O'Reilly Open Source Convention - August 1-5, 2005

Tutorial
Slony-I Replication Tutorial
Christopher Browne, Database Administrator, Afilias Canada
Steve Simms, Owner, Technically Sound

Track: Databases
Date: Tuesday, August 2nd, 2005
Time: 1:30pm - 5:00pm
Location: E143-144
This tutorial describes the "care and feeding" of the Slony-I replication system for PostgreSQL. Slony-I is an asynchronous replication system supporting multiple replicas "fed" by a single master. Special capabilities include the ability to shift the master node (to support maintenance), to abandon the master in favor of a replica in case of failure, as well as to add additional replica nodes.

Essential topics covered include:
- What is Slony-I?
- What are the sorts of use cases where it works well? Where is its use contraindicated?
- How Slony-I works
- Slony-I Concepts - Clusters, Nodes, Replication Sets
- Installing Slony-I
- Configuring a first Slony-I Node
- Adding Nodes to a Cluster
- Adding Tables to a Cluster
- Manipulating Replication Sets
- Failover and Moving Set Origins
- Maintenance and Monitoring

We'll also have material on the following topics that may fall more into a Q&A mode:
- Bandwidth Issues (e.g. - handling of initial COPY, ongoing updates)
- Which internal tables are useful to examine for diagnostic purposes
- Issues surrounding DDL updates