Research On Reliability Of ORACLE Database System

With the development of information technology, a variety of database-based application have been widely used in all fields. Software failures, especially those in mission-critical areas, can cause substantial problems for both the service provider and the consumer dependant on the service.Software robustness is one of the key indicators measuring software quality. The typical software robustness is the characteristics of the software when it runs into unexpected input and other scenarios like network failure, low level API failure, resource (memory, disk) shortage etc, and it has long been the concern of software researchers, engineers, and practitioners. The issues caused by software robustness is more and more complex when most of applications rely on database and when the scale of the database-based application keeps growing along with more features and more interpretability with other applications. Therefore, being able to accurately measure the robustness of a Database Management System (DBMS) will be vital when assessing or improving the robustness of existing applications, or when it comes to a point of choosing the proper type of DBMS for a new application. There has been enormous research in software robustness, and some of the researches have gotten meaningful exploration and result. However, the problem concerning the robustness of such DBMS is far from being solved. Although many efforts have been made in this research domain, people still find it difficult to make choices among various DBMS for robustness concerns.In this paper, we tried to analyze a real world mission-critical database-based application by collecting the operational profile of the application. The operational profile reflects all the types and frequencies of SQL statements executed in a statistical lifecycle time. We can map the operational profile to Operating System (OS) Application Programming Interface (API) and the frequency of all APIs invoked after integrating such application operational profile with our purposefully written testing SQL scripts. We constructed a reference model for DBMS. We followed the Ballista methodology in testing the robustness of OS APIs, and then we can get the robustness metrics of various DBMS by our Robust Model.At last, we set up a test environment on virtual machine using VMware Workstation: RedHat Enterprise Linux 5 Update 2 (RHEL 5.2, kernel version 2.6.18-8.el5) and three types of DBMS including Oracle 10g, IBM Informix Dynamic Server 9, and MySQL 5.1.7. We rebuild and execute Ballista testing suite to get the robustness metrics of the OS API. We finally get the robustness metrics of the three types of DBMS by running the application against Oracle, Informix and MySQL respectively and employing the Robustness Model we constructed.The results of our testing in this paper offered useful references for the robustness improvement/forecast of existing database based application, as well as for DBMS selection of a new application development.