Disk Subsystem Problem during snapshot

We have set up a new SQL environment that has to replace our current SQL
environment. On our current environment we apply a snapshot every night from
our production database (the production database is hosted on another
environment). This means every night approximately 15GB is being copied to
the subscriber. It takes a long time to do this (3+ hours), but is not a
problem in our environment.
Exactly the same snapshot has to be applied on our new SQL environment. In
this environment we are having serious problems with our Disk Subsystem when
applying the snapshot. In the new environment we use a shared HP MSA1000 Disk
System. When applying the snapshot, the load is so high on the MSA1000 that
it can’t handle all the requests no more. This results in a crash of SQL
server (SQL can’t find the tempdb anymore) and other systems that use the
MSA1000 (two more servers, so three in total) are having problems with the
disks also (event id 51).
We’ve contacted HP and Microsoft and the conclusion will probably be that
the load generated by applying the snapshot is to high for the MSA1000.
Microsoft’s explanation about how this can be, knowing that the exact same
process runs on an old environment, is very reasonable (although it’s still
hard to believe that old hardware operates better under high load than new
advanced hardware).
I’d like to know what the opinion from you all is about applying a snapshot
every night to a subscriber, which involves copying approximately 15GB of
data. Should it be no problem to this, or is it very uncommon to do this
because of hardware limitations? Is it normal that the MSA1000 has problems
of this kind with this, or should it be able to handle the load (should it
result in disks that cannot be contacted anymore or should it just perform
very slow)?
Your opinion is very much appreciated!
Kind regards,
Jan Martijn Schuur
"Martijn" schrieb:
...
> I’d like to know what the opinion from you all is about applying a snapshot
> every night to a subscriber, which involves copying approximately 15GB of
> data. Should it be no problem to this, or is it very uncommon to do this
> because of hardware limitations?
15 GB is a large amount of data for a nightly snapshot. Why don't you use
the incremental transactional replication?
Your other problems with this are not familiar to me. Sorry.