Grid Task Staging Table

In order to minimize database spikes caused by contention on the Grid Task table during large Grid Task creation processes, Platform introduced the Grid Task Staging Table. This table is an exact match for the Grid Tasks table except it does not have the indexing, referential integrity, and most importantly, doesn't have hundreds of processes continuously pinging it for more tasks.

All Task Generators – because this logic was deferred to Platform – load the Grid Tasks Staging Table (Grid_Tasks_Stg). At the end of the process, a Stored Proc (unstage_grid_tasks) is called that bulk transfers the data from the Staging Table to the real Grid Task table. This lowers the strain on the database because it is not having to communicate back and forth with the app server – it simply copies data from one table to the other.

Task Writers can take advantage of the staging table by explicitly writing to this table and then calling the stored proc at the end of their process. This will achieve exactly the same effect. The issue is that the engine has to be coded explicitly for it.

Within the IXM Engine Config, the config must set ChunkTaskWrite to true. If it is not set or set to False, then the Grid Tasks are written directly to the Grid Task table.