... tar is used to create a "collection" of files. It's the equivilant of packing files into a briefcase. Tar was invented/created in order to store multiple files on tape, since tape works best with contiguous files/access. You'd stream a tar off the tape and dump the files out of it onto your disk subsystem. So, in other words, tar performs no compression. There is very very slight overhead for the tar.
Bzip2, of course, compresses the file. If you're going for maximum compression, then bzip2 -9 blah.tar will work best. However, bzip2's disadvantage is the extreme compression time. Most of us that frequently deal with compression/decompression stick with gzip for this reason alone. If you're archiving for long time storage/don't plan to compress/uncompress often, go with bzip2. If you're going to be doing this often, go with gzip.
As to compression ratios, it's going to vary based on the file types as mentioned by a previous poster. Text will compress heavily, pre-compressed source material such as mp3s will not. You can "guess" the compression ratio based on past experience/knowledge of what you are compressing, but you cannot determine an exact compression ratio without actually compressing the data. That's why bzip2/gzip do not have an option to inform you of the compression ratio prior to compressing. :)
It's generally good form to pack multiple files that are associated into a tar, then gzip/bzip2, simply for "good housekeeping" reasons. The very slight overhead of tar is worth the ease. It also makes tape backups less painful - random access is NOT something tapes are well known for (to put it lightly.) :p |
