Link to home
Start Free TrialLog in
Avatar of Copyleft
Copyleft

asked on

Exchange 2003 - rednundanc, failover and/or /multiple Exchange servers

OK, heres the deal.

Our company has recently downsized, so I now have an unexpected excess of HP ML350 (single quad core Xeon/4GB RAM/300GB RAID1 SATA drives) and Windows Server 2003 Standard licenses. I also have two free Exchange Server 2003 licenses, as well as sufficient antivirus and backup software to cover all the boxes.

We recently experienced a serious outage where our exchange box, set up by my predecessor, was out of action for an entire day. Recovery could have been made that much swifter but the server was/is also being used as a file server  the catch 22 is that I was unable to take the entire system down for recovery until everyone was out of the network shares.

Magnificently, our ISP did not queue any messages during that period where the Exchange box was offline, so not only did our company lose a day of communications, they also appear to have lost a days worth of inbound emails.

As a result of this I have a mission to accomplish:

I want to implement some sort of failover system to keep an email service running in the event of the single Exchange server we currently use throwing a fit.

So, can some sort of redundancy be built in using multiple Exchange servers within one organisation ?

Is there such a thing as an equivalent of Exchange mirroring ?

Is it worth messing around with POP/SMTP accounts as a backup system in the event of Exchange failure ?

Webmail (ISP hosted) as a realistic alternative ?

What I am looking for is some recommendations on a robust system which includes alternatives to a single Exchange box for our organisation.

I am moving our web and mail hosting to a new provider over the next few weeks.

Today and tomorrow I am migrating all user shares off the Exchange server to a new file server I am bringing online so that its only roll is to serve emails, achieving the separation I need.

Where to go from here is the question. I dont necessarily need great detail: just recommendations and sound rationale.
Avatar of fishadr
fishadr
Flag of United Kingdom of Great Britain and Northern Ireland image

With the hardware and software that you have there is little you can do.

For high resilience and quick recovery you would normally have a clustered Exchange system so that if one server failed then the other would take over the operations. But you do not have the hardware and software for this.

Therefore if the main Exchange server fails you need to recover this (more than likely via restore methods).

There are many hardware and software options for mirroring the solution but this is all going to cost money.

I would suggest that you advise the management that with the current configuration it will take x number of hours to restore the Exchange server (from the previous backup) in a worst case scenario. They can then decide how much of an issue to the business this would be (if the server is down for this amount of time). If it is a problem then you will then be able to by software or hardware to implement a resilient solution.

The absolute minimum you should do is ask your ISP to queue the mail for you or switch to one that will. Therefore if the Exchange server is down the mail will be queued until you recover the server. I would not suggest going down the route of hosting the mail at the ISP unless you have a very small organisation.

At the end of the day it is not your problem and you can't work miracles - if a server fails then you implement a recovery process and you are limited to the resources the company has provided you with - the only way out is to spend money on more resources
ASKER CERTIFIED SOLUTION
Avatar of Exchange_Geek
Exchange_Geek
Flag of India image

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
Avatar of JayPeeAS
JayPeeAS

Installing a Clustered Exchange would be the obvious solution for your problem...but you're missing external sotarage (i.e. SAN, etc) but that's costly...although advisable.

You can also have ISP redundancy. Have 2 ISPs with 2 different fixed IPs and have one with a higher level of preference on your MX records.

You can also have something like www.messagelabs.com which will queue your messages in case your e-mail server is down and archive them for legal purposes as well...not to mention antivirus and antispam.

If the number of people at the company isn't very large (between 30 and 50 ppl) you can also go with a Hosted Exchange solution, either with your own server or just by buying accounts on a Hosted Exchange.

It's really up to you and how much money your company's willing to invest :)
Avatar of Copyleft

ASKER

Well, thanks for the information so far - very, very interesting, particularly the links in Exchange_Geek's reply, which look more or less exactly what I need/would like.

We have SAN in place already and a quite expansive network involving several other servers and site to site networks linked via LAN to LAN over WAN VPN, but the email side of it is something I had no involvement in until now (I inherited it as it stands) and given the 'spare' servers, operating system and exchange licenses we have I figured I could put something a little more robust in place.

You've given me some serious food for thought.

To be continued...

Can I verify one last point ?

Assume I have 2 physically identical servers.

Server 1 has Windows/Exchange 23k Backup Exec System Recovery and antivirus software installed.

Server 2 has last night's disk image of Server 1 applied to it daily.

Assume Server 1 goes haywire but the mailstore files are still accessible from within Windows.

Could the mailstore be easily transported over to what is in effect an exact clone, so that Server 1 can be brought offline and Server 2 (with an up to date mailstore) can step up to take its place ?

Just bouncing a theory around - I much prefer the idea of clustering described by Exchange_Geek
I believe you can, just as long as you get all the store database files across plus the logs. You'll have to replace the files on the second server with the store unmounted and then commit the log files me thinks. The server would have to have the exact same name, network configuration, patches...the works...an exact clone as you said.