ABSTRACT
Strengths: B-trees are designed for collections that are so large that the structure itself (not including the elements) cannot fit in main memory. When the data structure must reside in secondary storage, the dominant cost is the number of disk pages that must be accessed to locate the desired element. The B-tree is a generalization of a binary search tree designed to minimize the number of disk pages accessed by increasing the branching factor for each node. For the same n, increasing the branching factor makes the tree wider, and therefore reduces the depth of the tree. The branching factor is selected so each B-tree node fills a significant portion of the disk page, so each disk access yields as much relevant information as possible.
