Framework for Simulation of Distributed Systems

With the increasing popularity of the cloud computing environment, the problem of scheduling task graphs for execution within the cloud environment has become relevant. To develop and analyze different scheduling algorithms, you need a simple and fast way to measure the quality of the scheduler —— simulation of the environment. The cloud as a platform for computing has a high level of system instability, and the performance of provided nodes may differ significantly from the specified one, both in greater and less ways. Many simulation toolkits support the simulation of the cloud environment, but they do not take into account in their approach described instabilities. As a result, very often the predicted makespan differs from the actual one by several times. This paper presents an extension of the functionality of the WRENCH Toolkit, which can be used to collect task runtime statistics from the cloud environment and simulate similar instability during simulation.