The size of the dataset that can be clustered for the in-memory K-tree is limited by the size of computer memory, which on most systems limits the dataset size to about 10-20 Terabytes. However, the on-disk K-tree preforms clustering on disk rather than in memory thereby, dramatically increasing the size of the dataset that can be clustered.
/****************************************************************************************************************
the k-tree index file
HEADER EXTRAS INTERNAL NODES LEAF NODES =========|===========|====================================|================================================================== =========|FileList...|KeyAddressKeyAddressKeyAddress......|PixelKeyPixelKeyPixelKeyPixelKeyPixelKeyPixelKeyPixelKeyPixelKey.. internal node address = header size + (internal node size x internal node order) leaf node address = header size + internal nodes size + (leaf node size x leaf node order)
****************************************************************************************************************/