APM software (application performance monitoring). It was super expensive, pain in the ass to maintain, required a lot of training. But when someone skilled sat down with it - just wow. You could investigate particular users problem, and see that when she typed "s" in search form it triggered a few proxies, sql query which had proper execution plan, but it seems that connections pools are handled poorly.
It could visualize what is the actual architecture (not what we think it is) and show which connections are laggy, or more used, when it should be round robin.
It discovered undocumented connections and could show us laggy requests even if remote system was not monitored by APM - purely based on data from one side.
I could report to developers a particular line of code that is problematic from performance perspective (like "this takes 40% of time of the request, even though it's the simplest task in the process) without knowing much about that program or even coding in general (I'm more of an admin).
And for a desktop app I guess it's just called a debugger :> Unless said desktop app still uses http requests or something similar to webapps that can be attached to APM software. Then APM will know were it came from, but you won't get insight into the app itself, just the resulting connections.
It could visualize what is the actual architecture (not what we think it is) and show which connections are laggy, or more used, when it should be round robin.
It discovered undocumented connections and could show us laggy requests even if remote system was not monitored by APM - purely based on data from one side.
I could report to developers a particular line of code that is problematic from performance perspective (like "this takes 40% of time of the request, even though it's the simplest task in the process) without knowing much about that program or even coding in general (I'm more of an admin).