are these durations a business requirement? If so, you could use an attached timer boundary event (which could contain these values directly) and define what would happen if the timeout hits.
Or the other way around: The table can be edited by the business unit (in a DB or an Excel sheet or…), and then the process would read it at start and set the boundary event timers according to it.
I fully agree with @jonathan.lukas that the reaction to the timeouts should also be part of the process model.
Once the timeouts are hit, a mail should be sent informing that the process has not concluded in the stipulated time.
Ok, the embedded process with the timer boundary event works for the SLA of the whole process, but how can I configure the timer boundary event to get the values from the parametric table , using execution listeners?