9 Legend

 • 

20.4K Posts

August 17th, 2010 21:00

you can run step 2 as many times as you like. I had to migrate a couple of LUNs that were used for Oracle dump area, they were 1TB in size. I ran my initial copy during the day (since these LUNs were mostly used at night), i set throttle relatively low as not to impact overall array performance and then i would run multiple incremental's until the final day of cut-over. Data on the target will be "crash consistent" because your host is still up and running.  It will be truly consistent when you shutdown the host and perform your last incremental copy.

San Copy utilizes "tracking" bitmap (saved in RLP)  to track changes between updates, i am not aware how to get "delta" information from that bitmap.

9 Legend

 • 

20.4K Posts

April 18th, 2010 20:00

looks good to me. You will need to make that source clariion has access to target LUNs.  To create the session you would do this (play around with throttle setting, make sure you don't bring production host to its knees) :

NaviSeccli.exe -user Administrator  -password password -scope 0 -address 10.1.8.30 sancopy -create -incremental  -name Lun546to19C8 -srcwwn 60:06:01:60:99:51:0D:00:DE:6A:04:C6:09:49:DA:11  -destwwn 60:06:04:80:00:01:90:10:45:18:53:30:31:39:43:38 -throttle 10  -verify

Step 3)

NaviSeccli.exe -user Aadmin -password  password -scope 0 -address 10.1.8.30 sancopy -start -name Lun546to19C8  -copywholelun -nomark all

this will copy the whole LUN in bulk mode, meaning RLP will not be utilized.

Step 5)

NaviSeccli.exe -user Aadmin -password  password -scope 0 -address 10.1.8.30 sancopy -start -name Lun546to19C8

You may also want to explore using clones as the source of your SAN Copy session. Final thing that you need to be prepared for is to resignature VMFS file systems after the migration. VMFS gets UUID fromt he Clariion LUN and even though data is identical, UUID of the new LUN is different.  Search google for "virtual geek disk resignature" and you will see what i am talking about. Good luck.

2 Intern

 • 

152 Posts

April 19th, 2010 08:00

Hey Dynamox thanks for the response.

So in step 3 when we do the bulk copy and if there are any changes made to the source lun those changes will be copied to RLP is that right ? So in step 5 to copy the changes made to the source lun during the bulk copy we will run the SAN Copy session as you mentioned in your command is that right ? 

9 Legend

 • 

20.4K Posts

April 19th, 2010 09:00

don't forget to remove SAN Copy sessions when migration is completed.

NaviSeccli.exe -user Admin -password  password -scope 0 -address 10.1.8.30 sancopy -remove -name  Lun546to19C8

9 Legend

 • 

20.4K Posts

April 19th, 2010 09:00

in step 3 changes made to the LUN will not be copied to RLP. You will get all the changes in your final step (step 5)

2 Intern

 • 

152 Posts

April 19th, 2010 10:00

you said changes made to the source lun during the step 3 will not be copied to RLP. So if there are any changes made to the source lun  during the bulk copy how we are going to track those changes. If they are not copied to RLP where will they be copied ?. Becuase when we run step 5 only the changes made to the source lun will be copied right ? How will the session know from where to copy those changes made to the Source Lun.

9 Legend

 • 

20.4K Posts

April 19th, 2010 12:00

RLP will keep session information, which blocks changed since the copy but not the actual data.

By default, incremental sessions use EMC SnapView™ snapshots, which impose a copy-on-first-write
(COFW) penalty, as the source data changes during the SAN Copy transfer. The “nomark” option that is
available only through NavisecCLI does not perform COFW as a result of changes to the source LUN
during the SAN Copy transfer. Eliminating the COFW penalty is beneficial because the initial
synchronization step needed to replicate the full contents of the source LUN can be very time-consuming.
Increased performance can be expected since disk I/O normally used for the COFW operation is now
available for use by the SAN Copy session and any host application such as the Oracle database instance.
Incremental SAN Copy sessions started with the “nomark” option require an additional update session to
replicate a consistent state of the source LUN.

2 Intern

 • 

152 Posts

April 19th, 2010 14:00

so to make a consistent copy of the Source and Destination lun we run the step 5 after the bulk copy  . right ? 

2.2K Posts

April 19th, 2010 15:00

Yes, with all the hosts offline.

4 Operator

 • 

5.7K Posts

April 20th, 2010 00:00

My email reply didn't come through, so now my answer from yesterday:

Why not use Storage VMotion ? That way the VM’s can stay online during the movement of the storage/vmdk !

9 Legend

 • 

20.4K Posts

April 20th, 2010 05:00

you have to run VMware Infrastructure Enterprise edition to use vmotion and svmotion, smaller shops may not have it.

4 Operator

 • 

5.7K Posts

April 20th, 2010 06:00

Hmmm, didn't know that. Perhaps because we're always running that configuration

9 Legend

 • 

20.4K Posts

April 20th, 2010 06:00

from ESX 3.x to vSphere ..looks like regular vmotion is now available in Advanced version as well but svmotion still enterprise and ent +

http://www.vmware.com/products/vsphere/buy/editions_comparison.html

2 Intern

 • 

152 Posts

May 3rd, 2010 09:00

Thanks Guys for your responses.

We do the bulk copy from source to target lun using (nomark copy whole lun) when the host is online. When i do the incremental copy after the bulk copy is done is it compulsory that the host needs to be offline to copy the incremental changes or is it ok if the host is online and at the last cutover i make the host offline and run the final incremental session again. Please clarify ..

9 Legend

 • 

20.4K Posts

May 3rd, 2010 09:00

yes, you can run as many incrementals as you like ..to keep  target close to source as possible and then run your final incremental when the host is down.

No Events found!

Top