In terms of repositories, Peter Murray-Rust suggests that there are at least two differences. One might be the underlying technology (are the usual EPrints, DSpace, and Fedora the best choices for digital objects that aren't research papers?), the other is whether access can be open or not.
After posting his abstract for the opening keynote of Open Repositories 2008 on his blog, I commented on my surprise that he hadn't mentioned Open Data. This led to a a few more exchanges where he emphasised his view that data repositories based in institutions must focus on the needs of the data creator (scientist).
See
http://wwmm.ch.cam.ac.uk/blogs/murrayrust/?p=996What does “Open” mean in Open Repositories? - March 7th, 2008
Should Data Repositories be Open? - March 6th, 2008
http://wwmm.ch.cam.ac.uk/blogs/murrayrust/?p=994Repositories for Scientific Data (at OR08) - March 4th, 2008
http://wwmm.ch.cam.ac.uk/blogs/murrayrust/?p=992and also his brief observations about Australian developments (in between the others).
Robin Rice