I'm doing a personal project where i need to estimate the average waiting time for each customer in a queue system with s servers. In my case though, a customer can spawn its service on more than one server: imagine a parallel computing system which schedules multiprocess jobs: each job can spawn k<= s processes
So i am asking myself:
is it right to see any n-size customer as n single customers in the estimate of statistics? I.e. if there are jobs with 5 processes served per second i'd have a.service rate of 5 jobs per second, not just one
if the (1) was right, it keeps to be right despite any change in the scheduling policy?
could the size of a customer that enters the queue affect its waiting time?
I don t know where to find this information because seems that there are not so much studies upon my problem, so i want to ask if i'm wrong. Many thanks!
Maybe have a look at this lecture slides as a start. There is an example of MULTISERVER WAITING LINE MODEL on page 11, that should be helpful.