1

Total novice question. I have a server that hosts websites, contains databases, etc. It has several hard drives in RAID arrays. If a disk fails, the server continues until a replacement disk is installed.

However, what if other component(s) fail like RAID controller, RAM, or whatever? Is it normal to just have a bunch of spare parts on standby?

I was wondering if an identical server would/could be set up to somehow be a mirror image of the other server. And that, if the primary server ever failed, we could switch to the other one without missing a beat.

Is this a normal backup strategy? If not, what is normal? If it is, how would it work to get the data from primary to secondary real-time?

John Gardeniers
  • 27,458
  • 12
  • 55
  • 109
user1761600
  • 11
  • 1
  • 1
    Sorry, this isn't a good fit for our site, as it's too broad a question. Please see the [FAQ](http://serverfault.com/faq) for further details. – SmallClanger Oct 22 '12 at 08:32

2 Answers2

3

You could either use load balancing and fail-overs, or host the website in a virtual machine that runs on top of a number of physical servers. If a single physical server dies then the rest of the servers can host the virtual machines until the downed machine is repaired.

Depending on the enviroment you might have spare parts in place, but HP have a 4-hour service contract where they will get you a spare part within that 4 hour window. A lot of servers have hot-swap RAM and disks although most would require a power-off to replace a power supply.

It's quite a fascinating and very technical area - if you're interested, have a look at the Network Load Balancing and Clustering services in Server 2008.

tombull89
  • 2,964
  • 8
  • 41
  • 52
  • Thanks. We can definitely deal with an outage to turn off the server. We can even suffer an hour or two downtime. That 4 hr thing sounds nice, but our locations are pretty rural in the US. I guess having more spare parts at each location is the most normal way to go. I think we've been short in this area and caught scrambling in a failure. – user1761600 Oct 22 '12 at 08:46
0

A secondary server as mirror would be great for minimizing downtimes. But this will not replace backups. If database or some part of file system became corrupted or accidentally modified and then mirrored you will end up with 2 non functional servers.

Laurentiu Roescu
  • 2,266
  • 17
  • 17