Why are the data files divided into primary and secondary directories?
The data products are arranged such that the products necessary for most analyses (e.g. the CIAO threads) are in the primary directory. If you are interested in reprocessing your data, you will also need the level 1 files, which are in the secondary directory. The Introduction to the Data Products thread contains details on each type of file in the data distribution.
CIAO does not require the files to be separated into the "primary" and "secondary" directories in which they are distributed. Users may rearrange the files in whatever manner makes sense for their analysis process.