Presented by:


Shaun Thomas


Shaun is the author of the PostgreSQL High Availability Cookbook, and a frequent presenter at the Postgres Open conference on topics such as handling extreme throughput, High Availability, monitoring, architecture, and automation. He’s contributed a few PostgreSQL extensions as well as a tool for administering massive database clusters. On occasion, he’s even been known to guest lecture at the local university. His goal is to help the community make PostgreSQL a bigger, better database for everyone to enjoy.

No video of the event yet, sorry!

BDR: What does it do? Does it do things? Let's find out!

Non trigger-based Bi-Directional Replication has been a long-awaited Postgres feature for years, and with good reason. Who doesn't want multi-master freedom and indemnity? In regard to features, BDR adds quite a lot to Postgres that a DBA might leverage while building their dream architecture. Foremost among these is the fact BDR is specifically designed for high-latency environments.

International companies or internet-based businesses that want to reduce latency must all contend with the speed of light. With AJAX-enabled sites, a single web application can incur dozens of initial requests, and several more while the page is active. Each one represents a round trip that may exceed 150ms. Local application servers can prevent this, but a fully integrated stack depends on a local database as well. This is where BDR comes in.

This talk will tackle how BDR works in such a geographically distributed environment. We'll show how database nodes in Sydney, Dubai, Dallas, and Tokyo might interact. We'll cover a couple failure scenarios and how to recover. We'll present a mitigation scenario or two for communication conflicts. Most of all, we'll demonstrate why this is an area where BDR particularly excels.

That's what BDR does. Come to this talk to find out how.

2017 August 12 11:30
50 min
PGConf Local: Seattle