Abstract: A method and system are provided for continuously maintaining replicas of an active database in a backup system for disaster recovery purposes. Redo records transmitted from an active system are received into a dataspace work area in a backup system memory. Redo records in the work area for an uncommitted database transaction are grouped together. When a transaction becomes a committed transaction, the redo records for the transaction are sorted with redo records from other committed transactions according to database, block number within a database, offset location within a block, and sequence of occurrence. A plurality of update blocks from a backup database are read into a buffer in the backup system memory. The sorted redo records are sequentially applied to corresponding data records in the update blocks. The update blocks are then immediately written back to the database.

Abstract: A system for processing diverse and non-diverse result sets is described. The system includes a client process in a client computer system that constructs a client process execution request identifying a stored procedure and specifying constraints on the quantity of response data that the client process is capable of handling. The client process execution request is transferred from the client process in the client computer system to a server process in a server computer system. The server process invokes the stored procedure identified by the client process execution request. Execution of the stored procedure results in the generation of answer set data for a plurality of diverse query result sets. Result set generators obtain answer set data for X of the query result sets, where X is defined by the constraints. The server process then generates an initial response containing, for each of the X query result sets, an amount of the obtained answer set data consistent with the constraints.