Exchange Server

Dave · Guest

Hi,

I want to build two exchange servers, I would like one to be the working one
and the second to basically just be a replica of the first one but if the
first one crashes the second one picks up the ball and runs with it.

Can this be done in Exchange Server 2003, I am not really looking for a
clustered server just some form of having a carbon copy on a different server.

Hope this makes sense.

Thanks

Dave

Colby Holland [MSFT] · Guest

This isn't possible in the same environment. You would have to have a
separate domain/network for this to work.

Colby

--
Please do not send e-mail directly to this alias. This alias is for
newsgroup purposes only.

This posting is provided "AS IS" with no warranties, and confers no rights.

"Dave" <Dave@discussions.microsoft.com> wrote in message
news:1C84D274-0F3C-487C-BD3C-942B304BBB1E@microsoft.com...

Al Mulnick · Guest

Actually, that is almost the definition of a clustered setup. What's your
aversion to a cluster?

Al

"Dave" <Dave@discussions.microsoft.com> wrote in message
news:1C84D274-0F3C-487C-BD3C-942B304BBB1E@microsoft.com...

Benoit Boudeville · Guest

I guess he needs to perform mailbox snapshots just like Lotus Notes is able
to do.
Unfortunately there is no native way to do this, however if you have the
proper hardware (a SAN), then you can use SNA snapshots to make exact copies
of your disks to another set of disks or bay. You can then restore your last
system state backup to a cold server and use the exact copy, however this
requires a very sharp and detailed (and roughtly tested) procedure.

Because of how Exchange associates the mailbox GUID to the user object in
Active Directory I believe the could be a way (using a third party utility,
but I don't know any) to synchronize mailboxes, this would require having to
double user objects in AD (one "regular user+mailbox" object and one "mailbox
only" object)

Then you can swap regular mailbox GUIDs with the cold-mailbox GUID on the
regular user object. That could work, now this is hardly feasible in huge
environments.

Just some thoughts, never seen such a thing :)

Good luck,
Ben.

"Al Mulnick" wrote:

Al Mulnick · Guest

After reading that post, I think now would be a good time for you to check
out the disaster recovery whitepaper. You'll be particularly interested in
the "Dial-tone" scenarios. http://www.microsoft.com/exchange/library

While at it, you may want to have a look at the VSS documentation and how it
relates to Exchange/Windows etc. You'll find docs about it on MSDN and
http://www.microsoft.com/windows

-Al
"Benoit Boudeville" <BenoitBoudeville@discussions.microsoft.com> wrote in
message news:5798E9DC-3E7C-4BE6-8D6F-A3E98FF964EA@microsoft.com...

Benoit Boudeville · Guest

Dial-tone recovery isn't exactly what he needs, however it's a very efficient
way to bring back service online and then merge restored information stores
with temporarily ones, but you might still loose data if failure occurs at
the storage level itself (ie: loosing transactional logs).

"Al Mulnick" wrote:

Al Mulnick · Guest

Sure. But if you replicate data, you also will likely lose data right?
Unless you have a data center close enough that you can have in-synch data
100% of the time. Most solutions will put you within minutes, but not 100%
mirrored.

I think you bring up a good point: we need more information about the
requirements to really know what needs to be done.

-ajm
"Benoit Boudeville" <BenoitBoudeville@discussions.microsoft.com> wrote in
message news:2FEE2D0E-7F40-4381-91EF-AB89BA3F9419@microsoft.com...

Benoit Boudeville · Guest

This really depends on yoru SAN data replication, in general yes, inter-bay
replication has some delay since to not have latency issues when replication
data is buffered on the "other side". If your "source" bay or on volume is
lost, then time to flush write cache buffer can take long in very big
scenarios, that's why MS only recommends this situation in very specific
cases, and when everything has been checked (HCL, vendor info, etc.)

But no you don't loose data since you always have a "master" and a "slave",
data replication only occurs in one direction (ofc if you loose data at the
application level it's another story, you have to handle it in Exchange
itself)

"Al Mulnick" wrote:

Al Mulnick · Guest

Right. That's my point You will always have some latency loss in your data
set. Maybe only seconds, maybe only minutes, maybe longer (but then, what
would be the point?).

I still don't have a crystal clear understanding of what the requirements
are. We seem to be missing a few pieces of information for what the
original requestor wants.

Al
"Benoit Boudeville" <BenoitBoudeville@discussions.microsoft.com> wrote in
message news:8CCCEB93-DBA3-4B3B-87E1-43C85F668473@microsoft.com...

Benoit Boudeville · Guest

We might never know, since the thread seems to be "closed" (using MS's Web
newsgroups here and I see a green check box near your first reply, which
seems to indicate that it answered his question :))

"Al Mulnick" wrote:

Glenn L · Guest

Dave, if you are still monitoring this thread, I think what you are looking
for is offered by a product from XOSOFT.
www.xosoft.com
They have a product called WANSyncHA designed with Exchange in mind.
Basically it performs continuous replication of all database transactions as
well as database writes (basically all file system I/Os) to a second
exchange server in the same exchange org. This server remains offline until
the first one fails.
The product handles all the failover tasks.

--
Glenn L
CCNA, MCSE 2000, MCSE 2003 + Security

"Benoit Boudeville" <BenoitBoudeville@discussions.microsoft.com> wrote in
message news:C10916C6-C3E0-4AE3-B497-3E2B0A2DB703@microsoft.com...

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