RE: Exchange Server Redundancy
- From: "Mulnick, Al" <Al.Mulnick@xxxxxxxxxx>
- To: "'[ExchangeList]'" <exchangelist@xxxxxxxxxxxxx>
- Date: Mon, 8 Mar 2004 16:16:16 -0500
Microsoft implements "shared" clustering meaning that hardware resources
share hardware. This is a high-availability solution and not a
fault-tolerant solution mostly aimed at reducing hardware and upgrade
downtimes (upgrades as in service packs, hotfixes, etc). Implementing an
a/a/p cluster could certainly allow for a lot less downtime, but at a
significant cost in terms of hardware/software licenses as you've noticed.
5x9 availability requires so much more than just the software and hardware
it runs on, so I won't get into that in this thread. Suffice it to say that
5x9's availability is a team effort that encompasses 8 layers of the OSI
model ;-)
I for one would like to see the public folder replication scheme go away.
But for fun, let's assume that Microsoft did do that. What would that mean
in terms of availability that you can't get today? A public folder
replication interval is 15 minutes. Cluster failover is ~ 10 minutes or
less depending on implementation. I'll take clustering over the public
folder replication :)
There are software products out there that claim to be able to cluster the
servers geographically and provide nearer to instantaneous failover. I
haven't seen them myself, and can't vouch for them, but I have a hard time
understanding how they could be faster than a hardware failover and are
certainly going to add cost.
I think that understanding what the managers really want would be helpful.
Do they want fault tolerance, or just highly available messaging service?
If the latter, do they want to have the data that's in there all the time,
or would they prefer the service to the data if it came to it? Exchange
does have the ability to offer "dial-tone" (although I've seen some flaky
dial-tone so understand that Exchange can do better) service whereby if a
store fails, you can wipe the store, restore service in the amount of time
it takes to wipe the store and mount a new one with the same information,
and then put the data back over time using the RSG option. It's possible
and can work for a lot of people. VSS is another option that may be of
interest to them. Sounds like hardware failure is the most concern, and
clustering is really a good option for that. Gives a hot-standby host that
automatically fails over in case of hardware failure reducing the time to
heartbeat notification + time to bring the other node online or about 10
minutes depending on how implemented.
Just my thoughts. Glad you got it worked out.
Al
_____
From: Jason Merrique [mailto:j.merrique@xxxxxxxxxxxxxxx]
Sent: Monday, March 08, 2004 3:44 PM
To: [ExchangeList]
Subject: [exchangelist] RE: Exchange Server Redundancy
http://www.MSExchange.org/
Hi Al,
Bit of a delay, but this is a reply to Al Mulnick and Sonny Willis (Thanks
for your input on this chaps, your expertise is very much appreciated.)
The whole redundancy issue was brought up by management who (as always)
would like as near 100% uptime as is possible. As it happens, the hardware
failure last week only caused about 2 hours of downtime and safeguards were
put in place to insure that a repeat occurance would be highly unlikely. My
initial instinct was that it wouldn't be worth the investment in additional
hardware and licensing for the 99.999% (2 hours in the whole year) to 100%
improvement. So I'm left thinking that its just not worth it.
So clustering and cloning may not be the answer. But I would like to confirm
- Is there really no way to replicate mailboxes between servers using
Exchange 2003? Doing this in the same fashion as Public Folders would be
excellent. Maybe allowing a simple redirection to the secondary server if
the primary falls over? Or is this just wishfull thinking on my behalf!?
Sonny: I don't suppose you know how NeverFail copes with differences in
hardware and SIDs between the primary and secondary server? If its an exact
replica of the software on the primary, I'm not sure how the Secondary would
cope with this...
Thanks again guys,
Cheers,
Jason
_____
From: Mulnick, Al [mailto:Al.Mulnick@xxxxxxxxxx]
Sent: 02 March 2004 15:54
To: [ExchangeList]
Subject: [exchangelist] RE: Exchange Server Redundancy
http://www.MSExchange.org/
I'll offer both for what it's worth.
Clustering is not a redundancy solution. It's a high availability solution
(HA). Redundancy infers no downtime while HA infers that you can withstand a
hardware failure and can be up and running much more quickly than you could
with manual intervention.
