topic Re: Implement SAN redundancy MSA2040 in MSA Storage

Implement SAN redundancy MSA2040

SSKA — Tue, 17 Jul 2018 12:04:47 GMT

Hi all,

We have two MSA 2040 SAN with the information mentioned below:

Vendor : HP MSA 2040 SAN

Mode : SPS-CHASSIS

Number of Disks : 6 (4 disks in RAID 10 and 2 disks are spared)

Midplane Serial Number :

Part Number : 639410-001

MIdplane Type : 2U24-6Gv2

Enclosure Power : 125.28

This dual controller SAN is connected to 2 Oracle DBs server via FC ports to controller A and B which are in Active-Active ULP Mode for each server. We are supposed to load up second SAN which is entirely same as the first one to implement SAN redundancy.

Based on our investigation the best design is using two SAN switches to connect two ports from each controller module to the first switch to facilitate I/O traffic, and connect two ports from each controller module to the second switch to facilitate replication to get full redundancy and avoid the potential single point of failure inherent to using a single switch. But I want to know what if we eliminate the switch and just connect two SANs together? Please consider that our second SAN storage does not have any configuration, we want to make sure adding it to cluster will not affect our production environment.

Re: Implement SAN redundancy MSA2040

SUBHAJIT KHANBARMAN_1 — Tue, 17 Jul 2018 12:28:03 GMT

If you want to eliminate SAN Switch and implement replication then this is not supported as per HPE official document. Please refer the below document (page no 13),

https://h20195.www2.hpe.com/v2/getpdf.aspx/4aa1-0977enw.pdf

Hope this helps!
Regards
Subhajit

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

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

Re: Implement SAN redundancy MSA2040

SSKA — Wed, 18 Jul 2018 04:26:41 GMT

Thanks for your reply,

Do you know what happend if direct connection is used between two systems?

Regards,

Re: Implement SAN redundancy MSA2040

SUBHAJIT KHANBARMAN_1 — Wed, 18 Jul 2018 11:16:57 GMT

I am not sure........we follow what is officially supported because if we implement something which is not supported by HPE then any further issue arise or any furture issue we will not get support from HPE.

I would request you to mark my previous reply as solution so that whoever following this forum will get proper answer

Hope this helps!
Regards
Subhajit

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

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

Re: Implement SAN redundancy MSA2040

Shawn_K — Wed, 18 Jul 2018 16:40:44 GMT

Hello,

If you eleiminate the switch and direct connect the servers to the arrays you would not be able to use the MSA arrays replication feature. This would elminate the redundancy that replication would offer.

By havign the two arrays connected via switches you remove the possibility of a single point of failure taking down your SAN.

I would also suggest you consider future needs when you configure your raid sets. With only 6 disks in a RAID10 configuration you are likely going to be hitting space boundries soon. Purchasing an additional 6 disks and creating a second Pool B would load balance your IO across the two controllers and provide better performance.

Cheers,

Shawn

Re: Implement SAN redundancy MSA2040

SSKA — Sat, 21 Jul 2018 05:41:12 GMT

Hello,

Thanks a lot for complete advice.

Regards,

Re: Implement SAN redundancy MSA2040

SSKA — Sat, 21 Jul 2018 05:50:20 GMT

Hi,

just to make sure I got your mean by right, you said we can not use replication if connect the servers directly to the SAN? Servers are already connected to the array via FC ports. To add new array and implement redundancy between two MSAs we should connect servers via switch?

Thanks,

Re: Implement SAN redundancy MSA2040

Shawn_K — Sat, 21 Jul 2018 14:23:18 GMT

Hi,

To use replication you will need to have the servers connect to the switch(s) and then connect both MSA arrays to the switch(s).

Look at the following MSA 2040 User Guide starting with page 38. There is a diagram for a single switch and if you have two switches.

https://support.hpe.com/hpsc/doc/public/display?docId=c03792322

Be sure to correctly zone the switch. Your host IO traffic should not be on the same zone as your replication IO traffic. You do not want to have the switches be "wide open" so that every system can see every other system on the network. Correctly zoning the switches will reduce IO collisions and network congestion.

Good luck,

Shawn

Re: Implement SAN redundancy MSA2040

SSKA — Sun, 22 Jul 2018 05:01:50 GMT

Hi,

I got it, thanks a lot.