We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control
deepmind.google·6d ago
AI Engineering
Lobsters·Jul 2·5 min read
kernel asynchronous reads in PostgreSQL 19 (io_uring)
TL;DR
PostgreSQL 19 faced limitations with synchronous read calls, which hindered performance. The introduction of the io_uring method allows PostgreSQL to leverage Linux's asynchronous buffered I/O for improved efficiency.
✦ Why It Matters
Engineers can optimize PostgreSQL performance by utilizing io_uring for asynchronous I/O operations.
Key Takeaways
- Explore io_uring for improved I/O performance in PostgreSQL.
- Consider disabling seccomp in Docker for advanced I/O features.
- Utilize io_combine to enhance data prefetching efficiency.
- Monitor the impact of kernel-managed I/O on query execution times.
Full Summary
PostgreSQL 19 aims to enhance performance by addressing the limitations of synchronous read calls, specifically pread64() and preadv(), which can slow down query execution. By implementing the io_uring method, PostgreSQL can utilize Linux's asynchronous buffered I/O capabilities, allowing for more efficient I/O operations.
In a Docker container with Secure Computing Mode (seccomp) disabled, the author executed a query that benefits from io_combine, a technique that optimizes data prefetching. The results showed that the kernel now manages I/O operations directly, leading to the absence of PostgreSQL's I/O worker processes.
This shift not only streamlines the process but also potentially improves query performance. The findings suggest that leveraging kernel-level asynchronous I/O can significantly enhance database efficiency, particularly for workloads that involve large data sets.
Engineers can apply these insights to optimize their own PostgreSQL configurations and improve application performance.