eXtremeDB Transaction Logging with HA and Cluster

In some cases it is desirable to enhance eXtremeDB High Availability (HA) and/or eXtremeDB Cluster applications. For example, you might want one node of a cluster, or an HA replica, to be responsible for replicating data asynchronously to an external system in the “cloud” using Data Relay.

The general algorithm to enable database persistence via Transaction Logging in C, C++, Java, C# and Python applications is to perform the following:

• try to load a database snapshot if one has been saved, or else open an empty database
• apply the log file if it exists
• save a new database snapshot
• start transaction logging
• attach the node to a cluster or initiate HA
• perform the usual node activity until the successful conclusion or a crash occurs
• stop transaction logging, detach the node from the cluster and close the application

Also xSQL is capable of implementing transaction logging within a cluster node or with HA. Simply add a TL configuration section like the following to specify the database snapshot and log files:

     
    tl_params : {
        log_path : "db_file.log",
        snapshot_path : "db_file.img",
    },
     

This set of xSQL options enables the snapshot and log file persistence in the same manner as with the embedded API algorithm described above.