svm mobility

[rare siren]

anyone have any experience of using this? I have some NFS SVM's to move and the customer is after zero downtime, understand the cutover blip as with vol move but does is genuinely do what it claims and zero downtime for users?

[meager mason]

Hi Darren, Are you moving them to SVM's on the same cluster or to a new cluster ? What's the workload, files, vm datastores, etc ?

[rare siren]

Hi David, new cluster, all nas, NFS & CIFS. I have used the vserver migration method using identity preserve extensively but heard about the mobility product recently.

[lavish berry]

vserver migrate is svm mobility. I think you meant SVM DR (that has identity preserve/discard flags)

vserver migrate works fine from our experience, although the version restrictions are stricter than when using SVM DR. What ONTAP versions are your source and destination clusters running?

[rare siren]

sorry, yes meant svm DR that i have used a lot. I have looked at the pre reqs and will bring up the source and target to whatever is needed if we go with mobility. I was more interested in case experience re downtime before I go promising anything to my customers.

[carmine jackal]

Which is now called SnapMirror SVM (yuck, dislike the new name)

[meager mason]

during cutover, as long as the snapmirror update process doesn't take too long, then it will just be an update of ARP within your data switch. If you can trigger a snapmirror update before the cutover it will save some time.

[rare siren]

Hi David, does mobility not use sync snapmirror? is it possible to run an update on that?

[meager mason]

Hi Darren, apologies I saw svm DR in your 3rd comment and "if we go with mobility". You're right mobility uses snapmirror synchronous. Few other things with mobility, client cutover and source cleanup I believe is enabled by default, so it's hands off and full speed ahead, they can be changed with -auto-cutover false, and -auto-source-cleanup false. I haven't test drive mobility yet, but keen to give this a go in the lab.

[rare siren]

Hi David, thanks very much for clarifying. Yes I have done a few test "mobility" this week and we used the cutover and cleanup false params so as to control things. we had NFS clients mounted and constant pings and writes taking place. Linux admin said that he only saw a tiny blip, similar to vol move so things look great. thanks for your help.

[native musk]

Since CVO does not support snapmirror sync, guessing it won't support this feature as well