How often do you perform backups of the harddrives on your servers?

Ok, I backup in three different ways.
1. I'm using RAID1 to prevent downtime and loss of data in case of harddrive failure.
2. My second server is constantly trying to keep a backup of all the data on my first server so that I can stay online if, for example, the first servers power supply wrecks all the hardware in the first server.
3. Every day all vital data such as customer information and their webpages is backed up to a network server which puts it on tape. Every friday the friday-tape is taken to a location outside the NOC.

BTW, if you have any suggestions for me, they are welcome :D

This is going to sound like a really stupid question but I forgot what NOC stands for :)

we back up a diff every night offsite to one of our other servers, for clients we then mirror that data again to another server, we synchronize mysql databases real time for our customer databases and radius data for our outsourced radius stuff. That way we can recover pretty quickly in the event of a failure.\

I don't really trust tapes, they can be handy if you have physical access to your boxes, but with the cost of extra hard drives it is hard to justify the cost of tapes and tape devices.

NOC = Network Operations Center
(ie. data center)

This is how I would do it:

1) Raid 1+0 (just in case of harddrive failure)
2) Incremently tar files (only updated ones) to a backup drive on the server.
3) Set cron to run a script that transfers the tarred file to a Cheap backup server from another Noc.
4) Set cron to untar the files daily.
5) Set cron to tar all files to a backup drive on the backup server.
6) Set cron to run a script that transfers the incremently tarred file of the day to a similar server at home.
7) Untar the incremently tarred file on the home server. Tar all files on the home servers to the backup drive.

Both the backup and local servers would mirror the main server.

a) If all drives on Raid fail on the main server, I still have the files on the backup drive.
b) If that backup drive fails, I still have the files stored on the backup server at the other NOC.
c) If the main harddrive fails on this server, I still have the files on the backup drive of this server.
d) If both the main and backup servers fail, I still have the files on my local server.
e) If the harddrives on RAID fail on my local server, I still have the files tarred on the backup drive of this server.

If there's a transfer failure from point A to point B, the script will instead attempt to transfer from point A to C. If there's a failure in transfer, an email is sent... which sets off my emergency pager.

This, in my opinion better than having someone doing a tape backup for you. Much cheaper too.

We're using rsync to back up our stuff from one drive to another. rsync will also work acrosss a network to another server. The nice thing about rsync is that you only send what has changed.

Frank

Originally posted by ffeingol
We're using rsync to back up our stuff from one drive to another. rsync will also work acrosss a network to another server. The nice thing about rsync is that you only send what has changed.

Frank


Frank - is it possible to share your rsync script please ?

Originally posted by mark224
NOC = Network Operations Center
(ie. data center)

NOC's and datacenters are different things - an NOC is located in a datacenter, and staff in the NOC monitor the servers etc.

Regards
Matt

We back up nightly. In todays environment it's not even an option any longer.

Originally posted by voxtreme-matt


NOC's and datacenters are different things - an NOC is located in a datacenter, and staff in the NOC monitor the servers etc.

Regards
Matt

Yes, most people don't know the real meaning of the word is a room where you monitor the datacenter

Overall NOC has became the "lingo" word for DataCenter.

Originally posted by CRego3D
Yes, most people don't know the real meaning of the word is a room where you monitor the datacenter

I've seen someone from Exodus refer to a NOC -- *network* operations center -- and DOC -- *datacenter* operations center -- separately.

But I think that distinction is only going to be made by those operating long-haul networks, if ever.