Continuous Integrated Testing of Oracle Databases on CERN Agile Infrastructure

Project Specification

With the introduction of Puppet configuring machines hosting Oracle databases managed by IT-DB Group of the European Organization for Nuclear Research (CERN), it is now much easier to apply any changes and perform the upgrades. Doing so on production environments is always a risk, which has to be addressed, by creating proper test environment. With Oracle Real Application Testing and related features it is currently possible to thoroughly test impact of any change, before implementing it on production and avoid virtually all problems.

The aim of the project is to add a new kind of testing i.e. the Oracle Real Application Testing (RAT) to an existing framework – DBTest. With the addition of this module, the DBTest framework would be more robust and be more able to detect performance issues in databases due to modifications. The author within the time frame didn’t complete the RAT module but has written a significant part of the module. Also, the author made some improvements to the existing framework. Some of the improvements are starting a database before connecting to it, making the reporting of the framework easier to read and also enhancing the error messages. 

Abstract

Changes occur to production environments regularly and they pose a risk to production. This could be patches to the underlying kernel, migrating to another distribution or an upgrade to the Operating System version. It could also be parameter changes to the database or a change in the vendor. Sometimes, this leads to degradation in the performance of the database. The author extended a framework for database testing by including a module that utilizes a testing tool provided by Oracle – Real Application Testing. The Real Application Testing (RAT) allows workload of a database to be captured and replayed on a test environment. This document explains the Oracle Real Application Testing then gives a brief overview of DBTest – a framework that automates testing of databases. After this, the author presents how to use this tool to help detect performance changes in databases.

The result is a tool that makes databases performance testing easy. The author made improvements to the original framework and also wrote the rudimentary code for the Real Application Testing framework. Hence, the original framework, DBTest can now also start databases before connecting to them amongst other improvements. It is the aim of the author that subsequently, the Real Application Testing module can be completed.