The IT Gods of our company have implemented some really serious backup and security measures. Something called Bit9 randomly kills my compilation process, the Mozy backup locks the file right in the middle of a git pull --rebase origin master
resolving merge conflicts. The damned backup system takes two cores out of my 32 core machine 24/7 constantly. It backsup debug logs and regression suite temp files every 30 minutes. My unit testing script churns through 10 GB of scratch files every night. All I care about it is the the line "Unit test run: 12474, passed 12474, 100%". But all night long Mozy has been backing up the scratch files dutifully. Where it is storing it, how much it is charging our company
.... god only knows.
It is so bad that no developer in our company has used Mozy to recovery anything successfully. You launch the recovery dialog, wait for it to populate the file tree and recovery sets. It spins and spins and spins. We give up. It seems to be some tool meant for office application, some spreadsheets and documents all manually created. How many documents can you physically type up every day? It works at that load. Deployed on a development work station, that downloads some 45 GB of source code in some 20 repos, with daily pulls and rebases, developers maintaining multiple views, running regression suites and unit tests, we easily generate 10 GB of data per day per developer spanning easily 20000 machine generated files. All I touched were three source files, and approved one pull request, boom 10GB!
Yahoo emails are all manually typed. It does not have to contend with this level of machine generated derived objects and data files. But it is not dealing with a 200 developers, it is dealing with several million users. They could be churning through this much of data in their image files and video/audio clips. It will take significant effort and cost to recover anything from backups.
There is however, one developer who actually found something very interesting. We don't have wait for the recovery sets dialog to populate. If you know precisely the entire absolute path name to a file that was deleted you could type it as the wheel it spinning and it can be found relatively easily, he said. So yeah, if you know precisely the name of the blob that want to recovery, may be you can. But to go on a fishing expedition finding all files that existed on a particular day it well nigh impossible.