Convert to switched FC backend

Hi Gal, 

What I would do is remove 1 cable from each Unity SP and plug it into the switch. Zone the hosts with those ports. Do a rediscover and make sure the paths show up, then do the other Unity ports

I’d definitely do one “fabric/port group” at a time and run an array re-discover any changes. The most reliable way would be to check the ITLs for a sample of storage-volumes to the Unity. To do this, CD into the storage-volume and list out the details w/ the --full flag

Example:

VPlexcli:/clusters/cluster-2/storage-elements/storage-volumes/C2_CLARiiON1539_LUN_00005> ll --full

Name                           Value

-----------------------------  ------------------------------------------------

application-consistent         false

block-count                    262144

block-size                     4K

capacity                       1G

description                    -

free-chunks                    []

health-indications             []

health-state                   ok

io-error-status                ok

io-status                      alive

itls                           0x5000144260072010/0x500601643de0084b/5,

                               0x5000144260072010/0x500601653de0084b/5,

                               0x5000144260072010/0x5006016c3de0084b/5,

                               0x5000144260072010/0x5006016d3de0084b/5,

                               0x5000144260072011/0x500601643de0084b/5,

                               0x5000144260072011/0x500601653de0084b/5,

                               0x5000144260072011/0x5006016c3de0084b/5,

                               0x5000144260072011/0x5006016d3de0084b/5,

                               0x5000144270072010/0x500601643de0084b/5,

                               0x5000144270072010/0x500601653de0084b/5,

                               0x5000144270072010/0x5006016c3de0084b/5,

                               0x5000144270072010/0x5006016d3de0084b/5,

                               0x5000144270072011/0x500601643de0084b/5,

                               0x5000144270072011/0x500601653de0084b/5,

                               0x5000144270072011/0x5006016c3de0084b/5,

                               0x5000144270072011/0x5006016d3de0084b/5

largest-free-chunk             0B

locality                       -

operational-status             ok

provision-type                 legacy

storage-array-family           clariion

storage-array-name             EMC-CLARiiON-APM00111501539

storage-volumetype             normal

system-id                      VPD83T3:60060160c9c02c0020e02328f69be811

thin-capable                   true

thin-rebuild                   true

total-free-space               0B

underlying-storage-block-size  512

use                            used

used-by                        [extent_C2_CLARiiON1539_LUN_00005_1]

vendor-specific-name           DGC

The highlighted section are the ITLs. The first column are the VPLEX initiator ports and the second column are the array target ports. For the VPLEX ports ending in 1 are in PG1/”odd fabric” and the ports ending in 0 are in the PG0/”even fabric”. Once the first set of back-end paths are moved over, do an array re-discover and re-check this output. The number of ITLs should remain the same and you should see logins on both PG/fabrics.

Hi Anker,

Many thanks for your reply.

I have a couple more comments/queries please.

1) When you mentioned hosts - It is purely the back-end connections that are changing as the front-end connections are in a different fabric switches.

2) Currently Unity is connected in b/e ports 2 and 3 on each director - would it be recommended to re-run the connections from 2 and 3 or just build the new fabric using connections from 0 and 1?

3) Can all of this be done whilst the system is operational with no disruption?

Thanks

Gal

1) got it, then nothing has to be done there
2) either way works should work. Since its already logged into 2/3 it *might* be quicker to pick up those paths, but with rescans it should work fine with using connections 0/1

3) yes, this is why we are doing only one cable per SP and not both. Just ensure the ITLs are logged in via the commands above. Though as in all things storage, might be best to do it during low IO periods

Thats great, thank you for clearing things up - I will be doing this out-of-hours and at-risk period.

If you could bare with me for a couple of mins more just so I can finalise the order in my head...

Currently I have both Port 4's on the storage connected to Port 2's on the VPLEX backend.

1) FC-PORT-GROUP-2 is made up of A1-FC02 and B1-FC02.

2) Therefore, my first action will be to physically relocate the Port 4 SAN connections to FC Fabric A
3) Followed by the FC02 connections from each director's backend ports to FC Fabric A
4) Zoning would be along the lines of each director's FC02 ports to both storage controllers Port 4 ports (this would lead to a doubling of initiator to target relationships)
5) Run a rediscover array
6) Check ITL's on virtual volumes

7) If all okay then repeat process for FC-PORT-GROUP-3 and array targets on Port 5.

Welcome any comments or suggestions,

Thanks,

Gal

Looks good to me!