I don't know. I also need to store/analyse data (I wanna give python-pandas a try).
HDF5 seemed to be THE way to do it.
PyTables seems more performance oriented while h5py seems to be a much more simpler wrapper to the raw HDF5 C library:
So I was just wondering whether you have compared the two of them. Seems not, so I'll give pytables a try :)