Redundancy

T

The_Unbeliever

Honorary Master
Joined
Apr 19, 2005
Messages
103,193
Reaction score
10,233
Location
Nkaaaaandla
So I'm tasked with finding out about redundancy etc.

Now, we got the following servers (on site, not at the office) :

1x CentOS Linux server (Informix database)
2x Windows2008 server
1x Windows7 PC

Those 4 machines have been identified as highly critical, and need to be backed up. (disk images).

Backups must be automated (background) so that it have little or no impact on the system.

We're looking at bare-metal backup, so that we can restore to clean PC's and be up and running with as little fuss as possible.

What suggestions do you guys have floating around?
 
Backups are a very important start, however its not really redundancy.

Redundany would entail that a device can fail and people can keep on working as if nothing happened. Now this can become a expensive excercise, but a good one to try and keep costs down while introducing redundancy.

So what you would do is, instead of running a single server for a task, run two servers which can fill the same task. The most expensive way would be to get something like 2x ESXi vmware servers with heartbeat between them, split the servers over them, if 1 physical server fail, the virtual server/s are instantly migrated to the other physical server, and employees keep on working as if nothing happened, IT guy starts figuring out wtf is going on, in his own time without having everyone ask him the whole time when is the server back up.

Now there are cheaper ways of doing it, but it usually requires software to support redunancy, or using tricks like load balancers ect. Ultimately you want do double up everything, servers, network equipment, multiple interlink connectivity ect. and even 2 power feeds that are fed from different UPS's, generators ect.

So maybe a good start is, is all you want backups? If you want more, how far do you want to go and how much is your budget.
 
Redundancy can mean a lot of things. Do they have dual network cards that are teamed? Dual Power supply's is your raid configuration able to to allow you failure with one drive?

Those sorts of things are more of a failsafe to me than anything else.

Then you start looking at if these are virtual machines, can you move them over to an active server with as minimal downtime as possible?

Do you have the correct UPS's that will alert you when a server does fail or alert you if power failure and how long the UPS will last for?
 
I'd also look at virtualizing the setup over 2 identical hosts. As Tinuva points out redundancy and backups are 2 different things. Redundancy improves uptime. Backups are there when you suffer data/system loss. Virtualizing your systems can make backup/restore much simpler as you are HW independent.

If you want to be cheap and go with one host, then throw in proper RAID, redundant power etc. You can then recycle one of the old servers for a backup device to backup your VMs to disk at night and then stream them to tape for offsite storage during the day. Traditional backups tools like BackupExec have clients for VMs. BE is still excellent value for SME level IMO. There are lots of other fancy backup tools like Veeam when you go Virtual. The flaw in this plan is when you suffer a complete failure of the VM host (fire/theft etc). Business needs to decide what their risk is and what they are prepared to spend. I often have clients insisting on certain features till they see the cost and suddenly you get to what they actually need ;)

Edit: I thought I'd mention that once you go with 2 or more VM hosts for redundancy you need shared storage, which doesn't come cheap. VMWare 5 has introduced a "poor mans" shared storage option which lets you mirror the storage between to hosts which gives you a similar effect to having shared storage.
 
Last edited:
The_Librarian said:
So I'm tasked with finding out about redundancy etc.

Now, we got the following servers (on site, not at the office) :

1x CentOS Linux server (Informix database)
2x Windows2008 server
1x Windows7 PC

Those 4 machines have been identified as highly critical, and need to be backed up. (disk images).

Backups must be automated (background) so that it have little or no impact on the system.

We're looking at bare-metal backup, so that we can restore to clean PC's and be up and running with as little fuss as possible.

What suggestions do you guys have floating around?
Click to expand...

http://www.stratus.com/Products/ftServerSystems.aspx - servers with a minimum five 9's availability

http://www.windowsitpro.com/article/systems-management/product-review-stratus-ftserver-4500-129998 - review

We use Stratus, though mainly the bigger (non-Windows) boxes. Awesome is an understatement.
 
Last edited:
You must log in or register to reply here.
Back
Top
Sign up to the MyBroadband newsletter
X