Linux Backup Utility for Incremental Backups

And why are you not considering git itself? The strategy you describe, after one full and two incremental backups, has it's complications when you continue. It is easy to make mistakes, and it can get very inefficient, depending on the changes. There would have to be a kind of rotation, ie from time to time you make a new full backup - and then do you want to keep the old one or not?Given a working dir "testdir" containing some project (files, and subdirs), git makes by default a hidden . git subdir for the data. That would be for the local, additional version control features. For backup, you can archive/copy it away to a medium or clone it via network. The revision control you get (without asking for) is a side effect of git's differential storage. You can leave out all the forking/branching and so on. This means you have one branch called "master". Before you can commit (actually write to the git archive/repo), you have to configure a minimal user for the config file. Then you should first learn and test in a subdir (maybe tmpfs). Git is just as tricky as tar, sometimes.Anyway, as a comment says: backing up is easy, hard part is the restoring.Disadvantages of git would be just the small overhead/overkill. Advantages are: git tracks content and file names. It only saves what is necessary, based on a diff (for text files at least).I have 3 files in a dir. After git init, git add . and git commit I have a 260K .git dir.Then I cp -r .git /tmp/abpic.git (a good place to save a backup:). I rm the 154K jpg, and also change one text file. I also rm -r . git. Before restoring the files I can get the precise differences:Here I want to follow the git restore hint.After git --git-dir=/tmp/abpic.git/ restore *:The jpeg is back, and text file btext has not been updated (keeps timestamp). The modifications in atext are overwritten. To reunite the repo and the (working) dir you can just copy it back. The files in the current dir are identical to the .git archive (after the restore). New changes will be displayed and can be added and committed, without any planning. You only have to store it to another medium, for backup purposes. After a file is modified, you can use status or diff:And just like git knows about "more" in file 'btext', it will also only store that line incrementally. After git add . (or git add btext) the status command switches from red to green and the commit gives you the info. And you can really get at the contents, somehow:And then the first 4 hex hash digits To travel back in time by one commit it is:btext's blob has a different hash before the last commit, the others have the same.An overview would be: Instead of manually timestamped tar files you have commits with a message and date (and an author). Logically attached to these commits are the file lists and contents. Simple git is 20% more complicated than tar, but you get decisive 50% more functionality from it. I wanted to make OP's third change: change a file plus two new 'picture' files. I did, but now I have:So what did that Your Name Guy do exactly, in his two commits, shortly before 6 pm?The last commit's details are:And to check the second-to-last commit, whose message announces two pictures:This happened because I tried git commit -a to shortcut git add . , and the two files were new (untracked). It showed in red with git status, but as I say git is not less tricky than tar, or unix."Your debutante just knows what you need, but I know what you want" (or the other way round. Point is it's not always the same)

1. How to fix Time Machine's "This backup is too large for the backup disk”?

This happened to me recently; I was pretty annoyed because I expected that if there was not enough space, Time Machine should just delete the oldest backup.The problem turned out to be that the disk contained only one (large) backup, and did not have room to store the current delta (delta = changes to files since last backup).I realized this when I looked at the disk contents and saw that there was only one backup folder under Backups.backupdb. I used Disk Inventory X to look at the contents of my hard disk, and identified a few large folders which did not need to be backed up. I excluded these from the backup using "Time Machine" > "Options" > "Exclude these items from backups". The info from Disk Inventory X also led me to delete a bunch of large unused files.After the two changes listed above, the backup proceeded successfully. I did not have to launch the backup, it automatically retried. I presume that my changes brought the delta to a small enough size that it fit in the remaining free space on the backup disk.UPDATE: This happened to me again recently, and just excluding folders from the backup was not enough, so I found another trick. Let's say you analyze your hard disk contents and realize there's a 20GB folder (let's call it UselessStuff) which was being backed up. You exclude it from the backup as described above, but that does not delete the previous backups of UselessStuff from the disk, and you still do not have enough space for the backup to complete. Since you do not need any backups of UselessStuff, you can enter Time Machine, right click on UselessStuff, and click "Delete all backups...". In my case, this freed up enough space to let the backup continue successfully.

2. Backup large files in an external disk

Addonics manufactures a range of HDD duplicators that duplicates hard disks very fast without any connected computers. From naked disks, their solutions also support USB-to-USB copy, etc. Do check it out.We have one in the office, limited application, but it is the most frictionless way to duplicate large files (in fact, the whole disk) to another.

3. MSSQL Backup Question

There are two ways people backup databases. One is to do dump of the database (i am not a dba so i am not sure the mechanics of doing the dump) to a text file, then write that file to tape. The second is to use an agent that is aware of the RDBMS that you are using. Both of these methods will get you everything you need to restore the DB to a working condition. You can use backup exec, but you will need to make sure that the MSSQL Agent is installed on those systems. Since you have more that 250 server's i am guessing you have at least one dba on staff - I would ask them as to how backups of the SQL servers are currently being done. They should know some of the basics at least enough to get you started. They would know things like whether your predecessor used an agent, or backed up text dumps