Robust and efficient algorithms for storage and retrieval of disk based data structures

Data sets are often too immense to fit completely inside the computer's main memory and must instead reside on disk. If data set will be kept in main memory it will be very costly. A computer must retrieve required data and place it in internal memory to process it. Efficient data structures, like B-tree, B+ tree, are used to process large datasets. Nodes of these data structures are buffered in memory using data structures like array, linked list and trees. In this paper, we have proposed an algorithm which provide better results in case of disk based data structures which uses the concept of immediate modification i.e. when modification will be done in buffered structure (present in memory), an immediate modification will be done on disk. When the dataset size is large the CPU processing time is the crucial factor. Although, in case of disk based data structures the processing time is much less but if we try to reduce the processing time of CPU it will also gives beneficial results. Number of maximum disk accesses depends on height of B-tree which is O(log
_blocksize/2
n) [1], there is no change in the number of disk accesses. But the CPU processing time in case of search, insert, delete operation have been modified, reduced to O(log n) from O(n). The modified algorithms of B-tree are also provided in this paper.