I am testing for our migration to W2K and E2K. The proposal for our E2K
servers is clustered system with 4 Storage Groups and 4 DBs in each, should
total somewhere around 80 - 100 GB per E2K site. 4 sites in all. What I am
looking for:

1)
I have tested the Scheduler and utilizing the command file have found that
I have tested the Scheduler and utilizing the command file have found that
the server is taking quite a hit with all four storage groups backing up at
the same time. Is there a way to delay the start of each storage group ? In
other words kick one off wait "X" amount of time, then kick off the next ? I
was originally looking at multiple schedulers but that would require me to
have a different "nodename" for each scheduler. This is also required to
complete in a 12 hour window, with 65 server class machines backing up to
each of our 3 TSM servers at the same time.

2)
We have a long mail retention requirement for Legal purposes. I was able to
We have a long mail retention requirement for Legal purposes. I was able to
get them to agree to a Monthly scheme, but the documentation mentions that
archiving is not an option for the TDP for Exchange 2.2. What are other
sites doing to get long term retention? In Exchange 5.5 and TDP 1.1 we wrote
a script that ran once a day that queried the server and sorted them in
reverse order. That way we could keep the first backup for each month "in
reverse order that would be the last backup of the month" and deleted the
rest for that month. Other that policy settings has anyone figured out a way
to alter the retention ?

3)
There are multiple Domains in our proposal. For restore purposes I am
There are multiple Domains in our proposal. For restore purposes I am
suggesting we have a separate forest setup with a single box performing all
functions except the Exchange server. Then rebuilding the Exchange box when
ever a different domain has a restore request. Any other ideas out there ?

4)
I have not tested the cluster portion yet are there any obvious pitfalls to
I have not tested the cluster portion yet are there any obvious pitfalls to
avoid ?

If anybody has been through this before I appreciate hearing from you.

Duane Ochs
Systems Administration
Quad/Graphics Inc.
414.566.2375