Web Hosting Talk : Web Hosting Main Forums : Hosting Security and Technology : Backup solution

[Aushosts]

Which backup solution is better?
Raid-1 mirrored hard drives
or
2nd harddrive of backups

I will be running AMD Athlon(tm) XP 1800 1.53Ghz w/Cpanel

[apollo]

If you have 2 x 60GB drives, with RAID-1 you will have 60GB space in total (-OS files of course). But it offers data protection

Two drives in simple configuration will give you 120GB (- OS files) of usable storage space,but if one fails, you lose the data

[rigor]

My solution for you is this:

Raid-1 for redundancy
One spare drive for backup of everything.

Raid-1 will not protect your filesystem from futzing itself. Its hardware layer only. The day the filesystem corrupts completely for whatever reason you will thank yourself for having that extra drive holding all your data.
rsync is a nice utility to handle replication..

I know alot of people snide me saying i suggest expensive solutions, but i've been in real world situations where raid-anything didn't save my ass, filesystem corruption is a reality, it can happen, if it hasn't happened to you, count yourself lucky!

[Iggy]

I'd go with Rigor on this and one better. Set up a hotspare too. Drives are cheap now as well. And you might want to look into rdiff-backup as well http://rdiff-backup.stanford.edu/

Works really nicely for me. Gotta have Python installed though.

Thanks to Rigor for advice in the past!

Iggy

[mtvhosting]

I most definetely agree with Rigor.

[Aushosts]

Helpfull as usual

[ttjoum]

First of all i have to say that i agree with the "expensive" solution which was mentioned before.Because what you get is always depending on what you give.

But in case that you cannot afford this a simple solution that i use (because the boss doesn't want to hear about raid etc)
is that i make a bootable backup in cd's of the operating system(with mkcdrec) and then i just make frequently backups only the data.

[seg fault]

A second drive with backups is better in my opinion. Simply because, if you get hacked, you can have 100000 mirrors, and you will still have none in which you can use, since they are all hacked.

It's good to keep an old backup and recent backups.

You only need to backup your configurations and userfiles, since if you are hacked, you would want to get new source etc.

Another thing which is nice, is to have the second drive bootable with the OS in which you like so you can have the drives swapped right away and it's ready to go.

Good luck!

[sharper]

I would create a solution that would have 2 drives in a mirrored environment, then take a 3rd drive to run backups to. This way you have the mirrored drives for redundency, and the third drive to fall back on if you were to loose a RAID controller or something like that.

Shamon
Webolution

[DigitalIsles]

The one thing I think is missing in this conversation is what happens if your controller takes out all of the drives!!!?!?! The best solution in my opinion, which I know is even more expensive, is to backup to a second machine or put a tape drive in your machine to isolate that backup data from your live data.

-Robert

[Fujiwara Takumi]

amanda is possibly a good solution for replication, just a suggestion:

http://amanda.sourceforge.net/

[ttjoum]

Also be extra careful where you will put the backups because a friend of mine had them in a pc which was into the lan and always on and when they hacked him they also hacked the backup pc.
So try to save them in a safe "place".
I have a pc which does only that job and it is only on when i ftp the data into it.

[traktor]

Originally posted by psychalgia
amanda is possibly a good solution for replication, just a suggestion:

http://amanda.sourceforge.net/

Amanda's documentation seems to say it requires a "non-rewind" backup device, meaning a tape drive. Although Amanda uses a hard-disk staging area, I could find nothing in the doc to say it will or can back up to a hard drive.

This seems wierd to me. It would seem like Amanda could backup to a hard drive, but not according to my reading of the documentation.

Can anyone set me straight on this?

[traktor]

Originally posted by Iggy
<snip>And you might want to look into rdiff-backup as well http://rdiff-backup.stanford.edu/
Works really nicely for me. Gotta have Python installed though.

Hi, Iggy,

I think I have python installed. At least when I ask 'which python' I'm told 'usr/bin/python'.

The rdiff, if I understand the man page correctly, looks pretty handy, though the complexity of the restore commands scares me.

It looks like it can run a backup -- either to a second hard drive on our webhost machine, or via ssh to a machine here in our office -- and this backup would initially be configured to backup every darn thing in the specified webhost directorie(s).

Thereafter, when run, it updates everything that has changed since last time, and it also keeps copies of the old versions of changed files.

Is this essentially correct?

And it would appear that, since it is command-line driven, we could have a cron job driving a script which just specified one directory after another, until the entire machine was covered, right?

What do you do about special directories such as /mnt and /proc and /dev?

Any special requiredments for mysql databases (they might be in use during the backup procedure!)?

I would be very grateful if you could clarify this.