Techniques for the log-based replication of high-level procedures

A computer system having a processor and memory, the system comprising:

• the memory for storing a primary database, comprising a system table for storing data blocks;
• and
• the processor for executing a file system comprising code for running the primary database, wherein when executed, the code creates an annotated redo log to support replication of one or more procedures at a standby database, wherein a procedure can correspond to multiple transactions executed as a unit, the annotated redo log comprising for each procedure:
• one or more entry markers written as entries into the annotated redo log when the procedure begins executing at the primary database, the one or more entry markers comprising:
• procedure information, argument data for the procedure if any, and a corresponding transaction identifier,
• a plurality of redo records, at least one of the redo records comprising a copy of a change to a data block in the primary database and a corresponding transaction identifier, and
• an exit marker written as an entry into the annotated redo log when the procedure is done executing, the exit marker comprising an exit status for the procedure indicating whether the procedure was successfully executed at the primary database, wherein the entry markers, the redo records, and the exit marker are sent to the standby database which is configured to group the transactions associated with the redo records into a procedure based at least in part upon the entry and exit markers such that the changes of the procedure are applied or discarded at the standby database as a whole.

A computer program product including a non-transitory computer-readable medium having instructions which, when executed by a processor, cause the processor to perform a process for replication of one or more procedures stored at a primary database to a standby database, wherein a procedure can correspond to multiple transactions executed as a unit, the process comprising for each procedure:

• when the procedure executes at the primary database, writing one or more entry markers as entries into a redo log, the one or more entry markers comprising:
• procedure information, argument data for the procedure if any, and a corresponding transaction identifier;
• writing a plurality of redo records as entries into the redo log, at least one of the redo records comprising a copy of a change to the primary database and a corresponding transaction identifier;
• and
• when the procedure is done executing, writing an exit marker as an entry into the redo log, the exit marker comprising an exit status for the procedure indicating whether the procedure was successfully executed at the primary database, wherein the entry markers, the redo records, and the exit marker are sent to the standby database which is configured to group the transactions associated with the redo records into a procedure based at least in part upon the entry and exit markers such that the changes of the procedure are applied or discarded at the standby database as a whole.