The files xxx_2500_xxxx contains the 2500 most requested file for a period, while the xxxx_dir_count_xxxx contains statistics on a directory basis, counting the number of total requests and the number of different file requests for each directory requested.
-Note that the directorycount is not accumulative, i.e. it chows only the number of requests for files in the actually directory withoud regard of requests for files in subdirectories.
-Also remember, that althoug the statistics are supposed to be correct, theres no guarantee for this due to the need of automatic generation of the files.