Show me your data: Institutional Repositories

There has been some move in research to not just publish papers with the final results but to also release the raw data sets and even software for other researchers to verify the results and further discovery. There are even some futuristic claims that the data sets will be viewed as the ultimate results of research and the actual paper will be a secondary product.

Like to discuss what peoples experiences have been with Institutional Repositories. Has it been to showcase work, preserve for the future or to play an active role in furthering discovery?

If you have created or maintained an IR, what issues did you face, how well accept was the IR by the researchers. Does data set sharing play a role.