OTF queries typically consume high CPU resources as the engine requires additional CPU cycles to convert Iceberg data formats to Teradata native formats. On an idle system, a single-stream OTF query could potentially use 40-60% of the node's CPU. Therefore, it is recommended to run OTF workloads at medium or low priority. This can be achieved by creating a workload management rule with query classification on these OTF Iceberg functions:
- TD_OTFDB.TD_ICEBERG_READ
- TD_OTFDB.TD_ICEBERG_WRITE
- TD_OTFDB.TD_DELTA_READ
- TD_OTFDB.TD_DELTA_WRITE