Cluster resource adjustment based on an improved artificial fish swarm algorithm in Mesos

At present, research of Mesos is in the early stage. Because of the two-level hierarchical scheduling, Mesos doesn't take the whole cluster into consideration in resource scheduling, which could cause load imbalance and low resource utilization. To solve above-mentioned problem, this paper proposed a resource adjustment plan based on an improved artificial fish swarm algorithm, it adopts a new behavior strategy. The users can adjust the mapping between salves and containers to improve cluster performance indexes without affecting the tasks' execution. The test results show the proposed resource adjustment achieved significantly increased load-balancing and resource utilization.