Looking for Good Practices to Pair Individual Rows in a Dataset with Other Datasets

I need to pair individual rows of a compound type dataset with a
dataset of 3D points (three floats per row, anywhere from 4 to many
thousands of rows) that is unique to that row. Are there any standard
ways to accomplish this? I have considered adding a link to the
compound dataset which refers to the dataset containing the 3D points
but I would like to avoid having that information being part of the C
structure as my library seeks to entirely wrap all HDF function calls
and datatypes. I am also uncertain if there are any drawbacks to
creating so many datasets.

Thanks for your help,