Exchange 2003 Cluster Recovery Options
I am trying to determine what my best course of action is to recovery my 2
node (a/p) Exchange 2003 Ent cluster to my DR Site.
The Exchange 2003 Ent cluster backs onto a SAN that is synchronously
mirrored to the DR Site SAN. We have a high bandwidth low latency network
between our production data center and DR Site.
We have a live AD domain controller at the DR Site. Redundant internet,
firewalls, hot servers.
So my question is, should I look at geographically dispersing my cluster? I
can easily span a VLAN across the WAN. Can I just pick up my passive node
and move it to the DR Site? Or would I need to add a 3rd node for the DR
Site?
Or my other option is to setup a standby cluster (single node) pre-staged.
I was thinking of running this on VMware though (I understand support may be
an issue in this configuration).
Anyhow any thoughts or comments would be appreciated.
Mike.
ApprenticeRating: 0%, 0 votes
There's a support policy for replication of Exchange data to contend with,
and most async replication isn't supported. Host based replication suffers
the same fate.
#
The only solution I can think of that is supported is async replication of
snapshots, because it's not live data. In Exchange 2007, MS introduced log
shipping which is one way out of the support quandry. If you're still on
2003, look at the October 2006 issue of Technet Magazine.
#
It's basically a standby cluster with async replication of snapshots.
Was this solution helpful? Show your Appreciation by rating it:
ApprenticeRating: 0%, 0 votes
There's the thing you see.
If you won't use VMware because Microsoft might pull the rug on you
you have to use something else.
Microsoft will pull that same rug with whatever 3rd party geographical
clustering option you choose, i.e. Veritas.
They'll also get stroppy if you're using SAN based replication.
Honestly, you're damned if you do and damned if you do it another way.
All you need to do is to see what your support people and/or vendors
can and will support and how competent & professional they consider
them to be.
Was this solution helpful? Show your Appreciation by rating it:
ApprenticeRating: 0%, 0 votes
True enough. Another way of looking at this is if you have a real and
serious requirement for Microsoft support you will probably have a
Premier support agreement. If that's the case then you are supported
by Microsoft when virtualised.
If you have VMware and like the idea of VMotion you can have a cast
iron assurance that your vendor or VMware (EMC) will support you to
the hilt, better than Microsoft will because VMware will ensure that
your problem is fixed without refusing at the 3rd fence and demanding
that you put it all on physical hardware and regenerate the problem
before going any further.
Personally I would rephrase Rodney's last two sentences (probably the
first and only time I would ever presume so to do) to read "You will
need support on Microsoft technologies at some point. Will your chosen
vendor fully support the technology they sold you?"
Was this solution helpful? Show your Appreciation by rating it:
ApprenticeRating: 0%, 0 votes
Run the standby cluster (single node, pre-staged) on stand alone hardware?
If that is the case should I not then look at the geographic dispursion of
my cluster nodes? That almost seems easier, assuming I can move the existing
passive node.
In that scenario (geo cluster) I am just not sure how fail-over would work?
"Rodney R. Fournier [MVP]" wrote:
Was this solution helpful? Show your Appreciation by rating it:
ApprenticeRating: 0%, 0 votes
of Microsoft support is huge for some applications and organizations. I
would talk to your management before going that route. Exchange has issues,
as does every major application - you will need Microsoft badly at some
point. Will they support you?
Cheers,
Rodney R. Fournier
MVP - Windows Server - Clustering
# - Clustering Website
# - Blog
# - Cluster Training
ClusterHelp.com is a Microsoft Certified Gold Partner
Was this solution helpful? Show your Appreciation by rating it:
ApprenticeRating: 0%, 0 votes
VMware would be a great solution. Yes, support might be a drag from
Microsoft but your storage vendor and VMware themselves would drool
over you like a teenager at prom night.
Do you know about VMotion from VMware? You should and if you do, you
should be looking at using it.
Even if you don't use VMware the other solution that pops straight
into view is Veritas clustering. You ditch the 2nd node from your main
site and send it over to the DR location.
Your storage vendor is almost certain to have a Veritas specialist on
staff, it's part of their core business (ours has several).
Don't waste a moments thought of what Microsoft might say to you about
the virtualisation path though. The other vendors will take real good
care of you.
Was this solution helpful? Show your Appreciation by rating it:
Post a New problem for IBM DS4800 Hard Drive Array
Email this problem
Can you Help with these Network Storage problems?
Tags | Blog | Repair Directory | Extended Warranty | Partners | Jobs |
