No data loss system with reduced commit latency

A data storage device characterized in that:

• the data storage device contains a redo log file to which a processor in the standby database system which has access to the redo log file writes items of redo data belonging to a stream of redo data that the standby database system receives from a primary database system, the processor sending an acknowledgment of the receipt of an item of redo data to the primary database system prior to writing the item of redo data to the redo log file and the redo log file comprising:
• items of redo data that have been received in the standby data base system and written to the redo log file;
• and
• a no data loss indicator that indicates after an end of redo event whether there are gaps in the redo stream that was being written to the redo log file, the no data loss indicator including:
• a first component that indicates whether there were gaps in the stream of redo data that was being written to the redo log file prior to the end of redo event;
• and
• a second component that indicates no data loss only if the first component indicates no gaps prior to the end of the redo event and all of the items of redo data for which the standby database system has sent acknowledgments to the primary database system have been written to the redo log file.

A database system comprising:

• a primary database system coupled to a communications network;
• and
• a standby database system coupled to the communications network, each at the primary database system and the standby database system having a processor and persistent storage to which the processor has access and the primary database sending a stream of redo data items for transactions belonging to one or more sessions to the standby database system via the communication network, the standby database system responding after receiving a sent redo data item for a given session by sending a first acknowledgement to the primary database system via the communications system, and the database responding to the first acknowledgement by sending a second acknowledgement to the given session, the standby system writing the redo data items to a log in the standby system's persistent storage, the log file including a no data lost indicator that indicates that no gaps have occurred in the stream, and
• the database system being characterized by:
• an acknowledgement mode in which the standby database system sends the first acknowledgment to the primary database system prior to writing the redo data item to the standby system's log file;
• and
• the standby database responds to an end of redo event by writing any redo data items intended for the log file that have not been written to the log file thereto and, when no gaps in the stream of redo data items have occurred prior to the end of redo event, setting the no data lost indicator to indicate that there are no gaps in the stream of redo data items.