MSA Storage

MSA1500 - Active/Active Yet?

 
SAKET_5
Honored Contributor

MSA1500 - Active/Active Yet?

Hi Guys,

Just after confirmation - my understanding is that MSA1500 with the current FW V.4.98 does not support Active/Active controller configuration. Could someone please confirm this for me. I obviously do not get to play much with the MSA1000/1500s :(

Anyone knows when V5.x of the FW is expected and will it definitely bring the controllers to behave "Active/Active" for different LUNs?

Regards,
Saket.
45 REPLIES 45
Uwe Zessin
Honored Contributor

Re: MSA1500 - Active/Active Yet?

The next date for MSA Active/Active failover I got told was december 2005, but if I remember it was only for the MSA1500. It's my understanding that you can have different logical disks owned by different controllers, so it works somewhat similar to the EVA XL.
.
Luk Vandenbussche
Honored Contributor

Re: MSA1500 - Active/Active Yet?

Hi,

MSA1500 is not supported in active/active mode yet

see

http://h18000.www1.hp.com/products/quickspecs/11945_div/11945_div.pdf
John Kufrovich
Honored Contributor

Re: MSA1500 - Active/Active Yet?

Uwe is correct, December time frame. We have a few more kinks to work out. The 1500 will be initially supported. The 1000 in a few months after. Expect v6.xx FW.
SAKET_5
Honored Contributor

Re: MSA1500 - Active/Active Yet?

Thanks guys,

any particular reason as to why HP-UX is only supported in single controller configuration? Also, the limitation on the number of cluster nodes(Windows) that can connect to the MSA1000/1500cs - is it just a test/certification issue or one that is strictly technical in nature?
Vincent Fleming
Honored Contributor

Re: MSA1500 - Active/Active Yet?

The reason MSA is only single connect on HP-UX is because it's not active/active; I believe that the SecurePath (aka AutoPath) software was not modified to handle the MSA failover as-is is because: 1. VA7xxx exists, and 2. active/active has been scheduled to be here soon.

As for the Windows clustering, I'm not sure, but I'm sure it's a combination of both techical reasons, testing capacity, and the idea that if you have more than 16 nodes, you should be in an EVA instead of an MSA anyway - ie: the target market is too small to bother with the testing.

Regards,

Vince
No matter where you go, there you are.
Basil Vizgin
Honored Contributor

Re: MSA1500 - Active/Active Yet?

I hope, that MSA active/active will be different from VA realisation. In VA array all load via second controller goes to disk through first controller...
Vincent Fleming
Honored Contributor

Re: MSA1500 - Active/Active Yet?

Basil,

Other than cache-centric arrays, they all work that way... even from our competitors.

Regards,

Vince
No matter where you go, there you are.
Uwe Zessin
Honored Contributor

Re: MSA1500 - Active/Active Yet?

It depends on the type of VA:
true for the VA71x0, which has only one redundancy group (RG1).

The VA74x0 has two RGs. In a two-controller configuration, each controller is responsible for one RG - so you can divide your load through both controller's performance paths, unless you only make use of one LUN (leaving the other RG empty).
.
Basil Vizgin
Honored Contributor

Re: MSA1500 - Active/Active Yet?

Vincent, but why cannot we split cache for controller part and mirror controller part?
Yes, we loose half of the cache capacity, but wins in controller performance, which is weak (30K I/O's) in MSA.
Also can you point me to competitors models, which work in such "two pathes one controller" way, as you say? I didn't find such info nor in IBM, nor in EMC or even Adaptec documentation...
Thank you in advance!