基于Memcached的缓存系统设计与实现

发布时间：2018-05-20 01:00

本文选题：缓存技术 + 多副本　；参考：《大连理工大学》2012年硕士论文

【摘要】：随着网络技术的高速发展,高性能数据缓存系统的研究成为电子商务平台与电信服务提供商的研究热点,同时它也是其它业务高速增长的企业所必须面对的问题和挑战。它对其服务的低延时、高效性和高并发性的要求使其对系统的整体设计、性能指标等各方面都带来了巨大的挑战。随着网络技术的发展以及内存等各种硬件设施的价格越来越低廉,高性能的数据缓存系统得到越来越多的关注。本文首先对缓存技术做了介绍,包括缓存技术的原理、分类及其常用的替换算法。然后针对一些高性能缓存系统做了详细的介绍,包括系统的网络拓扑结构、通信协议及一些常用的分布式缓存系统。接着在对memcached分析基础上,设计了一个基于memcached的高性能缓存系统。在设计过程中考虑了网络拓扑结构,缓存替换与热点问题等方面。本文设计的基于memcached的缓存系统分为三层,即网络层,系统核心层与存储层。网络层负责外部客户端的连接与连接的管理,主要功能是解析命令与处理高并发连接。系统核心层负责整个系统的核心功能,主要功能是数据存储的负载均衡,与热点数据多副本生命周期管理。并且为了有效防止局部的高峰访问,本文提出了一种有效的多副本管理策略。通过该策略不仅可以有效防止高峰访问时,单台服务器成为整个系统的瓶颈,而且可以充分发挥其它空闲服务器的性能。存储层是具体的memcached节点,负责具体数据的存储。本文针对高性能缓存系统应用场景,进行了功能与结构上的合理设计与实现,其测试结果也进一步表明了,本文设计的系统有很好的实用性。
[Abstract]:With the rapid development of network technology, the research of high-performance data cache system has become the research hotspot of e-commerce platform and telecom service provider. At the same time, it is also the problem and challenge that other enterprises with high business growth must face. Due to its low delay, high efficiency and high concurrency, it brings great challenges to the overall design and performance of the system. With the development of network technology and the low cost of various hardware facilities such as memory, high performance data cache system has been paid more and more attention. This paper first introduces the cache technology, including the principle of cache technology, classification and commonly used replacement algorithm. Then, some high performance cache systems are introduced in detail, including the network topology, communication protocols and some commonly used distributed cache systems. Then, based on the analysis of memcached, a high performance cache system based on memcached is designed. The network topology, cache replacement and hot issues are considered in the design process. The buffer system based on memcached is divided into three layers: network layer, system core layer and storage layer. Network layer is responsible for external client connection and connection management, the main function is to resolve commands and handle high concurrent connections. The core layer of the system is responsible for the core functions of the whole system. The main functions are load balancing of data storage and multi-replica lifecycle management of hot data. In order to prevent local peak access effectively, this paper proposes an effective multi-replica management strategy. This strategy can not only effectively prevent peak access, single server becomes the bottleneck of the whole system, but also can give full play to the performance of other idle servers. Storage layer is a specific memcached node, responsible for the storage of specific data. Aiming at the application scenario of high performance cache system, this paper designs and implements the function and structure reasonably, and the test results further show that the system designed in this paper has good practicability.
【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP333

【相似文献】