Two Windows 2003 servers mirrored with data r

Microsoft Windows server standard 2003 r...
June 2, 2010 at 15:25:46
Specs: Windows 2003
I have a client that have one server running two apps whiched crash and data loss occured with a week downtime. They want mirrored servers with data replicated so when one server goes down the other takes over with minimal downtime

See More: Two Windows 2003 servers mirrored with data r

Report •

June 2, 2010 at 16:17:54
First, you want to setup a separate File Servers do not host the files on your DC or AD server. Then you can setup File Replication Services.

The best results come from servers with the drive partitions and space setup exactly the same so you don't have to worry about one of the servers running out of space without you knowing about it.

Then you setup you Home and Share directories with the exact same ACLs. Then go in and setup a the DFS Management under Administrative Tools. Go down to replication and create a new group. Follow the wizard. This will setup a replication. Start Loading the files on you Source File Server and they will start to replicate to the backup file server with in 5 minutes or what ever you have the sync set to.

If you do this with a live system MAKE SURE YOU DO A COMPLETE BACKUP WITH ACLS AND ALL PROPERTIES. If you do the replication the wrong way it will delete all of your files in place of the structure on the other server.

Now if you want the system to be completely fault tolerant then you may want to use Name Spaces. I try to steer clear of them because when they mess up they messup big time

With Name Spaces you can use Name Space referencing in your drive mapping script in place of UNC pathing. What this does is use the equivalent of a pointer to the files so that if one of the servers was to go down the users would still see their files through the pointer regardless of which server went down.

WARNING Name Spaces have built in replication and if you use it make sure to remove DFS Replication Groups First or you will get into a replication storm and wipe out both servers. You even get a warning if you try this that says that Unpredictable Results may occur. How do I know this, because I did it, One of the biggest mistakes of my life.

Also, note that I have tried using Name Spaces for a little while and it seems to be buggy to me. Some strange problems came up when using them. I went back to using DFS Replication groups and just switch the server reference in my Login Scripts if I have to bring one of the servers down. Its not seamless but it only takes a few seconds for every one to log out and back in.

I recommend you do some reading on how Name Spaces work and use test servers before you start but when it works, it is a wonderful thing. Especially if you have remote offices that you can replicate to to have off site backups. I run my backup scripts off of the replication server so it does not slow down the active file server and it works well for me.

As with any thing this complex I recommend you get a consultant.

Report •

June 2, 2010 at 16:42:26
Thankyou for your reply another question using this techique how about when the server goes down and the need to access line of business apps, also they have only one server now can you do this with two servers or do you need more because I see you mentioned not keeping data on DC or AD server

Report •

June 2, 2010 at 17:16:22
What you want to do is not cheap. You need a cluster but that requires a financial investment in a SAN with two servers running Enterprise [not standard].

We [my staff and I] found that a product called Marathon provided the failover without having the SAN investment. It was still about $8-10K just for the software. You would spend more than that on (2) Enterprise/SAN hardware.

Fact your client is only running one server tells me all of the above is out of their league financially. It also appears they do not even have the basics.


1. Raided drives [with hot spare]
2. Nightly backups tested daily as good [untested=no backup]
3. Image of the server for a bare metal restore
4. Server hardware with onsite maintenance [HP/Dell not IBM 5x24x365 they come out and replace broken hardware]

With these in place any tech worth their water weight can have you backup in 4hrs or less.

Ace talks about DFS but with server based apps that does not apply. DFS is great for data replication. Do note that the more complicated the system the harder to repair.

I would suggest you use the KISS principle in your case.

Report •

Related Solutions

June 3, 2010 at 08:45:45
"how about when the server goes down and the need to access line of business apps"

Like Wander Said it depends on how much your budget is and how much down time costs to your company. If it is so bad that it could cause mission failure then yes setting up multiple servers at multiple sites is what you need but you have to take into account the cost of the Servers (mine usually start at 3000 and go up as I add drives and perifials) and the cost of the sites.

Cheaper solutions would be SANs or even cheaper would be NAS but NAS would not help you with the fact that your DC and AD went down and what manages the authentication on your domain. These would just store the data until you get your systems back up.

Also note NAS is not fault tolerant as SANs but the more tolerant you get the more money you will spend.

Report •

June 3, 2010 at 08:56:59
NAS can be fault tolerant by using raided drives in the nas unit. NAS would not address the OP's issue.

Report •

Ask Question