【作者】 徐文芳；

【导师】 刘宏伟；

【作者基本信息】 哈尔滨工业大学 ， 计算机技术， 2011， 硕士

【摘要】 高性能、高可用服务器,可以应用于电信、金融、工业、能源、及政府部门等多个涉及国家安全及国计民生的领域。而目前国内对高可用的研究主要还是通过冗余技术实现,造成较大的冗余度,因此研制具备自适应、可扩展、可重构、自配置能力的高性能、高可用服务器,能够完善我国信息化建设,对国家的经济、社会安全有着战略性的意义。而服务器管理是研制高性能、高可用服务器的关键,为了实现对整个服务器集中、持续、有效地管理,设计了高性能服务器管理硬件平台。为了实现高可用性,采用多层次架构,设计了本地管理模块LMM和全局管理模块GMM。LMM主要负责对本地节点的状态收集和分析及故障的检测和处理,GMM主要负责监控和管理整个系统,它本身并不直接监测计算节点的软硬件资源运行状态。为了实现对服务器系统内计算节点运行状态的全面监控,采用了带内外相结合的监控方式,实现对服务器系统内部件运行状态的实时监测。此外,GMM的双机热备份技术,多通道的通讯网络冗余技术,保证了系统的高可用。热插拔功能的设计更能明显缩短管理系统的平均维修时间,进一步提高系统的可用性。本文研究了基于IPMI的服务器管理技术、常见的高可用技术,根据高性能、高可用服务器的需求,总结了服务器管理平台的功能需求,然后针对此需求提出了LMM和GMM双层管理架构,并完成了对LMM和GMM的详细设计。最后采用ARM7最小系统+外围通讯接口群的嵌入式系统完成了LMM的硬件实现,并对板卡各个模块及主要功能进行了测试。更多还原

【Abstract】 High-performance and high availability servers can be used in many fields which involved in national security and people’s livelihood such as telecommunications and so on. Nowadays, the research of domestic on availability is redundancy, resulting in greater redundancy.So developping the high-performance and high availability servers with adaptive, scalable, reconfigurable and self-configuration capabilities can improve information construction in China,and have a strategic significance on the country’s economy and social security.Server management is the key technology of developing high-performance and high availability server. In order to achieve the focused, sustained and effective management of server, the paper designs a hardware platform for high-performance server management.The design uses multi-level architecture to achieve high availability-LMM(local management module) and GMM(global management module). LMM is mainly responsible for the state of the local node, GMM is responsible for monitoring and managing the entire system. The design using a combination of in-band and out-of-band monitoring ways to monitor the compute nodes in real-time. In addition, GMM’s hot backup technology, and multi-channel communication network redundancy ensures system availability. Hot-plug feature is designed to shorter mean maintenance time, so as to improve system availability.The paper studys the server management technology based on IPMI , the common high-availability technologies, and the needs of the high-performance, high availability server .The paper summarizes the functional requirements of server management platform and then puts forward a double-layer management structure-- LMM and GMM, the detailed design of LMM and GMM is finished at the same time . Finally the hardware of LMM is implemented using a ARM7 minimum system + Peripheral Communication Interface Group , and the main functions of each module is tested.更多还原