CPQ Data table bulkification and performance

We have a requirement in CPQ to create a rule engine which takes in input some parameters and returns which products/services are allowed. These rules are only dependent on the input parameters, so given the same input parameters the output will be the same.
The solution requires to use a CPQ data table which can be interrogated via two possible paths:

1. inside a transaction, via an action/triggers that calls a BMQL function
2. in headless mode using the Get Data Table Data API. In this mode, we want the data table performance to handle at least 8k queries per hour