Re: replicating information store
- From: "John Fullbright [MVP]" <fjohn@donotspamnetappdotcom>
- Date: Tue, 10 Oct 2006 13:10:02 -0700
Well, you would be talking either shared storage or some form of
replication. Shared storage would be a traditional MSCS cluster that would
provide protection against host server component failures with the exception
of storage which is shared. Shared storage means shared bus, which
typically means two servers in the same datacenter. In the event of storage
failure or data corruption, no protection is provided.
With replication, you replicate the data from the hos or at the storage
device level to a second host/storage device either synchronously or
asynchrounously. Replication will involve third party products, and it
important to understand the different types of replication and the MS
support policy as pertains to them. You can find this in KB 895846,
http://support.microsoft.com/kb/895847/en-us
I believe that a few host based replication solutions have already been
mentioned on the thread. From the storage device, you'll look to you
storage vendor, or a third party clustering vendor that provides an agent to
support your storage platform. A specific example of leveraging the native
replication of a storage vendor to implement standby clusters can be found
in the current issue of Technet Magazine:
http://www.microsoft.com/technet/technetmag/issues/2006/10/FailoverClusters/default.aspx
This design falls within the support of 895846 because snapshots, not live
data, are what is replicated asynchronously. The design can provide an RPO
as low as 5 minutes, and RTOs in the 2-4 hour range. Moving to a clustering
technology like VCS (with a storage agent) can reduce the RTO to the 15
minute range, however the RPO would still have a lower limit of about 5
minutes for async.
To reach an RPO lower than 5 minutes, you would need to move to a synchonous
replication technology. With synchronous replication, a write isn't a write
until both ends commit it. This can cause significant performace
degredation as the distance increases (speed of light you know). Because of
this, synchronous replication is typically limited to distances of 80 KM or
so. If you're servers are in the same datacenter, or on the same campus, or
even two campuses across town, this solution is viable and would provide the
lowest RPO. As distance increases, you'll likely find yourself considering
asynchronous solutions with a bit higher RPO. In any event, if MS
supportability is important to you, you'll want to select a solution that
falls within the guidance of 895846.
John
"Drew Campbell" <DrewCampbell@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:294A9E14-6457-4E09-B197-8FD08970F5FC@xxxxxxxxxxxxxxxx
One of our clients is using Exchange 2000, and we just installed it on a
second server for quick disaster recovery. We want the second server to
be
completely identical, so when one goes down, network users can retreive
their
mail from the other one instantaneously. I was able to set the public
mailboxes to automatically replicate to the other server, but I can't
figure
out how (or if it's possible) to set up automatic replication of private
mailboxes. Can anyone help? Thanks.
.
- Prev by Date: Re: Replicating the information store
- Next by Date: Exchange 2003 Enterprise install (with upgrade to Exchange 2007 to follow fairly quickly) hardware advice
- Previous by thread: Re: Replicating the information store
- Next by thread: Exchange 2003 Enterprise install (with upgrade to Exchange 2007 to follow fairly quickly) hardware advice
- Index(es):
Relevant Pages
|
Loading
