Self-Managing Databases—If Not Now Then When?
Presented by:
Johann Schleier-Smith
Many software developers hesitate to choose Postgres because running it well can be challenging. While it is easy to get started, at certain levels of scale or application complexity it becomes challenging to set the 300+ Postgres configuration knobs properly.
From the early days of databases, people have sought to automate their configuration and administration. These efforts gained traction in the 1990s, when Microsoft created the AutoAdmin research group. Today, both Oracle and Microsoft have tools that help administrators run their databases better. Unfortunately, these have a spotty track record and can reduce database stability. This is because they are based on dated techniques and do not benefit from the AI advancements of recent decades.
We have been studying how to adapt the technology of self-driving cars to turn Postgres into a fully self-managing database. Self-driving cars operate in complex and safety-critical environments, often involving scenarios that require at least as much good judgment as running a mission-critical database.
In this talk, we review the history of database automation, then dive into opportunities to reimagine it using architectures adapted from today’s commercially available self-driving automotive technology.
- Date:
- 2024 April 19 09:00 PDT
- Duration:
- 50 min
- Room:
- Santa Clara
- Conference:
- Postgres Conference 2024
- Language:
- English
- Track:
- Essentials
- Difficulty:
- Intermediate