We therefore strive to persuade application server administrators to set up Static Connection Pools, where minSize=maxSize


I agree and so does the real world performance team. We encourage customers to choose appropriate static sizes for connection pools (DRCP or otherwise).

However, from a software *delivery* perspective, we have a number of conflicting things we try to tackle:

1) some customers run on 2 cpu machines, some run on 200 cpu machines
2) some customers want to utilise (say) 10% of their overall server capacity for DRCP, others will want to use 90% of their capacity with DRCP
3) we want the DRCP to be a simple as possible to get customers up and running with it,

Because of (1) and (2), we can't really pick a (static) default based on (say) number of CPUs, because it will suit some but not all customers.
Because of (3), we don't want to have DRCP delivered totally unconfigured so that customer shave to set a bunch of settings to get it to work.

So we go with a compromise (a dynamic pool).

But as people become more familiar with it as they use it, we encourage static pool sizes for all the good reasons you're already aware of.