Oracle Real Application Clusters

8.1 Introduction

In the previous chapter we looked at some of the key features that should be considered while designing databases for Oracle, and specifically about designing databases for a clustered solution using RAC. We looked at the various key features that are available within Oracle. These key features help the enterprise in meeting the various business requirements like availability, scalability, and manageability. For example, features such as partitioning provide spreading out of the data instead of concentrating the entire data for a given table into one data file. Partitioning helps in spreading I/O among the various data files which otherwise would have caused users to queue one behind the other to get access to the single data file. These queues cause waits on the disk while retrieving data and thus cause performance bottlenecks. Spreading data across many spindles would noticeably reduce these performance bottlenecks.

We also looked at the various indexing features that Oracle provides. Selecting the right indexing strategy is important to achieve the ideal performance while retrieving data from the database. However, selection of this strategy depends on various factors, such as, is the data retrieved using finite criteria like in an OLTP system or is it retrieved based on a set of range values like in a DSS system?

Finally, we also looked into a key new feature introduced in Oracle 8i that helps in the overall performance and scalability of a data warehouse application. The materialized views are a boon in disguise that helps develop complex join operations into more meaningful denormalized structures for easy data retrieval. Updates to the materialized views happen in smaller chunks as data is loaded into the base tables and when the view is refreshed. At runtime it is only this view that the user will have to access to retrieve the data.

What good is a design that is not implemented on a well-configured platform even though it is done with utmost care and has selected the correct features? The underlying database platform should be installed, configured, and managed well to help harvest the potential of a good database design.

Careful planning of the installation, configuration, and administration of the underlying database helps to achieve the overall goal of the business application, basically efficient use and performance. It does not matter how well the system has been laid out, structured, designed, and developed; if the system is not installed, configured, and managed efficiently, achieving the desired result could be far from reality.

In this chapter we will look into the steps taken for installing, configuring, and managing the RAC environment. The first step to the installation process is to follow a standard that will provide a consistent way to define disks and directory structures. One such standard developed and recommended by Oracle is the Optimal Flexible Architecture (OFA). This architecture or standard is widely followed among most customers using the Oracle RDBMS product.