Originally Posted by DeanM
But if they have 50TB of movies, then certainly they had the money to purchase all those movies on DVD/BD (unless they are pirated??).
If they had the money to purchase the movies, then shouldn't they have the money to do 1:1 backup??
My backup is weekly snapshot as well. No fancy backup software, just robocopy... such that restores are easy using standard file copy.
And yeah... Online backup services for large collections is just not feasible. Both in cost and, as you allude to, bandwidth efficiency issues.
As others have alluded to.
In a software RAID setup where drives remain in native format, the chances of having more than 2 hardware failures at the same time are rare. Rare enough that in the event that you lose some actual data, that the job of reacquiring (via download/ripping/etc.) the lost data should only be a portion of that 50GB+ collection.
So lets run some hypothetical numbers.
50TB collection of ripped movies and music. Lets say that is on a Norco 4224 with a mixture of 3TB and 2TB drives. So at the current going rate of ~ $60/TB works out to be $3000 in hard drives, add in the extra $1000 for hardware and you are looking at $4000 additional base cost to mirror a 50TB collection, then add in electricity etc. and it only goes up. So in order to maintain this mirrored 50TB collection, you are spending in excess of $8000.
Compare that to a 50TB collection with say 5TB worth as parity so a total of 55TB in drives for a cost of $3300 + $1000 hardware and you are looking at a bit over half the cost for protection from two drive failures and in the event that 3 drives fail, you have to rerip say 2.5TB worth of stuff again. If not having to re-rip 2.5TB worth of content is worth > $4000 to you, then sure, by all means go for it.
Then there is the issue of maintenance. Having two arrays mirrored so as to provide a true 1:1 backup means that you have to copy to both locations, lets say that with a collection this size, you are likely to invest some time into setting up some form of sync job between the two.
Lets say you have this sync job to run every night (which means that for 23 or so hours of the day, you don't technically have a 1:1 backup anyway ... but I digress)
Somehow a program on your machine wipes out a bunch of movies on your "live" server (e.g. my movies software bug mentioned before)... if you don't notice this before the sync job is run, you will lose those movies from your "backup" server as well - unless you set your machines to only sync additions and not deletes/updates. In which case revisit the numbers before and account for a 50TB server being backed up with an even larger backup server.
Scenario number 2... same as above but you lose one drive of data due to hdd failure on your live machine. You replace the drive and kick off a manual sync job to restore the missing content. Unfortunately, data rot has taken place on your backup server and you replace some perfectly good copies with some silently corrupt ones. OR a drive on your backup server fails whilst the data is syncing, so now you have to manually determine which data you have that is good on which server and replicate across both servers.
I know that what I have outlined sounds ludicrous and unlikely to happen, but then it is only slightly more unlikely than the scenario of having 3 drives lost in a 50TB dual parity server failing.
TLDR; only people with small collections can consider 1:1 backup of movies/music to be a necessary endeavour, that or they have too much money to spare. For most other people, even those who have money, the value of a dual parity setup for media and 1:1 (or even more) backup of photos/documents etc. is more than adequate.