Presently we have a local Vplex and 2 VNX arrays. The virtual volumes created from 2 VNX array are mirrored
with each other on back end. Virtual device having 2 VNX leg which are mirrored.
Now we are planning to replace the two VNX array with two new vmax array. Can some one help me and
let me know the best approach to follow for the migration.
Vplex supports 2 leg mirror devices. So how we can add third leg and how the cut over will be if we add one more leg?
What will be the high level steps if we use local mirror option.
I just want to know high level steps needs to follow to cut over 2 VNX with 2 VMAX.
Thank you !
You are correct. VPlex supports 2 legs only.
You have Two basic choices to migrate over to the VMAX.
1. Drop one leg from the VNX, then add one VMAX leg. When you are in sync, repeat the process for the remaining VNX leg.
2. Create new mirrored virtual volumes on the VPlex from the VMAX. Then using host based tools migrate to the newly formed devices. Most enterprise Unix systems have non-disruptive migration tools built in to their LVM systems. If your dealing with VMware then you could storage motion your VMs.
Sent from my iPad
Thanks. I will go with option 1 but I have a few more queries.
1. Drop one leg from VNX: I have to use device detach-mirror command to drop VNX leg.
My query - Do I need to specify device name to detach local mirror?
When to use --discard option and --force option?
For attach-mirror we always specify virtual device name as a source device whereas
we specify device name as a source device when we do detach-mirror?
device detach-mirror --device /clusters/cluster-1/devices/device_leg_to_keep --mirror /clusters/cluster-1/devices/device_leg_to_drop --discard --force
Out of the cli guide:
If the --discard argument is not used, this command detaches an up-to-date child from the parent RAID 1. The detached child becomes a top level device with the same virtual volumes as its parent.
If the --discard argument is used, this command removes the child from a parent RAID 1. The detached child will become a top level device but without any virtual volumes or guarantees about its data.
My mistake. You must use the force option when you use the discard option.
Sent from my iPad
One other way to do this would be to create a new raid one device that uses the VMAX as mirror legs. From there go to mobility central and do a device migration from the VNX based raid-1 device to the VMAX raid-1 based device. This would be non-disruptive to the host, driven 100% by VPLEX (and not the host CPU), and ensure that you always had at least 2 copies of data.
i try to use the suggestion provided by Don, but my distributed device/virtual volume was not a choice i could pick even after removing from consistency group.
5.2 vPLEX admin guide says,
Limitation, Device migration between distributed devices are not supported.
Am i doing something wrong?