Although tar cJf archive files... as detailed by Zacharee1 and by heemayl is usually what you'll want to do, another way is to pipe tarred data to the xz command:

tar c files... | xz > archive.tar.xz

Since Ubuntu's tar supports the J option, this alternate way is specifically useful when you wish to pass options to xz.

In this example, I tar and xzip some TIFF files with a high level of compression (-9 to xz) and verbose output (v to tar, -v to xz):

ek@Io:~/Pictures$ tar vc *.tif{,f} | xz -9v > pics.tar.xz

page001.tif

page002.tif

page003.tif

page004.tif

page9087.tif

page3la.tiff

quux0000.tiff

  100 %       207.3 KiB / 290.0 KiB = 0.715

This could, of course, also be done in two explicitly separate steps:

ek@Io:~/Pictures$ tar vcf pics.tar *.tif{,f}

page001.tif

page002.tif

page003.tif

page004.tif

page9087.tif

page3la.tiff

quux0000.tiff

ek@Io:~/Pictures$ xz -9v pics.tar

pics.tar (1/1)

  100 %       207.3 KiB / 290.0 KiB = 0.715

Those two ways are not actually equivalent in how they operate, though the .tar.xz files they produce in the end should be the same (and were, when I tested it).

• In the first, the output of tar is piped (|) to the input of xz. xz receives data from tar almost immediately, and no intermediate uncompressed tar file is ever created. This is to say that the first way is essentially equivalent to tar cJf archive files..., except for the additional arguments passed to xz.


• In the second, an uncompressed tar archive is created by the first command, then compressed by xz in the second command. (xz automatically deletes the original file when it's done, unless invoked with -k/--keep.)

For further reading, see this post by Rafael van Horn and the tar and xz manpages.