Disk Redundancy across different server

Question

I have 3 servers, all with the same specs:

• Intel CPU
• 8 GB RAM
• Linux or BSD
• Single 2TB desktop SATA with more than 10K Hours of operation, with only less than 300 GB Used

My provider cannot install a second hard drive, but can guarantee me that the drive will be replaced immediately in case of failure, with another equally crappy drive.

The likelihood of drive failure is high, and since I can't use RAID, I was thinking about keeping a back up of each machine on all the other machines, so that there are always 2 copies on 2 different drives, plus the original.

I would synchronize the drives every hour, with rsync, to guarantee some sort of redundancy, since bandwidth inside the DC is free, so it would be much cheaper than offsite backup. (A daily offiste backup is kept anyhow).

What do you think? Any suggestion?

Answer 1

You don't list your OS, so it's tough to say, but if you're using Linux you should look at DRDB.

Answer 2

One thing to consider is setting up virtualization and copy the VM's over, since you didn't say what these systems are doing. If the overhead won't hurt for your application it may be easier for redundancy to have your systems rigged with heartbeat monitoring and fencing to keep the VM's running in case of failure. VM's also encapsulate your server.

The mechanism to use for this would be dependent on what you want to do and how to do it and your budget. You can get distros that use KVM and DRDB to make this automatic.

Using in-datacenter transfers as "free"...it may not cost you anything financially, but it can incur network and disk I/O that can in turn affect your system performance, perhaps noticeably.