Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I'm setting up some new Windows 2012 servers to replace old ones currently on 2003. One of my concerns is to try and have a suitable DR plan to get them back up and running if we have a major failure. I plan to have the following:

Server 1: Domain Controller, File/Print server, SQL Server, Web Server

Server 2: Exchange (2013) Server

Server 3: "Backup" Domain Controller

I will also have spare server hardware (identical) to be used in case of failure.

My main issue is around the backup of Server 1. Firstly, just to say there will be a maximum of 100 users and the file/print and SQL will probably only be used by about 25. I know in a totally ideal situation the DC would be that and nothing else but this is a relatively small setup so I don't see a problem with having all those services on one box.

In an ideal world I'd like to run a disc clone and then have that available to restore in case of total server failure (followed by then applying backups of database, user files etc). However as I understand it you should never restore a clone of a domain controller as then all the numbering is out of sync and replication will fail etc.

I thought of prepping the server but taking a clone just before promoting it to a DC. However reading the notes on SQL Server (I will be using Express 2012) it says that while it will run on a domain controller you can not promote (or demote) the server while it is installed. I.e. I can't install SQL and then promote the server.

So it seems currently my DR plan would have to be restoring a Win2012 clone and then rejoining it to the domain, promoting it, reinstall SQL etc i.e. quite a lot of steps.

One thought I had is if I did go against the rules and cloned the whole system, then in case of failure could I do a full restore but then immediately remove the "backup" domain controller from the domain (and later reintroduce it) (to get the replication numbering back in sync)? That would I assume have lost any AD changes since last backup but given the size of our organisation they are likely to be minimal.

Any other thoughts welcome.

share|improve this question

2 Answers 2

Have you considered using VMs instead of trying to stuff everything on the hardware ? That would make your system easier to backup and manage. You might need a couple of additional windows server licenses but not only will you get around the issue of having multiple services running on the same OS but you'll also gain the ability spread the load more efficiently between physical machines.

It will also make DR much easier assuming you have a good backup policy.

share|improve this answer
    
I have indeed wondered about using VMs although am hesitant, e.g. if I had the DC in one VM and everything else in another VM on the same server I'm still worried if the whole server died (as opposed to just one of the VMs). –  george May 7 '13 at 12:40

Two things spring to mind here - you could separate the services on the one box by using virtualisation - HyperV springs to mind as the obvious example, and that would solve your problem nicely with regards to SQL and DC functions tripping over each other.

The other suggestion I'd have would be to abandon the idea of using disk images for backup - SQL server has its own rich backup functionality, even in the Express edition, which you should take advantage of to backup databases to files, and you then can use a "traditional" backup system to backup those files plus the shared files (which will allow you to recover just one deleted file with no hassle easier than you can from an image) and the "system state" of the server, which is what you need to do a disaster recovery of a DC.

share|improve this answer
    
SQL server agent is not included with SQL server express which makes it more difficult to create backups on schedule –  Stephane May 7 '13 at 8:31
    
@Stephane - good point, forgot about that. Luckily a question about that specific bit has been asked and answered on Stack Overflow: stackoverflow.com/questions/487675/… –  RobM May 7 '13 at 8:59
    
@RobM please see my answer about re VMs. As for the disk images backup, that is only for the real disaster scenario for if the hardware failed and I needed to (quickly) provision a new server. I still intend to do normal SQL backups and backups of data files to a NAS drive. And yes I do need to consider the lack of SQL Server Agent in Express. –  george May 7 '13 at 12:42
    
@george re: your comment about it all being on one server, that's true but you haven't lost anything - you were in the same position over a total hardware failure without virtualisation, and by not looking at virtualisation because of that you're losing the potential benefits - this is like refusing to play the lottery in case you win big and have to pay tax on the winnings. There's lots of reasons not to virtualise (and/or play the lottery!) but that isn't one of them. –  RobM May 7 '13 at 19:58
    
@RobM - fair point on the total hardware failure. I did already plan to have by "backup" DC running on ESXi. Would it be logical then to have a 2nd ESXi box that runs the DC in one Win2012 VM and the SQL/IIS/File Server in a 2nd Win2012 VM? I believe a single Win2012 licence actually covers 2 virtual machines on the same host. The other thing to say is will my hardware be enough? It's going to a tower server with SATA drives (mirrored RAID). Originally for the file server I was to have 2x500GB RAID for the OS and then 2x1TB RAID for data - don't know best practice for this virtually? –  george May 9 '13 at 15:59

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.