Disaster Recovery Plan and Exchange 2003

[geo1s2]

Hi Everyone,

I work on a school campus and we are going to implement a DRP to provide contingency for the school in the event of fire, flood, etc.

We have a MS 2003 Server PDC, member servers and MS 2003 Exchange server

We currently have a mirror for our data server in another building, so our data is covered. But we do not have a mirror for our exchange 2003 Server.

My question is what is the most cost effective way of having email contingency in the event of a disaster.

Taking into account that:

a) We only need essential users to have email access immediately after a disaster

b) Our budget may/may not stretch to an exchange server.

c) We currently use an Ironport to filter emails, would we require an additional Ironport at the contingency site? Or could we get by on a software option?

d) Is there a way to set up the essential users on a POP3 basis temporarily?

If you can help me on some of my questions, I'd appreciate it.

thanks,

George.

 

[SeanFlynn]

Is there any chance to upgrade to Exchange 2007? If so, you can use the clustering technologies built into Exchange 2007 to replicate the data to the secondary server without a need for third-party software.

Other than that, you can look at a product like DoubleTake to replicate the data.

I would be curious to hear if the upgrade to 2007 is out of the question, then we can discuss other options.

John Price

http://www.ChampionTechGroup.com

http://www.learnITstuff.com

 

[geo1s2]

Thanks John,

Upgrading to Exchange 2007 may well be an option. I have also been looking at double take but that is in the region of £1,000 so maybe the upgrade will be viable. I will look at the costing and get back to you.

This may sound like a naive question but if we were to take either the upgrade option or use double take would we have to have like for like hardware or could we get away with purchasing a less expensive server or robust desktop?

As it would not be a permanent solution just good enough to keep the business running in the initial time after a disaster.

Thanks again, George

 

[SeanFlynn]

It depends on the number of users, how big your information store size is, etc. Also keep in mind these two items:

1) Exchange 2007 is 64-bit only
2) To use the built-in clustering functionality, you must use Exchange Enterprise edition.

By your last response, I would think that both of these factors may lead you toward using a different solution since they will require a pretty decent investment.

I do not have personal experience with doubletake, so I can't help you very much there, sorry.



John Price

http://www.ChampionTechGroup.com

http://www.learnITstuff.com

 

[geo1s2]

As automatic failover is not required, we need to operative 12 hours after a disaster. Would an option be to image the complete Exchange 2003 server with something like Storage craft Shadow Protect which can restore an image hardware independently on to a server at our contingency site(update the image on a weekly basis so loss of data will be minimum)

In the event of a Disaster switch on the secondary Exchange 2003 server with the up to date backup on it.

Do you think this is a realistic option?

 

[Zelandakh]

12 hours after a disaster? Yikes. I'd get a server with redundant disks made by a proper manufacturer and document the build. Make sure you are backing up properly and then have a second server on the domain with the OS fully hardened but no Exchange.

In the event of server failure, you can either build the real one again or build a new one. Depends on problems. Not too bad if you have that much time to play with.

 

[geo1s2]

Thanks Zelandakh,

Yes it does seem like at awful lot of time. It's not the solution I would prefer as I would like to go down the route of the Exchange 2007 server using Cluster continuous replication.

I have to look at the costing for this.

Anyway thanks for your help. Also thanks John for your input.

 

[58sniper]

DoubleTake and Never Fail would likely work fine in this scenario. I agree - you can recover from almost anything in 12 hours. Granted, it might not be full functionality (say, if you have to restore TBs of data), but access to send/receive new email is quite quick.

2007 clustering is a nice option, but do your research before trying to stretch a cluster across WAN links. Make sure you understand what's involved in failing over AND failing back. YMMV.

Pat Richard
Microsoft Exchange MVP

 

[geo1s2]

Thanks Pat,

The contingency site is on the campus connected by fibre which
has a gigabit link.

What I have to take into account for the 2007 clustering is the cost of a new server and the cost of the upgrade to exchange 2007 enterprise.

Is never fail reasonable in price?

Geo.

 

[Zelandakh]

Neverfail would be the cheapest option here.

Sounds like SCR is an option too.