I'm running Enterprise Vault 8 SP2 on Windows 2003 Enterprise. I"m in the process of moving my placeholders to a replacement server and it's slow then you know what. I'm using this command to move the placeholders FSAUtility.exe -m -s \\source -d \\destination. It could take a month at the rate this is going to move the placeholders. I think this is unacceptable and is serious flaw with EV. Any help is greatly appreciated.
We have archived about 1.5 tb of data. It seems to take an average 4 to 6 seconds each placholder move. The servers are on a GB network. The fsautility is also a single thread utility, so you can't run multiple instances of the utility without it crapping out.
I don't know if this will help you, but EV8 SP3 which is due out imminently will make this operation multi-threaded so it will improve performance.
One of the reasons for the slowness is that in the background the archived file is being exported and imported in to a new archive as well as the placeholder being moved. This was intended for when you needed to move placeholders and also move content to a new EV server. EV9 is likely to include enhancements to FSAUtility so that if source and destination are being managed by the same EV server and the same Vault Store, then the placeholder will be moved without the need for export/import.
The multithreading was added in 8 SP 3... multi-threading in internal testing made things go about 3x quicker. The real ‘fix’ is planned for EV 9, where we only move the placeholders and don’t export/import the underlying archived files – which is what happens now.
There is a registry key that you can use to up the amount of threads which you can play with and see if you can find a more optimum value:-
1. Create a DWORD registry value FSAUtilityFolderProcessThreads under the key HKLM\Software\KVS\Enterprise Vault\FSA. Valid values are from 1-20 and the default is 10.
2. To ensure that the new registry key value has taken effect check the Dtrace with the string "Number of threads for processing -m option are" for the number of threads being processed
As Rob said there were improvenments made in EV8 SP3 but further improvements will be added to EV9.