When users are allowed to alter any copy of a replicated table, asynchronous replication of changes must deal with conflicting changes introduced at diferent table copies. We discuss a high volumn, low latency replication protocol that insures that all copies converge to the same, most recent state despite conflicting changes at different table copies and the delayed arrival of replication messages between participating sites.
About the speaker:
Dr. Bruce Lindsay is a recognized expert on many aspects of Relational database technology, He has contributed to the technology of distributed DBMS, query execution algorithms, database parallelism, and system performance. Dr. Lindsay is an IBM Fellow, and an ACM Fellow.
After receiving his PhD from the University of California at Berkeley, Dr. Lindsay joined IBM Research in San Jose where he participated in the development of System R (the first Relational DBMS prototype) and R-star (a distributed RDBMS). Dr. Lindsay also worked on the design of DRDA (a remote DBMS access protocol) and was the architect of Starburst (an Extensible RDBMS). In the 90's, Lindsay led the port of the Starburst technology into the IBM DB2 products and has contributed to many aspects of the DB2 database. Dr. Lindsay was appointed an IBM Fellow in 1996 and is currently engaged in development of database replication technologies.
Contact information:
Bruce Lindsay
IBM Almaden Research Center
650 Harry Road
San Jose, CA 95120
bgl@almaden.ibm.com