Book Image

PostgreSQL High Availability Cookbook - Second Edition

By : Shaun Thomas
Book Image

PostgreSQL High Availability Cookbook - Second Edition

By: Shaun Thomas

Overview of this book

Databases are nothing without the data they store. In the event of a failure - catastrophic or otherwise - immediate recovery is essential. By carefully combining multiple servers, it’s even possible to hide the fact a failure occurred at all. From hardware selection to software stacks and horizontal scalability, this book will help you build a versatile PostgreSQL cluster that will survive crashes, resist data corruption, and grow smoothly with customer demand. It all begins with hardware selection for the skeleton of an efficient PostgreSQL database cluster. Then it’s on to preventing downtime as well as troubleshooting some real life problems that administrators commonly face. Next, we add database monitoring to the stack, using collectd, Nagios, and Graphite. And no stack is complete without replication using multiple internal and external tools, including the newly released pglogical extension. Pacemaker or Raft consensus tools are the final piece to grant the cluster the ability to heal itself. We even round off by tackling the complex problem of data scalability. This book exploits many new features introduced in PostgreSQL 9.6 to make the database more efficient and adaptive, and most importantly, keep it running.
Table of Contents (18 chapters)
Title Page
About the Author
About the Reviewer
Customer Feedback

Defusing cache poisoning

Not every DBA has experienced disk cache poisoning. Those who have recognize it as a bane to any critical OLTP system and a source of constant stress in a highly-available environment.

When the operating system fetches disk blocks into memory, it also applies arbitrary aging, promotion, and purging heuristics. Several of these can invalidate cached data in the presence of an originating process change such as a database crash or restart. Any memory stored by PostgreSQL in shared memory is also purged upon database shutdown.

Perhaps the worst thing a DBA can do following a database crash or a restart is to immediately make the database available to applications and users. Unless storage is based on SSD or a very capable SAN, random read performance will drop by two or three orders of magnitude as data is being supplied by slow disks instead of by memory. As a result, all subsequent queries will greatly over-saturate the available disk bandwidth. This delays query results...