PGCon2007 - Confirmed

PGCon 2007
The PostgreSQL Conference

Emmanuel Cecchet
Day 3
Room SITE A0150
Start time 13:30
Duration 01:00
ID 20
Event type Lecture
Track Replication
Language English

PostgreSQL replication strategies

Understanding High Availability and choosing the right solution

This talk introduces the key aspects you need to consider when building Highly Available solutions based on PostgreSQL. Current key replication and clustering technologies for PostgreSQL are presented to show how they address issues such as load balancing, performance scalability, failover/failback and upgradeability.

High Availability is a broad term that encompasses many concepts. We propose to first outline the key factors one should consider when building an HA solution from MTBF, MTTR, downtime cost to planned and unplanned maintenance. According to your needs, there are several strategies available to achieve availability from simple master/slave replication to multimaster clustering with transparent failover. We propose to show how the existing replication technologies for PostgreSQL (Slony, Sequoia, pgpool, PGCluster) fit different needs. We use different real life scenarios and use cases with various fault tolerance, performance, load balancing and availability requirements. We show how each technology can be used in these scenarios with its pros and cons.