Name resolution method for a distributed data base management system

Abstract

At start-up, input/output bus is scanned to detect any unprocessed I/O devices and each device is scanned to locate any unprocessed data base partitions. If any are located, these partitions are stored in memory and the partitions are marked as processed. During run-time, a message is received indicating that a new I/O device has come on-line. The new I/O device is scanned for any unprocessed partitions. The information regarding the partitions is stored in memory and the partition is marked as processed. The process also functions during data base operations to make connections between a client and a data base. A sequential process routes a connection request message through one node at a time around the network until all of the tables requested have been connected. If the message is returned to the originating node with out being completed, a failure is reported to the client. A parallel process broadcasts the connection request message to all of the nodes on the network simultaneously. Each node will then make local connections with any requested tables residing on that node and reply to the originating node. Once all of the replies are received, the originating node will collate the replies and assign tables to the client. In the event of duplicate table names, the rejected ones will be issued a disconnect.