基于云计算的分布式缓存系统的研究和实现

发布时间：2018-01-07 15:05

本文关键词：基于云计算的分布式缓存系统的研究和实现　出处：《重庆大学》2012年硕士论文　论文类型：学位论文

【摘要】：互联网的高速发展以及全球网民数量的快速增长，让网络服务的数据量以指数级增长。云计算作为一个新型的面向服务的计算模式，它具有资源高扩展、强大的计算和存储能力等特点。因此，本文基于目前云计算IaaS，Paas以及SaaS的三层基础框架的思想来构建和部署分布式缓存系统。本文首先介绍了该系统研究的背景及意义，国内外研究成果及现状。随后分析及介绍了云计算的特点及架构、云计算的典型应用及关键技术。在以上基础之上形成了以下本文的主要内容： 1.基于Dynamo系统框架的基础上提出了DCache分布式缓存系统，并结合当下传统关系数据库的局限性进行需求分析，提出了分布式系统所需支持的三个需求：高并发读写、海量数据的高效率存储和访问、对数据库的高可扩展性和高可用性。 2．对DCache分布式缓存系统的数据缓存层进行设计实现。本系统的数据缓存层技术采用memcached协议的内存分配和数据删除方式，并设计出多种存储模式以满足各种应用场景。基于该技术设计并详细介绍了缓存层与数据层的功能框架。 3．本系统采用BDB嵌入式数据库作为本系统的持久化存储，并采用基于内存索引的顺序持久化模式（IFS）来解决BDB数据库面临海量数据存储与读取时的效率低下问题。IFS模式可以将随机写变成顺序写，将创建、修改、删除都演化为文件的追加。而内存中存放数据的索引，，永远指向最新的数据,因此大大提高了海量数据的存储及读取速度。 4．本文针对高性能分布式缓存系统DCache的海量数据应用场景，分别在JAVA和.net环境下设计、编写测试用例。并用每秒响应次数（IOPS）、带宽（throughput）这两个指标来评估数据访问能力，成功率来评估系统的可靠性。最后的测试结果证明了DCache系统在并发量和数据处理方面的优势，已经满足了预期的面对海量数据时所需要达到的需求。
[Abstract]:The rapid development of the Internet and the rapid growth in the number of Internet users worldwide, so the amount of data network services to increase exponentially. Cloud computing as a new service-oriented computing model, it has the characteristics of highly scalable, powerful computing and storage capacity. Therefore, this paper based on the cloud computing IaaS, Paas and SaaS the three layer based framework to build and deploy a distributed cache system.
This paper first introduces the background and significance of the system, the research results and the current situation at home and abroad. Then it analyzes and introduces the characteristics and architecture of cloud computing, the typical applications and key technologies of cloud computing.
1. based on the Dynamo system framework is put forward on the DCache distributed cache system, and combined with the current limitations of the traditional relational database needs analysis, put forward three requirements needed to support distributed systems: high concurrent read and write, high efficiency of mass data storage and access to the database, high scalability and high availability.
The data cache layer 2. of the DCache distributed cache system was designed and implemented. The system data cache layer using memcached protocol memory allocation and data deletion, and a variety of storage modes are designed to meet a variety of application scenarios. This design and introduces the function of frame buffer layer and data layer based on.
3. this system adopts BDB embedded database as the persistence of the system, and uses the memory indexed sequential persistence model based on (IFS) to solve the BDB database facing massive data storage and read the problem of low efficiency of.IFS mode can be written into a sequence of random write, modify, delete all will be created for the evolution of the file append. While the memory index of the data is stored, always pointing to the new data, thus greatly improving the massive data storage and reading speed.
The application of massive data scene 4. according to the high performance distributed cache system of DCache, JAVA and.Net respectively in the design environment, write test cases. And the second response times (IOPS), bandwidth (throughput) of the two indicators to evaluate the ability of data access, the success rate to assess the reliability of the system.
The final test results prove the advantages of DCache system in concurrency and data processing, and have already met the expected needs of massive data.

【学位授予单位】：重庆大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP333

【相似文献】