topic Re: Loop Switch Resilience in MSA Storage

Loop Switch Resilience

Schwarzie — Thu, 17 Nov 2011 15:35:02 GMT

Theoretical question for someone ....

EVA 8000 in a 2C 12D configuration. 2 loop switches that connect all drive enclosures in the lower half of the cabinet die/go offline. The 2 remaining loop switches that connect the top half remain operational throughout.

Say that the EVA is fully populated and split down the middle equally into 2 Disk Groups.

What would happen to the VDisks, data traffic and host connectivity?

Thanks for any input given.

Re: Loop Switch Resilience

Torsten. — Thu, 17 Nov 2011 19:33:16 GMT

A failure of 2 switches would be very unlikely, but fatal.

Split the disk groups into the upper and lower area would be a bad design, because of RSS group layout.

Re: Loop Switch Resilience

Schwarzie — Fri, 18 Nov 2011 09:09:56 GMT

Completely agree. However, the two disk groups are split with a virticle line. ie DiskGroup1 is made up of bay's 1 to 7 and DiskGroup2 is made up from bay's 8 to 14.

Is the outcome still the same?

Re: Loop Switch Resilience

Torsten. — Fri, 18 Nov 2011 09:17:23 GMT

If you would loose both lower switches for any reason, you loose half of the disks - this would be fatal.

Re: Loop Switch Resilience

Schwarzie — Fri, 18 Nov 2011 09:58:54 GMT

Thanks for your input.