Exchange Server

me · Guest

- Back ground
My company consists of three major and several minor exchange sites all
running Exchange 2003

A Major site consists of
2 front end servers – load balanced
2 back end servers – clustered
SRS server
Bridgehead – smtp server
And has the ability to send messages to the internet
Holds 2000 users

Minor site
Front end server – load balanced if two
2 back end servers – clustered
Bridgehead server
And uses the major site to send and receive messages
Holds up to 1000 users

- My objectives are to be able to
Send and receive messages within four hours
Email for the last 90 days within 48 hours, data older than this is held
in another system

I am formulating a DR plan in my head that consists of the following,
please bear in mind that what I say DR I mean a complete loss of a
physical office (bit it minor or major).

Basically I need to be able to recreate a minor or major site in DR.

My thinking is this…

- Perquisites that need to be in place in the DR site.

Network links

Active Directory infrastructure, so I do not need to worry about DC & GC’s.

Exchange server hardware, which will be “warm” in the DR site and :
running and built to a windows OS level
in a workgroup
ensure that they use the same drive letters for system ad application
partitions
ensure that IIS, SMTP, NNTP, ASP .net are all installed.
Exchange installation files are store directly on the server (same time
than installing over the lan)

My plan is to use the /disasterrecovery switch along with the dial tone
option to be able restore messaging functionality within the four hour
period.

I see the high level steps to achieve the four hour timeline is to

In AD reset computer accounts for the exchange servers that are being DR’ed
Rename the exchange servers in DR to the exact name of the servers being
DR’ed.
Reboot and join to the AD, reboot.
On each server run the exchange setup/disaster recovery.
Recover the IIS system state
Reboot.
Mount the stores ( that do not contain any EDB’s )
Stores should mount giving users empty mailboxes that they can send and
receive messages.

Once environment stable and users are happy with there access,
Use backup application to restore mailbox EDB’s to the recovery storage
group and exmerge old data into the new mailboxes.

Does my high level thinking have any major holes in it ? is this
possible, are there dimplier ways to achieve this ?

I will appreciate people taking the time to read and understand this,
along with any constructive comments.

Thanks
P

Al Mulnick · Guest

- My objectives are to be able to
Send and receive messages within four hours
Email for the last 90 days within 48 hours, data older than this is held
in another system

Just so I'm clear, you mean to restore 90 days - (the delta of last backup
to failure time) correct?

Clients are always a concern. What client type do you plan to use? How will
the clients interact with the disaster site? How will they realize that the
existing site is down and should instead use the newly restored site in your
plan and network? Presumably the clients could be in the disaster site, so
you my thinking is that you may have some alternative client types
available?

It sounds to me like you're talking about a warm site for Exchange recovery.
You could have that warm Exchange server as a built but cold server not yet
joined to the domain at all. On activation, run through the steps and bring
it online etc. That would be easier to deal with.

Restoration of the disaster site should also take into account the EXACT
scenarios you want to use as triggers. I say that because if somebody comes
by after you and wants to resurrect the existing site or declare a disaster
situation that doesn't include the removal of the Exchange and DC's, then it
could get really interesting when they reconnect the environment. There
should be a plan to remove those servers from the environment to ensure that
they never ever can return to the general population under any
circumstances.

Since these are clusters, be sure to take that into account when staging
your warm servers.

Other ideas: You're one of the few that can have a warm site for AD. Some
would consider using virtualization to restore a DC/GC to, then restore the
Exchange server bits as needed. Possibly using virtualization for that as
well depending on the usage patterns expected.

My last thought for now: I've always approached this as an exercise of "who
needs what at what time"? Prioritization is helpful because not all users
need email as much as the others. For example, the c-level execs may need
email quickly to communicate with the rest of the company and the business
analysts to coordinate RTS. The janitor may not need email as soon. These
types of definitions can make a difference to your planning. I assume
you've defined all of this, but I didn't see it mentioned and it may be that
all users need the exact same RTS in your environment.

My $0.04 (USD) anyway.

	Exchange Server Forum Index -> Design	All times are GMT
Page 1 of 1