Presented by:

2feb05a2159447c6d104919ee1300a87

Bernhardt Garlipp

Quant Engineering Solutions

My passion in live is problem solving. This leads to a wide area of interests ranging form robotics,astonomy to boardgames.

My work background is - BSC Computer science degree (Hons) - Head of Development at Quant Engineering Solutions - Full stack developer (Python, PSQL, HTML, Javascript) - Creating software for the medical industry

No video of the event yet, sorry!

Dev-Ops. Imagine the following "Developer commits new feature. Feature passes Automate test. Feature gets Deployed. Oh no ... everything is slow."

This talk is aimed to provide methods and tools on how to identify slow and problematic queries on a database cluster and how to address and solve these problems. The second part of the talk will be aimed at lessons learned in our company and how we addressed these issues to solve these day to day problems

We will dive into

  • pg_stat_activity and pg_stat_statements (setup and interpretation)
  • Analysing why a query is slow (use of visual tools)
  • How to optimise your query (Index, joins, ect)
  • Lessons learned (Real live examples)
  • Why more hardware doesn't always solve your problems.
  • (Lessons learned) "Idle in transactions" why this is bad if transactions aren't ended
  • (Lessons learned) High access tables slowing down overtime (autovacuum issues)
  • (Lessons learned) Don't disable autovacuum (basic overview of autovacuum, why it is good)
  • (Lessons learned) autovacuum settings for frequantly accesssed/updated tables to adress bloat
  • (Lessons learned) Why Temp Tables are a bad idea and practical alternative suggestions.

Date:
2019 March 20 09:00
Duration:
50 min
Room:
Sugar Hill
Conference:
Postgres Conference
Language:
Track:
Ops and Administration
Difficulty:
Easy