Efficiency evaluation model of parallel processing in distributed environment using Split-Join queue

A model is proposed allows to evaluate the efficiency of multi-threaded processing of tasks in a distributed computing environment based on a queuing network. The processes of separation, parallel processing of subtasks and aggregation of results are modeled using a queuing system with the discipline Split-Join. The initial data of the model are: the arrival intensity of requests, the processing tasks stage duration distribution and the number of parallel streams. The efficiency evaluation of multi-threaded task processing as an additional distribution function is obtained, which characterizes the processing probability for a time not exceeding the directive period, is able to conclude if that multi-stream processing of the required data volumes is possible under specified conditions.