HPE MSA 2052 MPIO

[ccane]

Hi guys,

I have connected directly (iscsi, dac cables) 2 servers with vmware 6.5 to my storage, server1 to ports A1 B1 and server2 to ports A2 B2, on array i have one pool on a Storage A.

In vmware i have configured vSwitch with 2 vmkernel and iscsi software adapter, path policy is round robin and I see from each server 2 path but one of them is Active and the other is Active (I/O).

Failover is working properly, but I was thinking that this array is Active/Active, and with mpio host should use all avaible path to array, but from my testing it using only one vmnic (one path) at the same time.

I assumed that I would be able to utilize 2 10Gbps connections from every host, by connecting them to 2 controllers A and B, and both controllers works simultaneously.

How to connect 2 servers to array to use Active/Active mode and have link with more than 10Gbps?

Any suggestions would be great

Thanks

[Shawn_K]

Hello,

Checking SPOCK for the correct configuration for an MSA 2050 iSCSI array with ESXi 6.5, direct connect is not supported by the iSCSI initiator or VMware.

https://h20272.www2.hpe.com/spock2/Cont/ConfigsetView/index/99256

Troubleshooting an unsupported configuration becomes problematic since HPE has not tested nor qualified these types of configurations.

Cheers,
Shawn

I work for Hewlett Packard Enterprise. The comments in this post are my own and do not represent an official reply from HPE. No warranty or guarantees of any kind are expressed in my reply.

[ccane]

thank you for responding

Can you confirm that my assumptions was right, that in normal circumstances with transceiver modules transfers from one host to storage would be on two controlers by two path at the same time?

[Shawn_K]

Hello,

In a supported configuration having server_1 connected to controller host ports A1 and B1 while server_2 could connect to controller host ports A2 and B2 would offer the ability for failover.

How your configuration is reporting might not lead to correct failover should it be needed.

Cheers,
Shawn

I work for Hewlett Packard Enterprise. The comments in this post are my own and do not represent an official reply from HPE. No warranty or guarantees of any kind are expressed in my reply.

[SUBHAJIT KHANBARMAN_1 · Accepted solution]

The storage system uses Unified LUN Presentation (ULP), which can expose all LUNs through all host ports on both controllers. The interconnect information is managed in the controller firmware. ULP appears to the host as an active-active storage system where the host can choose any available path to access a LUN regardless of disk group ownership. When ULP is in use, the controllers' operating/redundancy mode is shown as Active-Active ULP. ULP uses the T10 Technical Committee of INCITS Asymmetric Logical Unit Access (ALUA) extensions, in SPC-3, to negotiate paths with aware host systems. Unaware host systems see all paths as being equal.

 In a typical round-robin configuration, MPIO will load balance between both ports on the server and across all optimized paths to a Volume, however it will never drive I/O down an un-optimized path whilst even a single optimized path remains active.  This preferential behavior is an intentional product of the how the Unified LUN Presentation, or ULP feature, takes advantage of Asymmetrical Logical Unit Access, or ALUA extensions, to support optimal path selection to the controller which has ownership of the Volume, and therefore the underlying Pool and Disk Groups.

 ULP can make a volume available via any selected array host port so to improve availability and performance, but it is the ALUA protocol which ensures that I/O need not be brokered by a partner controller which would otherwise increase latency.  This happens on a per volume level rather than at the array level, and it is for this reason that the array should be configured in a balanced manner, i,e, that I/O is distributed across two equally or at least similarly provisioned pools, and therefore multiple volumes.

In your case as you have two paths so one will be Optimized path and that will show as  Active (I/O) and another path will be Active only.

Hope this helps!
Regards
Subhajit

I am an HPE employee

If you feel this was helpful please click the KUDOS! thumb below!

***********************************************************************************

[ccane]

If i understand correctly, when I have one volume on storage A it's optimized for controller A and it is presented as Active (I/O), second path to controller B is not optimized that why is only Active. To achieve on a host two path with Active (I/O) I need to make another volume on storage B witch wil be optimized for controller B, is that correct?

[SUBHAJIT KHANBARMAN_1]

Yes your understanding correct

Hope this helps!
Regards
Subhajit

I am an HPE employee

If you feel this was helpful please click the KUDOS! thumb below!

***********************************************************************************

[ccane]

thanks for the clarification