Storage Architecture Design And Performance Analysis Of The ORACLE 11G RAC OLTP Massive Databases

With the continuous informationization development of enterprises, the data volume stored in database as the core component of the information system is growing in a geometrical progression. The efficiency of transaction principles of the transactional OLTP (Online Transaction Processing) database is very slow in the face of mass data, and it is hard to meet the business requirement of the continuous increasing concurrent users. With the trend of global informationization, mass OLTP database performance problems have also become one important concern within the enterprises. For enterprises, maintaining the system stability under high concurrent users and guaranteeing the system response time for single user not exceeding the SLA (Service Level Agreement) signed with business departments become an urgent problem that needs immediate solution. For IT companies which provide services to customers, huge economical loss will be made if cannot satisfy the indicators specified in the SLA.The key points of this study are the detailed storage architecture design and performance analysis of the Oracle 11g RAC OLTP massive database. Focusing on the two popular storage subsystem architecture SAN (Storage Area Network) and NAS (Network Attached Server), this study discusses the database performance relevant factors in two aspects, design and implementation. This study also performs the performance testing and optimization on the whole database system.This thesis firstly introduces existing theory of storage system and database, labors the different disk technology, advantages and disadvantages of different RAID techniques and various factors affecting the performance of the storage subsystem, mainly focuses on the key technologies and architecture of Oracle RAC database. On this basis, this thesis makes the overall architecture design of the Oracle RAC database, including the physical system architecture design, storage subsystem design, network architecture design, and database file system design. Furthermore, this thesis makes a detailed performance analysis for storage subsystem on existing experiment environment.This thesis then performs optimization combining with specific examples on the implementation of the database system in physical experiment environment, including the storage subsystem, server, database, network, etc. It detailed introduces and discusses the specific tuning techniques. This optimization solution is general to all Oracle database system performance optimization.The last part of this thesis performs repeated TPC-C benchmark testing on the massive databases built with different storage subsystem architecture, analyzes the overall database performance by utilizing the Oracle AWR (Automatic Workload Repository) report, finds the system bottleneck and optimizes corresponding resources. The system's overall architecture has been verified under multiple test runs, and the system remains stable and maintains good response time under high concurrent users. Finally, this thesis compares and analyses the best test results and performance statisticses obtained from the databases built with the two kinds of storage architecture, thus to provide the reliable design reference to enterprises who need building massive databases.