The Design Of Ganglia-based Data Center Monitoring Platform

In recent years, the development of cloud computing technology and services are thriving. As cloud computing fundamental infrastructure, data centers obtain more and more attention, at the same time, data centers monitoring technology has become the focus of cloud-related research areas. Monitoring platforms has its own particularities in data center environment, clusters of equipment, dynamic growing management scale, diversification of monitoring objects, high reliability requirements, these applications presents challenges for data center monitoring system design and development.This research project utilizes the Modular Data Center system in Nanyang Technological University, Singapore, designing and implementing data center monitoring platform. The platform learns from the framework of large-scale distributed cluster monitoring system-Ganglia developed by University of California, Berkeley. For ICT infrastructure, power supply system, air-condition environment system, the three objects categories, the platform re-develops a high-performance data storage system based on MongoDB. In addition, the monitoring platform increases user interface based on virtual reality technology. The entire system above, achieves the targets of comprehensive monitoring, high-performance high-availability database cluster, revolutionary virtual reality user interface.On the basis of comprehensive investigation, combining with professional IT maintenance experience, the writer deals with data center monitoring platform requirements analysis, clarifies desired system performance and design goals. The writer believes that the platform should have6objective features, including scalability, low overhead, extensibility, robustness, manageability, to satisfy the application requirements in product deployment environment.Then, the paper elaborates hardware platform and overall software system solutions. The hardware platform is Modular Data Center system located in campus of Nanyang Technological University, which is composed of a modular data center room, Network Operations Center, users’and developers’equipment. Meanwhile, the article demonstrates overall architecture of software systems from two aspects. From the dataflow side, there are five steps including data collection, data transmission, data storage, data visualization, user interaction. From the perspective of subsystem modules, there is five parts including ICT infrastructure monitoring subsystem, power supply monitoring subsystem, air-conditioned environment monitoring subsystem, data storage subsystem, user interaction subsystem.In addition, the paper describes specific implementation of data center monitoring platform. For ICT infrastructure, power supply systems, air-conditioned environment system three types of objects, the paper introduces implementation of data collection and transmission. For the data center storage subsystem requirement, MongoDB database cluster is deployed, with schema-less data storage solution. In the user interaction step, the project utilizes3D modeling technique and real-time interactive3D engine, implementing virtual reality user interface.Finally, the paper demonstrates function and performance testing for monitoring platform prototypes. For the data collection and transmission section, user interaction part, the paper focuses on functional tests. Considering of high performance requirements for data storage, the paper designs benchmarking platform based on Apache and PHP, accurately measuring MongoDB performance indicators, obtaining accurate, scientific test results.This paper designs and implements a complete data center monitoring platform, based on Modular Data Center hardware system, Ganglia-based framework, integrated comprehensive equipment monitoring, high-performance data storage system, revolutionary user interaction, achieving a comprehensive, efficient and innovative system design targets.