07-24-2014, 05:41 PM
AVL vs Binary tree !!
They are different in that an AVL tree is implemented with memory-based solutions in mind, while a B-tree is implemented with disk-based solutions in mind. AVL trees are not designed to hold massive collections of data, as they use dynamic memory allocation and pointers to the next block of memory. Obviously, we could replicate the AVL tree's functionality with disk locations and disk pointers, but it would be much slower because we would still have a significant number of reads to read a tree of a very large size.
When the data collection is so large that it doesn't fit in memory, the solution is a B-tree (interesting factoid: there is no consensus on what the "B" actually stands for). A B-tree holds many children at one node and many pointers to children node. This way, during a disk read (which can take around 10 ms to read a single disk block), the maximum amount of relevant node data is returned, as well as pointers to "leaf node" disk blocks. This allows retrieval time of data to be amortized to log(n) time, making the B-tree especially useful for database and large dataset retrieval implementations.
They are different in that an AVL tree is implemented with memory-based solutions in mind, while a B-tree is implemented with disk-based solutions in mind. AVL trees are not designed to hold massive collections of data, as they use dynamic memory allocation and pointers to the next block of memory. Obviously, we could replicate the AVL tree's functionality with disk locations and disk pointers, but it would be much slower because we would still have a significant number of reads to read a tree of a very large size.
When the data collection is so large that it doesn't fit in memory, the solution is a B-tree (interesting factoid: there is no consensus on what the "B" actually stands for). A B-tree holds many children at one node and many pointers to children node. This way, during a disk read (which can take around 10 ms to read a single disk block), the maximum amount of relevant node data is returned, as well as pointers to "leaf node" disk blocks. This allows retrieval time of data to be amortized to log(n) time, making the B-tree especially useful for database and large dataset retrieval implementations.