Open Replicator - guide

Steps for hot pull are referenced in EMC Solutions Enabler Symmetrix Migration CLI guide. Since there are SCSI reservations on the disk, you will need to do complete shutdown of the cluster, create OR session and then bring it back up. Pace and Ceiling are explained in the same document. Quorum will be migrated using Windows cluster admin, this procedure is online.

Thanks Saurabh & Dynamox

One more question, if there are 100's of servers to be migrated using Open replicator and for example 1000 of Luns are allocated for those servers. So how many of these can be migrated in a single shot and which factor decides the capacity that can be migrated? So if i have to start the migration what is the decent capacity to start with?

Regards

Dinesh

i think the new code supports upto 1024 sessions but i would not want to test this limit . I think the right approach would be to migrate cluster by cluster.
use ceiling If you plan to migrate too many hosts at once. you would not want to choke Front ends.

regards,

Saurabh

i have done hundreds of devices (at slow pace though not to impact production, using -precopy). When app was shutdown i would increase pace so it would finish faster and then activate the session.

Thanks once again. Suppose if i do open replicator hot pull from clariion to VMAX, kick start open replicator session for 1 cluster & wait till 100% migration is done and then go with the next cluster. Thats what you meant. Right? Suppose if FAs or Clariion ports are not highly utilized then can i combine few servers and migrate?

Regards

Dinesh

after smaller devices are completed... we can increase the pace to 3 (maximum)... if you have dedicated ports for OR then there is no issue at all.. on an average for 1.5 TB data to complete the migration it takes around 2 hours.

even if you  have dedicated ports for OR you might be putting too much load on the source devices. We had a couple databases where we started OR at pace 5 and DBAs were screaming that response times went up ..so we had to back it down.

@dynamox , that is true in some cases.. if the db is shutdown and the host is down we can use the highest pace.. on a live db we should take things into consideration.

Thanks guys.

I have to do hot pull from CX to VMAX. For example lets consider we have  planned to migrate 10 hosts by this week. Consider 7 hosts is going to use port group A (4 FA ports) and 3 hosts is going to use port group B (2 FA ports).

1. Can we able to use 2 OR sessions (one for each port group)?

2. When is the best time to schedule downtime for hosts? (means whether in the beginning itself shut down the hosts, move the hosts to vmax, kick start the OR session with donor update and bring up the hosts) or (start the OR session with donor update, allow the data to migrate, if so how much data? and bring down the host and move it to VMAX and bring it up)

To Answer Your questions as far as my experience goes.

1. Can we able to use 2 OR sessions (one for each port group)?

I will always suggest to do one by one.. instead of doing 2 together.

2. When is the best time to schedule downtime for hosts? (means whether in the beginning itself shut down the hosts, move the hosts to vmax, kick start the OR session with donor update and bring up the hosts) or (start the OR session with donor update, allow the data to migrate, if so how much data? and bring down the host and move it to VMAX and bring it up)

Downtime may be evenings or whenever people are using it less time.  From the beginning itself shutdown the hosts, (both sides) start OR session after 10 to 20 % of copying is complete you can start the host at VMAX side, mount the volumes. If you are using donor update then allow the data to migrate, after all the data copied  bring down the host and move it to VMAX and then bring up.

KD8EWE wrote: If you are using donor update then allow the data to migrate, after all the data copied  bring down the host and move it to VMAX and then bring up.

i am confused with this paragraph, he is coming up on VMAX and pulling data from CX already ?

@dynamox wrote.

i am confused with this paragraph, he is coming up on VMAX and pulling data from CX already ?

Nope... donor update happens on both sides..  The steps are as below.

1. Shut down server.(both sides Clarrion and Vmax)

2. Symrcopy create [use donor_update,  update the writes to source lun

3. Symrcopy activate

4. Make the server up on VMAX side

5. Start the applicaiton you should not  make any changes on the server unless the new luns are larger than the old ones.

if he is doing hot pull why does he need to "after all the data copied  bring down the host and move it to VMAX and then bring up."