Presented by:

I am a Software Engineer working in PostgreSQL at Datadog.

No video of the event yet, sorry!

Ensuring high availability is crucial for databases to achieve resiliency and durability. However enabling high availability while running PostgreSQL on Kubernetes poses unique challenges navigating complexities such as increased write latency, node failures, leader elections, handling replication lag etc. This talk will focus on

  • Architectural components of how at Datadog we run PostgreSQL on Kubernetes.
  • What is High availability?
  • Why did we want to achieve high availability and what were some of the challenges we faced.
  • What alternative did we consider to ensure high availability?
  • How did we test the impact of synchronous replication using Patroni to achieve high availability.
  • How we implemented synchronous replication.
  • How do we monitor synchronous replication and address related issues?
  • What has been the impact of synchronous replication in our production systems?

Date:
2024 November 6 10:40 PST
Duration:
50 min
Room:
Ops: 421
Conference:
Seattle 2024
Language:
Track:
Ops
Difficulty:
Medium