webinar register page

Webinar banner
Optimizing Postgres I/O Performance and Costs
We take a deep dive into I/O optimization in Postgres, and how tuning I/O helps reduce overall database costs. We will review the different sources of read and write I/O in Postgres, including queries, indexes, internal cache usage and background processes. Lastly, we’ll also take a look at relevant configuration settings and how to tune them.

Here is what we will walk through in this webinar:

- Common causes for I/O spikes in Postgres and how they affect infrastructure costs
- How to map high read and/or write IOPS to database activity
- Using EXPLAIN plans to find heavy hitting I/O intensive queries
- Caching in Postgres, and how the buffer cache hit ratio can be improved to reduce I/O
- Reducing write I/O from inefficient or unused indexes with high index write overhead
- How inefficient data distribution affects index performance and causes high write IOPS
- The checkpointing and background writer processes in Postgres
- Tuning fsync overhead through tuning of synchronous_commit
- Tuning memory configuration settings like work_mem to reduce temporary file I/O
- Optimizing Linux Readahead and tuning effective_io_concurrency
- Finding the root cause of high I/O costs of cloud databases such as Amazon Aurora
- How pganalyze helps improve database performance and reduce I/O costs
- Upcoming work in the Postgres community for asynchronous I/O, and why it matters


* Required information


Lukas Fittl
CEO and Founder of pganalyze @pganalyze