not sure I understand the last post ?? Here in the uK we have a number of "large" databases sponsored by government for various reasons, many are "developed" by US...

There's also a bug in sp4 that restricts awe memory, you must apply a post sp4 fix.

Read memory use from sysperfinfo table. You must decide which memory method you wish...

been there, book, T shirt, baseball cap, etc. etc. kick raid 5 into touch

I'd seriously suggest you rely upon sql backups. I've used sql server on SAN's for around 5 years and have always used sql backups.  There's some useful white papers available...

turn off the reserve physical memory.

set max memory = , say, 3.25 gb

leave min memory

set awe on

you are correct that you can use awe with 4 gb ram  but you...

Rudy, I do apologise for using stupid acronyms - I've done quite a bit of san work of late, including snia ( http://www.snia.org ) certification so I have loads...

I sense a measure of confusion here ?  a dbcc dbreindex against a table which has a clustered idnex will drop and recreate the clustered index and all secondary indexes....

Generally duplicating data is bad news, a single source of truth is usually best.

You say you're sorting the data, in that case a clustered index will achieve that, using a secondary index will require you to use an order by statement which may...

as an independent DBA I encounter this often. If I'm to perform dba tasks on prod systems then I need the access to do the task. I know SOX prefer...

pai isn't it? I figure you're running sp4 sql2k ?  check out the readme about chnages to monitoring in sp4. It may or may not be an issue - possibly...

sata disks will not give you the performance of scssi disks for random i/o, however as you're only specifying 15 spindles it's more of an external raid array.

Raid 5 is...

I'd be very wary on relying on any SAN based backups for recovery, most do not fully support COW and in a disaster/failure a recovery may well return inconsistant databases....

The point many overlook with this type of question is whether or not the underlying disk subsystem is a bottleneck.

I assume server 1 and server 2 are physically seperated...

things like data loads and index rebuilds can bloat a tran log - techniques to avoid this include:-

1. switching to simple recovery mode
2. batching of jobs
3. increasing frequency of log backups

I've used all...