Recently at work i’ve finally gotten around to sorting out a backup scheme.

This make sense, as we have 200+ macs each with anything from 500mb~80gb of data on that need to be backed up, usually in classes of fifteen or so… it’s been hard for me to believe that there hasn’t been a backup solution in place. Up until now it’s been down to the students and the staff to make sure that they back up their own work which is [1] good, because this teaches the kids to look after their work and [2] bad, because in a production environment, this will be taken care of for them, and they don’t really understand how to, or why.

First I looked at upgrading a [high priority] class from Tiger to Leopard, and Time Machine, to allow them all to have a completely un-manned backup, something that takes care of itself. Later I found that Time Machine doesn’t handle backups over AFP/SMB too well, even with hacks running. And buying 500gb external drives for each machine is out of the question.

Enter Rsync, I’d tried a couple of different approaches, one being Psyncx, which doesn’t like leopard much at all, and Rsyncx, which started shitting all over the place at random intervals. Instead I tried a front end wrapper to Rsync which makes things very easy, and after setting up a few backup schemes at home over AFP shares; it works perfectly!

This my friends, is the night i’ve finally bashed down a demon that’s been haunting me for a few months.

In all of this I’m aiming to solve the backup solution, but at the same time, my own objective is to educate the staff involved, and in turn the students themselves in how to handle things like this, and in a limited way, how they work. The solution will pass through from end user, all the way up to myself and then up onto the main cluster’s backup – but along the way, the teaching and support staff involved will have access to be able to use the backup solution on a day-to-day basis should they need it.

So, the current layout is, fifteen-to-twenty macs per room, in each room is a teacher’s machine, that isn’t anything special, same spec as the student machines. We have some spare G4 gigabit Powermacs laying around that i’m going to turn into staging servers in each room – truth be told, anything from a stock ATX windows or linux server can be used, but the G4’s have enough network grunt and a pretty face too, it’ll be good to put them to use, as there’s not much they can handle in the classrooms these days.

at 4pm the students leave, and the machines log out the current user ten minutes later

at 4:30 the student macs [depending on network load and amount of data to shift] will singly, in pairs, groups, or all at once, start copying designated folders and their contents over to the staging server in their room. Running off their own gigabit switches, the different departments shouldn’t slow each other down.

This will happen each evening before lights out at 6pm~9pm depending on the evening’s activities on site.

On friday evenings, the staging servers will copy all the backups that they’ve received during the evenings up onto the cluster, at first this is going to sting and take a really, really long time, but after the first sync, the backups will only copy up new files and folders that have been created.

The bit I have to work out now, is how i’m going to set quotas, per-machine, and per-room/department.

of course, i could just get a Terrabyte worth of Time Capsule for each department – the trade off would likely be control over ease of use, plus a big safety net in the shape of a warranty! maybe i should talk the boss into letting me buy one of the 500gb models for a test run…

we shall see how it goes