Disaster recovery -v- regular backups

Random

regi in another thread wrote:

Do you want backups, or disaster recovery? Backing up your files and your database with a maintenance plan is easy enough - imaging your server for DR is a bit more complex.

So what's the major difference? One basically means that you have all your files available, but you'll need to reconfigure your OS/System etc if **** hits the fan.

And the other means that you can litterally turn it on and it'll be like nothing ever happened?

aidan_walsh

One basically means that you have all your files available, but you'll need to reconfigure your OS/System etc if **** hits the fan.

Thats essentially my understanding of it. OS files generally aren't included in backups. Disaster recovery would mean creating a full snapshot image of the server at a given point (much like a Ghost image), which you would be able to restore in the event of an otherwise irreperable OS failure.

regi

I'd see DR as a one-click way to get my server/system back to exactly the way it was when the last DR backup was taken.

Regular backups, I'd see as file or database backups, used for playing around with, restoring particular bits (as I did with the boards database today ) and for long-term storage.

DR backups, particularly on a Windows server, can be _huge_. Regular backups might be tiny. Sometimes, it makes a lot of sense to do a DR backup on a weekly or monthly basis and then regular backups very regularly inbetween. Thus if the proverbial hits the fan, you can do a DR restore and then a data restore to the latest cut.

Ruu_Old

So the DR backup is something like system restore for Windows or something?

Random

Seems to me to be like having a completely separate server sitting right next to it (or in a safer location) that can be plugged in with a moments notice with only minimal or no downtime/effect?

regi

Or a bare-metal backup product, or taking system images with something like Ghost. You could maybe even class running one vmware instance on one server, and then backing up that image as one.

Capt'n Midnight

Ruu wrote:

So the DR backup is something like system restore for Windows or something?

Automated System Recovery can be used in XP or Windows Server. BUT It wipes the whole drive ( or just a partition I can't remember ) and only restores windows so it's main use is to restore the OS to the point where you can use Ntbackup to restore everything else or install your [strike]favorite[/strike]least hated backup software.

http://hacks.oreilly.com/pub/h/1196

Tom Dunne

ciaranfo wrote:

Seems to me to be like having a completely separate server sitting right next to it (or in a safer location) that can be plugged in with a moments notice with only minimal or no downtime/effect?

That would be a hot-standby.

phil

Disaster recovery is more of a "plan". It's a number of "what ifs" and you implement different measures depending on what goes wrong. They should be written by an individual or group of individuals in your organisatin who would be equivalent to "system administrators". They're an essential part of system administration.

Backups are a strategy to help in the event of a DR plan being activated. They're complimentary to each other. Again, "hot standbys" are part of a DR plan. Plenty of places I've seen have crazy DR strategies (well crazy as in they cost a lot of money) one of the most impressive I've heard was hot standbys being available in a mobile truck.

Phil.

jesus_thats_gre

Surely you would have an image of the OS drives and seperate backups of the drives containing the database etc..

Depending on what happens you would take appropriate steps to recover from a failure. To be honest, backing up the OS and the database in the same dataset just baffles me....

skywalker

regi wrote:

I'd see DR as a one-click way to get my server/system back to exactly the way it was when the last DR backup was taken.

One click? seriously?

regi

No, not really

But something predefined, so that the restore could be very easily performed.

phil

DR is normally a "strategy" for larger corporations. In the event of this, do this. In the event of this, do this. etc.

For smaller places, it's normally "make backups. Restore if necessary. If not possible to restore, run around like a headless chicken."

One-click DR solutions are common, but moreso with bigger corporations. Again, it depends what happens and what part of the strategy you execute.

Phil.