The announcement of Postgres-BDR 3.0 last month at PostgresConf US had been long-awaited in the PostgreSQL community. Complex use cases from customers have driven the development of BDR far beyond the original feature set, resulting in a more robust technology than ever imagined – so we’re happy to say that it was worth the wait.
Postgres-BDR (Bi-Directional Replication) enhances PostgreSQL with advanced multi-master replication technology that can be used to implement very high availability applications.
For an introduction to Postgres-BDR covering an overview of its complex architecture and its common use cases – 2ndQuadrant held the “Introduction to Postgres-BDR” webinar as part of its PostgreSQL webinar series.
The webinar was presented by Simon Riggs, Founder and CEO of 2ndQuadrant, who is also one of the committers of the PostgreSQL project. Those who weren’t able to attend the live event can now view the recording here.
Some additional Questions and Answers are shown here:
Q: Is it necessary for all nodes to replicate data to all other nodes?
A: BDR has a fully connected network of nodes based on mesh topology. All nodes replicate data to all other nodes. You can, however, choose a subset of tables to replicate within a node.
Q: Can I use a star topology instead of a network (all with all)?
A: BDR has a fully connected network of nodes based on mesh topology, which is the only topology currently supported.
Q: Are there any tools available that help you resolve conflict visually?
A: BDR resolves conflicts automatically. Also, it keeps a log of resolved conflicts as well as any unresolvable problems for later assessment. These conflicts are visible using GUI tools as well as command line access.
Q: Does Postgres-BDR has Lock conflict management? If yes, how it works between the different nodes?
A: It is a deliberate design choice by Postgres-BDR to not propagate lock information since it would cause an unacceptable performance loss across long distance network links, as occurs in more tightly coupled clustering solutions. This allows BDR nodes to operate independently apart from replicating data changes.
Q: Can you elaborate on backup/recovery options in BDR3?
A: You can get more detailed information by downloading the BDR3 white paper from this link.