High availability data replication of smart large objects

In a database including a primary and a secondary server and a high availability data replicator adapted to synchronize the secondary server with the primary server, a smart large object application program interface residing on the secondary server, the smart large object application program interface comprising:

• a smart large object read module adapted to read data from a smart large object without acquiring a lock on said smart large object, said read data being communicated to a client;
• an exception module adapted to send an exception to said client responsive to a synchronizing event of the high availability data replicator modifying said smart large object;
• an update module adapted to update said smart large object responsive to said synchronizing event;
• and
• a cache module adapted to create a memory cache of said smart large object, said smart large object read module accessing said memory cache during a read, said cache module being further adapted to update said memory cache responsive to said updating of said large smart object.

An article of manufacture comprising one or more non-volatile storage media encoding instructions for performing a high availability data replication process that synchronizes a secondary server with a primary server of a database that includes smart large objects, the process comprising:

• ordering log entries of a smart large object modifying operation performed on the primary server in a selected order wherein a log entry corresponding to updating a large object header of said smart large object is consistent immediately upon execution;
• transferring log entries including said log entries of said smart large object modifying operation from the primary server to the secondary server;
• identifying a read operation accessing said smart large object, and communicating an exception and error code to a client associated with the read operation;
• replaying said transferred log entries on the secondary server, the replaying of said log entries of said smart large object modifying operation being performed in the selected order without locking said smart large object on the secondary server, thereby insuring consistency of the data being transferred from the primary server to the secondary server, wherein the replaying of the log entries includes modifying said smart large object on the secondary server;
• and
• communicating an exception including said error code to said client.