You don't need a rate limit if everything is running smoothly. Have you experienced any issues? You probably have the API_INT semaphore pools using the the default 4 simultaneous threads. How many nodes do you have? 2 million transactions against 2 nodes (8 API_INT semaphores) would sound like the system would benefit from somekind of baseline analysis. Are there some bottlenecks and if, where are they?
How closely have you monitored the performance of the integrations? If you do not monitor and measure the performance, it is hard to predict what way you are going with the performance, capacity and usability. Below is an example from an instance that is running a hundred integrations, 120 Midservers, a large Event management installation and gets about 175M API-calls per month. I have been monitoring the system/integrations since day one and we have no real issues with API-capacity or performance. Altogether I have +200 reports, scripts and several dashboards that monitor the data. Currently we are running the system on 12 application nodes.
BR, Jan
