EV - preparing for EV Server Migration - Copying Partitions

[SYMAJ]

As I prepare to perform an EV9 to EV10 migration (including 32bit to 64bit platform) I have a question regarding the copy of the vault partitions. 

I have relocated the SQL databases from one of the 2 EV servers to a seperate SQL 2008R2 server - all completed OK. 

I have the two new EV servers prepared in terms of Windows and EV pre-reqs, with a 'vanilla' install of EV10.0.4 completed (not configured).

As part of the migration we obviously need to have the Indexes and partitions in the new location.  These are currently on LUNs presented over fibre channel from a NetApp filer - in the new environment the EV servers are virtual servers - and the data will be located into VMDK's under ESX control.

My plan is as follows:

Stop all partition collections by setting start and end time to the same value

Amend Mailbox and Desktop polices to stop user additions and to stop the move of items by users being reflected in the vaults.

As I have over 2TB of partitions to copy I cannot do this during a single outage in which backup mode would be effected, so with the above being done I planned to copy the closed partitions to the new location over a week or so utilising xcopy with the relevant switches (/X/O/K/E/H).  Once all closed partitions are copied then the final one would be done during the migration outages along with the indexes. 

However, after copying a number of partitions I compared the number of files/folders in the various partitions and some vary.  The variances are minimal, but surely they should be the same !  My fear is that the original partitions (live) are being accessed and archived items which exist in collected vaults (CAB files) are being extracted upon access - thereby equating to the original archives having a small number of additional files over the copied ones.  Is the a valid assumption ?  If not - what would be causing the difference in the original and copied partitions ?

As I do not have the opportunity to copy / move all 2TB of partitions in one session - is there an alternative to the above method ?  All of the technotes / forum articles involve putting EV into backup mode during the copy/move - I can't do this.

Any pointers appreciated.

AJ 

[Rob_Wilcox1]

Why can't you do the copy in one session? If you have disabled archiving etc, then there isn't a need to go in and out of backup mode, right? Do you allow deletions? That could affect the item count too. Are there the same number of CAB files on source and target partitions that you have tested?

[John_Chisari]

You assumption is most probably correct - the way you could check is to count all files, but exclude any ARCH* in it's extension.  All *.ARCH* files are files extracted from CAB's and are not required to be copied over.

A tip you may want to look at using Robocopy instead of using xcopy - use Robocopy on the new server to copy files from the older server.  You can use the /MT switch in robocopy which allows multi threaded copies.  It will definitely help especially with the Index files.

[SYMAJ]

Rob, one session is not practical as we are a journaling site also and to copy 2+ TB, plus indexes, plus export/ import etc etc is too long. I have not stopped archiving yet either. I am only looking to take the closed partitions at present. John, thanks for the advice on ARCH files and ROBOCOPY. AJ