website back up

[dvtimes]

Do you have your servers back them selves up?

Just got mine to back up. Its a one of payment of about $90.

I think its woth doing to save the hassell.

Do others do this?

[ApacheAnderson]

Where are you servers backing up to?

[dvtimes]

A hard drive.

[-HF]

Where are you servers backing up to?

the D: partition of the HDD.

[dvtimes]

The harddrive is part of the server or somthing.

its an extra hardrive.

[dvtimes]

What i mean is the server is backed up in case it crashes or whatever.

just saves me uploading 80gigs or more of data (as it would take ages).

[ApacheAnderson]

What i mean is the server is backed up in case it crashes or whatever.

just saves me uploading 80gigs or more of data (as it would take ages).

Well if it's on your server then it's only going to protect you from a Disk failure.

You need to have it backing up to another off-site server as well if you're really worried.

I have 2 servers that backup to each other, both have RAID 1 disk arrays and I download the important files to my desktop once a week.

[gawdi]

Check that they can do a restore too........ all very well paying for backups but if theres a fault somewhere (as happened to me) then your backups are worthless..

[topclass]

Yup I do, I learned the hard way, my server was attacked a while ago and the little cunts deleted everything.

I was a cheap arse and started off with static sites so didn’t see the point. But learned the hard way and ended up loosing some sites, like toplists which is a shame.

At the moment my host backs up once a week and that’s about $10 extra, well worth it. However as my sites grow a little more I can see I will have to get mirrored disks or daily backups.

[Rosie]

No idea about the techie specs, but my host, serverprovider.com back up our sites daily for us I believe. Which is invaluable when some daft bitch webmistress deletes pages accidentally from both her own hard drive and the server :noway2:

[redwhiteandblue]

No idea about the techie specs, but my host, serverprovider.com back up our sites daily for us I believe. Which is invaluable when some daft bitch webmistress deletes pages accidentally from both her own hard drive and the server :noway2:

That's impressive - it should take a PhD in numptyness to be able to do that.

[Rosie]

That's impressive - it should take a PhD in numptyness to be able to do that.


Sure does. It takes a special kind of moron to delete index.htm from both, doesn't it? :blush:

[Jase]

Dont mix up backup, replication and redundancy - it all comes down to what you want to achieve.

If you want to protect against hardware failures of components like disks etc then technologies like RAID are ideal - RAID wont protect against things like deleting a file accidentally or losing a whole server or datacenter. Its also useless at being able to roll back from a virus attack for example.

If you want to protect against a server failure (whole box or something like a motherboard failing) or a datacenter failure then you can either use some sort of replication to another box or traditional backup (to tape or better still to disk which will give quicker restore times). The benefit of replication being it can be real time so the data loss in the event of a failure is minimal - this might not be applicable to most here as even though sites are fairly dynamic they arent changing every second like an online shop would be etc where an hour of lost data equals potentially hundreds of lost transactions. The downside to traditional backups are they happen on a schedule so if you have a failure just before the backup is due to start then you lose the data from the last time you backed up till that time - again might not be a major issue especially if backups are daily. As I said before backup to disk (while no quicker than backup to tape) is more reliable and a LOT faster when you need to do restores.

The ideal scenario for bigger sites would be load balanced front end web servers with redundant components (in order of likelihood to fail those being disks, power supplies, fans, network cards then memory) and then possibly clustered backend database servers if applicable - every night do a daily backup to tape (make sure the tapes go off site!!!) and if possible also to local disk.

The next level would be to also have some sort of replication to another datacenter (doesnt need to be real time for most people in this business though I know there are a few exceptions). Ideally you would want to have manually instigated automated failover to save all the hassle with working over different subnets etc when you have to reconfigure DNS etc.

While its the utopian scenario, you dont really need to have the same number of servers on the DR site - in the event of a true DR you need to realise that you will have some loss of service and that performance etc will suffer and plan accordingly. Obviously dont have ten servers being hammered on the main site being replicated to one somewhere else or it will fall over and die once you fail over!

As with everything test it and test it regularly - test your backups work, test you can do a complete restore, test that your replication works and that if you failover it works and how long it takes.

Finally - Im hungover to fuck and have no idea why Im writing all this though this is my proper job!! Im sure I have missed something but if I have I will add to it at some point - probably when my I find my head again....... :cheers2: