I've written papers where the raw data was small enough that it could be put into tables and included within the online supplements. This is pretty much ideal.

I've written papers where the raw data was small enough that it could be put into tables and included within the online supplements. This is pretty much ideal. Unfortunately, a lot of experiments generate enough data that no publisher will store it for you. Given what a racket the scientific journal business is and that scientists pay thousands for each publication, it really should be expected that journals will store and curate any data pertinent to a paper they are paid to publish.