Re: Microsoft Clustering on Exchange 2003

It all comes down to log replay when you mount the store. This is really a
function of how many logs you have and how many of them contain uncommitted
transactions. A big part of that is designing your storage for optimal log
replay times, and the other part is frequent backups to keep the logs
truncated. The backup solution I use is the one provided by the storage
vendor I work for. In a design, I typically backup every two hours with
verification on a seperate verification server, and the logs are truncated;
all without impacting end users. The storage is sized in accordance with
"Optimizing Storage for Exchange Server 2003". The particular storage
products I work with have unique advantages for write optimization and
mitigation of comingling.

If the problems you have is log replay time on service startup, switching
your clustering vendor won't solve a thing.





"Mike" <mjj2u2@xxxxxxxxx> wrote in message
news:1138724958.217505.49690@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
> The thing that takes so long on the fail-over is the time it takes to
> remount all the information stores. Yes the actual server may fail over
> but the Exchange services aren't available for around 10 minutes after
> that. You say that you are running Exchange servers and you haven't
> seen this problem? This surprises me especially if you are running them
> over WAN's! Could you men passing the active node to a different server
> as apposed to failing the node? It is a bit faster for me if I "Move"
> the active cluster from one node to another, but when it actually fails
> it takes about twice the time. Also, regardless of the transition it is
> mounting the information stores that takes so long. If you don't mind
> me asking, how long does it take to mount your stores?
>
> As for how would other cluster servers solve this problem.... Some of
> the clustering services we are looking at actually use one OS on two
> physical servers. In essence it writes the data to two clones
> simultaneously. If for any reason one becomes unavailable the other
> will answer for it immediately. The key here is that you don't have to
> remount drives, or information stores. They are already up and running.
>
> Have you had any experience with alternative clustering software for
> Exchange?
>
> BTW, thanks for your comments!
>
> Mike
>


.

Relevant Pages

  • Re: Concurrent Storage Question
    ... Subject: Concurrent Storage Question ... Application Cluster) is taking care of clustering. ... storage at the same time but if I halt one server then storage ...
    (AIX-L)
  • Re: Thats it?
    ... Local storage dies with the machine. ... Rod ... MVP - Windows Server - Clustering ...
    (microsoft.public.windows.server.clustering)
  • Re: Network topology for Small Business HA - Clustering?
    ... Windows 2008 has additional restrictions around the storage w/r/t Failover ... may rule out clustering for you. ... MVP - Windows Server - Clustering ...
    (microsoft.public.windows.server.clustering)
  • Exchange 2003 inetinfo taking out SMTP once a minute!
    ... I have Exchange Server 2003 running on Windows Server 2003. ... problem definately seems to be related to the mailbox stores on this server ... each in their own storage group - one of them also houses the ... * Moved mailboxes in both stores into new stores (in different new storage ...
    (microsoft.public.exchange.connectivity)
  • Exchange 2003 inetinfo taking out SMTP once a minute!
    ... I have Exchange Server 2003 running on Windows Server 2003. ... problem definately seems to be related to the mailbox stores on this server ... each in their own storage group - one of them also houses the ... * Moved mailboxes in both stores into new stores (in different new storage ...
    (microsoft.public.exchange.connectivity)