Static-dynamic algorithm for managing asynchronous computations in distributed environments

Abstract

The paper addresses a relevant problem of computation scheduling in scientific applications (distributed applied software packages) executed in distributed environments. Forming an optimal schedule of jobs for executing of applied software (modules) is an NP-hard problem. Therefore, in practice, heuristic methods of scheduling are often used. In this regard, we propose a new static-dynamic algorithm for managing computations in heterogeneous distributed environments. The results of operating the proposed algorithm are simulated in comparison with other scenarios for computing management. They show that applying the algorithm makes it possible to achieve a rational balance between the scheduling time and the computations makespan.