Can’t disclose how it will work. But early tests indicate that if you take a reasonable sized RAR archive compressed to the max and place it in my container, you should expect to achieve a further 15-17% compression.
Now the plan is to finish the library and post the service as a website where you can compress and uncompress a single file. I will probably put a limit on the file around 50 Mb mark. This will allow me to gather vital statistics (without killing my server).
If all checks out I will add support for multiple files and release it as open source. Still thinking about the last step though 😛