Oracle Real Application Clusters

Clustering technology has been around for a long time. It was in the early 1980s that Digital Equipment Corporation (DEC) first introduced the clustered configuration called the VAXCluster. Around the same period, I started my computer career working on the PDP-11 systems, also from DEC. PDP-11s were the first platforms that were used by Oracle Corporation to develop the first relational database in 1978, when Oracle released Version 1 of the Relational Database Management System (RDBMS) product. This commercial relational database entered the market nearly 8 years after Dr. E.F. Codd first introduced the relational database model in June 1970 through his white paper entitled, ''A Relational Model of Data for Large Shared Data Banks'' (published at IBM, San Jose, USA).

After the initial set of VAXClusters entered the market in the early 1980s, DEC introduced a relational database called Rdb.^[1] This database used the VAXClusters as their primary platform, thus becoming the first commercial clustered relational database. It was during the late 1980s that I started my career in the relational database world with Rdb on VAXClusters as my first venture into the clustered arena. In 1986, Oracle also introduced the first clustered database solution called Oracle Parallel Server (OPS) with Version 5. In spite of OPS supporting a clustered environment, it did not take advantage of the clustering technology that was available, for example in VAXClusters, namely communicating via the cluster interconnect. This caused a serious performance issue, until Version 9.0 when Oracle, after inheriting some of the clustered database knowledge from Rdb and the technology from Compaq (HP recently acquired Compaq, who had previously acquired DEC), introduced Real Application Clusters (RAC).

All of this experience from VAXClusters to Rdb to OPS to RAC created this motivation to share the knowledge in the form of this book.

About this book

Compared to its predecessor (OPS), the architecture behind the RAC implementation has changed significantly. The book will discuss in detail subjects that I have never seen discussed elsewhere, for example the internals of RAC operations, cache fusion, fusion recovery, fast reconfiguration of RAC, etc.

Areas such as partitioning, types of indexes, tablespace management, transaction management, backup and recovery principles, and internals of the recovery operation have been covered extensively in this book. What good is a book on RAC that does not provide discussions on tuning? In my view, the coverage of extensive tuning, in over three chapters, starting with tools and utilities such as SQL Trace, STATSPACK, and Oracle Event and Wait interfaces, covering single instance tuning followed by cluster tuning, including tuning the cluster interconnect and cache transfer, makes this book unique.

Throughout the book, examples are provided with file dumps, followed by discussions and analysis of problem solving. The book also provides discussions on migrating to a RAC environment.

Oracle Real Application Clusters provides comprehensive coverage of the features, technology, and principles of Oracle's RAC product, which is part of the Oracle 9i release of their RDBMS product. This book is divided into four main parts (Concepts, Architecture, Operations, and Appendices) covering various areas of the application/database design, development, and maintenance life cycles. Starting with the fundamentals, an introduction to the concepts of the various hardware architectures and the clustering technology available, the book discusses the pros and cons, leading into discussion of the RAC technology. After the formal introduction of the concepts of clustering and configurations, the book discusses the various theories of asynchronous (parallel) and synchronous processing and how they could be applied to the database tier of the enterprise architecture.

^[1]Oracle Corporation currently owns Rdb.Oracle purchased Rdb from DEC in 1994.