MSA 2050 iSCSI Controller Links

[QFlux]

I'm hoping someone can just confirm something for me. We have a MSA 2050 with twin controllers and 8 links in iSCSI setup. These are connected to a set of switches then onto a set of VMWare Hosts. For reasons too long to go into I need to disconnect and re run one of the links.

As I understand it this should not distrupt access from the hosts to the MSA's storage although will create a reduction in avalible bandwidth. Is this correct or have I miss understood?

Thanks

[Shawn_K · Accepted solution]

Hello,

The short simple answer is you can remove a link from the MSA array.  The array is designed to handle this type of a disruption from the array's side and will process all host IO correctly.

You need to be aware of what this type of disruption will do to the hosts. If your hosts do not have a redundant path configured via a multipathing option then the host(s) will loose access. For example, if each configured host is using only a single path or has only a single path from a switch configured to mulitple hosts, then all hosts connected via that one link will loose access.

Be aware of your hosts configuration first prior to removing the link. Additionally, check all multipathing configurations after you recable a new link to ensure you have the multipathing options correctly configured for the new path.

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.

[QFlux]

Morning Shawn,

Thanks for getting back with such a complete answer. Had not thought to check the multipath setup however it appears we're safe as all the hosts are targetting all 8 adapters correctly.

While I'm thinking about our MSA do you or anyone else know if I'd see any benifits from changing the controllers over to FC mode with 16G modules vs sticking on iSCSI and upgrading from the 1G Copper to 10G Fibre modules?

Thanks

[Shawn_K]

Hello,

Changing from iSCSI to FC requires a bit of work. You will need to consider that you are changing initiators from an IQN-iSCSI to a FC layout. So all host will need to be unmapped and then remapped. It is not hard but be aware that will add time when doing that kind of controller swap. You will also need to check SPOCK to confirm the swap of NICs to HBA to make sure your servers can accept a supported HBA that works with the MSA2050. Just make sure all the components along the host IO path are compatible from array - switch - server.

Now to upgrade speed from 1G to either 10G (iSCSI) or 16G (FC) will depend on your applications and bandwidth requirements plus where your systems are physically located. If the array and server(s) are in the same data center then FC may offer you some speed upgrade availablity. However, if your servers are located at a distance then the use of iSCSI and uplinks becomes expensive. Some databases benefit from the higher bandwidth but then may become bottlenecked by write speeds to the drives. So before upgrading check your drive speed, bandwidth needs, performance requirements and how much your servers can actually push down to the array. This will help you make an intelligent choice towards what your buisness needs.

That being said - who doesn't enjoy faster?

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.

[QFlux]

Thanks Shawn you've been a great help.