of products our COGS have to be in line for us (not surprisingly!)
> > to make a decent margin.
> > 
> > I know that the Adaptec 2045 MSRP is about $250, LSI 3801E about
> > $365 and LSI 3442E-R is about $269.
> > 
> > I haven't enquired LSI or Adaptec about OEM pricing.  Do I need to
> > that or will the operations team follow up?  Kindly let me know.  I
> > am fine either way.
> > 
> > Thanks.
> > 
> > Narayan Venkat
> > Vice President, Marketing
> > ONStor Inc. (www.onstor.com<http://www.onstor.com>)
> > Tel: (408) 963-2404
> > Cell: (408) 221-4297
> > 
