PARTITIONING METHOD FOR HIGH-SPEED BLAST SEARCH BASED ON A PC CLUSTER, CAPABLE OF SUPPORTING OPTIMUM SEARCH FOR A DIVIDED DATABASE

摘要

PURPOSE: A partitioning method for high-speed blast search based on a PC cluster is provided to store a rank database in a master server to access from each PC node to an NFS, load a partition of the database into a memory, and distribute a query to PC nodes in cluster middleware to arrange the database, thereby constructing a cluster computer with high performance for blast processing with an inexpensive PC.;CONSTITUTION: Cluster middleware(110) is composed in a server(100). A rank database(140) is arranged in the server. A PC node(120a~120n) accesses an NFS(Network File System)(130). A divided database is arranged in a cache memory of each PC node. The cluster middleware distributes a user query to the PC node. The cluster middleware integrates the execution of search with a search result. Each PC node transmits the search result to a master server.;COPYRIGHT KIPO 2011