For 30 users you want to weigh the costs. You can achive this, but wouldn't
it be cheaper and almost as fast to have the parts on hand and just replace
them if they go bad? Hardware is fairly reliable these days and that can be
more cost effective and easier to work with. Clusters are only a little
more difficult than a standalone (IMHO) but they do have additional
licensing costs for many third party applications that you use to support
your infrastructure. AV, Backup, etc fit this category typically.
As for active/active clusters with other apps on them. You can do this. It
is possible, but again the complexity goes up with each application as does
the requirement to monitor more closely and provision more appropriately in
case of failover (think about it, if you fail over a node due to hardware
failure, you have to have the resources to absorb the added burden. This
often indicates bigger hardware components to handle the load.) Multiple
apps on an Exchange server are not always straightforward and easy as
Exchange likes to live alone as do other apps. I'll reiterate that it can
be done, but is it worth it in your environment?
, and safeguards
check out http://www.microsoft.com/exchange/library
<http://www.microsoft.com/exchange/library> for some helpful docs.
Al
_____
From: Jason Merrique [mailto:j.merrique@xxxxxxxxxxxxxxx]
Sent: Monday, March 01, 2004 7:15 PM
To: [ExchangeList]
Subject: [exchangelist] Exchange Server Redundancy
http://www.MSExchange.org/
Hi All,
One of our exchange servers experienced a hardware failure this morning
(excellent start to the week!), and as it was the main mailbox store, it
caused a few hours of downtime for most of our users. To prevent this from
happening again, I'd like to have a certain amount of redundancy. I
understand that in a cluster of exchange servers this is possible as they
effectively act as one logical exchange server.
Is it possible to achieve the same level of redundancy without setting up a
cluster, by synchronising the mailboxes between the two servers?
Would a cluster be overkill for our two servers and 30 odd users?
As part of an exchange cluster - would it be possible for the server to take
on other tasks as an individual?
If you could offer personal advice, or a recommendation of literature, that
would be excellent.
Cheers,
Jason
------------------------------------------------------
List Archives: http://www.webelists.com/cgi/lyris.pl?enter=exchangelist
Exchange Newsletters: http://www.msexchange.org/pages/newsletter.asp
Exchange FAQ: http://www.msexchange.org/pages/larticle.asp?type=FAQ
------------------------------------------------------
Other Internet Software Marketing Sites:
Leading Network Software Directory: http://www.serverfiles.com
No.1 ISA Server Resource Site: http://www.isaserver.org
Windows Security Resource Site: http://www.windowsecurity.com/
Network Security Library: http://www.secinf.net/
Windows 2000/NT Fax Solutions: http://www.ntfaxfaq.com
------------------------------------------------------
------------------------------------------------------
List Archives: http://www.webelists.com/cgi/lyris.pl?enter=exchangelist
Exchange Newsletters: http://www.msexchange.org/pages/newsletter.asp
Exchange FAQ: http://www.msexchange.org/pages/larticle.asp?type=FAQ
------------------------------------------------------
Other Internet Software Marketing Sites:
Leading Network Software Directory: http://www.serverfiles.com
No.1 ISA Server Resource Site: http://www.isaserver.org
Windows Security Resource Site: http://www.windowsecurity.com/
Network Security Library: http://www.secinf.net/
Windows 2000/NT Fax Solutions: http://www.ntfaxfaq.com
------------------------------------------------------
------------------------------------------------------
List Archives: http://www.webelists.com/cgi/lyris.pl?enter=exchangelist
Exchange Newsletters: http://www.msexchange.org/pages/newsletter.asp
Exchange FAQ: http://www.msexchange.org/pages/larticle.asp?type=FAQ
------------------------------------------------------
Other Internet Software Marketing Sites:
Leading Network Software Directory: http://www.serverfiles.com
No.1 ISA Server Resource Site: http://www.isaserver.org
Windows Security Resource Site: http://www.windowsecurity.com/
Network Security Library: http://www.secinf.net/
Windows 2000/NT Fax Solutions: http://www.ntfaxfaq.com
------------------------------------------------------
Other related posts:
