Hi guys, I think I am missing something when it comes to jobs and billing.

I feel like recently some of my job runs have started to have quite high difference between the time each step takes and the overall billed time and I am wondering where that might be coming from? It is quite obnoxious considering both inconsistence in load time and billing. 

This used to be usually few seconds, which I assumed was like time between api-calls, inicialization between components and what not, but lately it got kinda unreasonable in dozens % of overall time jobs take. This bigger overheat is also mostly happenning in only this one job, which is nothing complicated and with generally pretty low amounts of data flowing through.

You can see such cases in the pictures.