A simulation cloud monitoring framework and its evaluation model

•A framework is provided to design the monitoring system of Simulation Cloud.•Two strategies are provided to reduce monitoring cost and balance load.•An evaluation model is developed to check its feasibility. Simulation Cloud can help users to carry out the simulation tasks in various stages quickly and easily by renting instead of buying all the needed resources, such as the computing hardware, simulation devices, software, and models. A monitoring system is necessary, which can dynamically collect information about the characteristics and status of resources in real time. In this paper, we design a Simulation Cloud Monitoring Framework (SCMF), which is a Monitoring Framework based on Simulation Cloud. The main functions of SCMF include: 1. Collecting performance information of Simulation Cloud (including physical resources and virtual resources). 2. Processing the collected performance information, providing ranking information about resource consumption as the customized service to service layer. 3. Detecting abnormal behaviors on Simulation Cloud in real time. The SCMF is based on hierarchical design. It consists of Root Monitoring Node (RMN), Federation Monitoring Node (RMN), and Main Monitoring Node (MMN). There is only one RMN in SCMF. It is responsible for collecting metadata about Simulation Cloud. For robustness, there are several FMNs in a federation. One is primary FMN and others are backup FMNs. MMN is implementing on every host in Simulation Cloud., MMN is responsible for collecting performance information about the host and virtual nodes. In the paper, it designs Sequence-Bucket strategy, which supports quick response for ranking information about resource consumption. It also designs two strategies: Rank-FMN (Federation Monitor Node) strategy and Huffman-Like Strategy. Huffman-Like Strategy combines small federations to reduce total consumption of SCMF, while Rank-FMN strategy is a load balancing strategy, which relieves the bottleneck of FMNs and spreads the loads equally among FMNs. The characteristics of SCMF are real-time, scalability, robustness, light weight, manageability, and archivability. Meanwhile, we design evaluation models for SCMF, which can provide quantitative results of monitoring accuracy and monitoring cost. The simulation results show that SCMF is accurate, low cost and can response in real-time.