When you set distributed=true in a task request, the task runs in distributed mode. In this mode, the system may create multiple sub-tasks to process different portions of the data in parallel. These sub-tasks run concurrently across available API nodes (as long as those nodes are free to handle periodic tasks).

The maximum number of sub-tasks that can be created is controlled by the taskPartsCount parameter. However, this value represents the upper limit — not the guaranteed number — of sub-tasks. The Reltio platform evaluates optimal performance conditions and starts only as many parts as necessary, up to the value specified. This is to avoid performance degradation caused by excessive partitioning, especially in cloud environments such as Azure CosmosDB.