I think you're missing the whole ownership chain. I guess you could use the o/s  "run as"  to impersonate your service account. It's not really a route I'd recoomend though.

you might want to try

master.dbo.xp_servicecontrol 'MSSQLServer'

Ah well .. two ways to decrease deadlocks ..

make reads dirty

increase spindles

The basis to reduce deadlocks, assuming the code isn't aweful, is to speed up transactions, if the disk...

shrinkfile will only attempt to remove unused space from the datafiles .. you need a certain amount of free space in your data files for growth and work space for...

I'm sort of confused by your claim of 100 deadlocks/min -- for a lightly used database that's very high. How are you measuring your deadlocks?

To get information on deadlocks enabale...

I'm not sure that microsoft fully support sql server on vm ware - from experience I wouldn't use vm ware in production. 2003R2 comes with virtual servers, maybe that's the...

you need to backup the transaction log

I wouldn't think a backup would roll back as what could it roll back to ? - possibly the transaction log I suppose. You could try stopping and restarting the...

you might want to try batching your input.

are you using sp_xml_preparedocument  , if so this can give memory issues in certain conditions.

to be honest using adv server with 2gb ram...

If you have a dedicated SQL Server then you might want to monitor with a view to increasing the memory given to sql server.

There's absolutely no point in leaving memory...

if you'd like to send me a private message I'll send you some stuff to help -- too long to post

I know this will sound contrite but one of the fundamental facts of performance tuning is that you learn by actually doing it!!

If you try the two methods you...

sp_helpfile executed against the database will give filenumbers, tran log is usually 2 with primary filegroup being 1.

in qa against database use these commands

checkpoint

go

dbcc shrinkfile(fileno)

go

nothing strange here, turn your connection pooling back on.

with regard to what - a cluster gives you failover so you don't really have DR. if node dies, cluster fails over, rebuild node or get new node, re-introduce to...