PROJECT TITLE :

Advanced Block Nested Loop Join for Extending SSD Lifetime - 2017

ABSTRACT:

Flash technology trends have shown that greater densities between flash memory cells increase read/write error rates and shorten solid-state drive (SSD) device lifetimes. This is important for enterprise systems, inflicting such issues as service instability and increased total cost of ownership (TCO) as a result of of SSD replacement. So, various studies have centered on decreasing the number of the DBMS writes. But, there has been no analysis that centered on decreasing the number of temporary writes, which are primarily created by be a part of processing. In DBMSs, there are two major be a part of-processing algorithms, i.e., hybrid hash be a part of (HHJ) and kind merge be a part of (SMJ), proven to be the simplest consistent with DBMS workload; but, the 2 algorithms produce temporary writes of intermediate results. Thus, we tend to instead look to the block-nested loop be part of (BNLJ); it is well-known that the two algorithms are better than BNLJ, but BNLJ creates no intermediate result writes. It's reasonable to use BNLJ for a serious be part of algorithm if its performance will be enhanced kind of like those of HHJ and SMJ, considering BNLJ's advantage of extending SSD lifetimes. Thus, in this paper, we tend to propose an advanced BNLJ (ANLJ) algorithm which will match the performance of the 2 main join algorithms.