@martius

- Posted using MobyPicture.com

- Posted using MobyPicture.com

- Posted using MobyPicture.com

So, what does this picture mean?

By default, volumes are spread across three members in one pool. So if you have six arrays in a pool, a different combination of three members will be used to handle different volumes.

There is a maximum of 16 nodes in a Hyper-V cluster which means every node in the cluster will get 6 persistent rervervations on a volume (96/16 = 6).

These 6 persistent reservations per node are then spread over a maximum of 3 members in a pool. We will then get 2 connections per member. If we apply redundancy this will result in 1 connection per NIC per Hyper-V cluster node.

If the above is correct then my previous post (http://www.delltechcenter.com/thread/4007957/Microsoft+Windows+2008+R2+CSV+and+Equallogic+SAN/ or http://www.martius.nl/?p=1896) on how to calculate the persistent reservations in wrong. This is because of the spread over max 3 members which actually means no mather how many Equallogics you are using(max. 16 in a pool) only 3 will be used to locate the volume.

The new calculation would then be:

Number of Equallogics   = 3 (max spread)
Number of NICs    = N
Number of MPIO    = M (default HIT = 2)
Number of Servers   = S

3 * N * M * S = 96

The numbers will then be:

3 * 2 * 1 * 16 = 96

Obvious question then is why should you use the DSM of Equallogic above the use of the standard Microsoft Multipath capabilities? The Equallogic DSM has a default MPIO per NIC of 2 while only one is available.

Another question is if this is the way Equallogic supports a 16 node Hyper-V cluster?

What are your thoughts? Please share them in getting this Equallogic – Persistent Reservation mystery clarified.

- Posted using MobyPicture.com

- Posted using MobyPicture.com

- Posted using MobyPicture.com

- Posted using MobyPicture.com

- Posted using MobyPicture.com