Learn how to allocate RAPCache database CPU resources within a production environment.
Consider a RAP system where all of these activities are being executed concurrently:
| Activity | RAPCache database execution classes | Ideal CPU/engine resource allocation |
|---|---|---|
| A publisher is continuously transmitting market data, which is loaded into the RAPCache and RAPStore databases. The data is stored in a four-way round-robin-partitioned table in the RAPCache database. | This activity requires one user-defined execution class. | Four dedicated engines, one per partition. |
| A real-time market data analysis application, a derived-data generation application, and a derived-data modification application. The three activities periodically query the RAPCache database to obtain results. | Each activity requires one user-defined execution class. | One dedicated engine for each user-defined execution class. |
| The generated derived-data is stored in a four-way round-robin-partitioned table in the RAPCache database. | This activity requires one user-defined execution class. | One dedicated engine. |
| Transfer rules are defined in the system to periodically transfer derived-data from the RAPCache database to the RAPStore database. | This activity requires three user-defined execution classes. | One dedicated engine for each user-defined execution class. |
| Truncation rules periodically remove market and derived-data from the RAPCache database. | This activity requires one user-defined execution class, as well as the user-defined classes created for derived-data transfer. | Four dedicated engines, one per partition. |
The total number of engines the RAPCache database requires for all these activities is 14. However, if only 6 engines are available, distribute the limited resources amongst the activities to achieve optimal system performance.