Today’s day 1 of the conference. I’m looking forward to:
- PostgreSQL 16 and beyond
- Patroni 3.0: What’s New and Future Plans
- Logical Replication – handling of large transactions
- Visualizing Postgres I/O Performance for Development
- All the networking opportunities. Which are already starting as I type this, nursing my coffee after breakfast.
The U. Ottawa campus is very nice, as is Ottawa. Every PGCon being held here is fine with me! PyCon US cycles its conference through different cities, spending two years in each one. That distributes the logistics load. But it also means venue selection and acquisition, local team assembly, and learning the ropes of conference running are continual. PGCon always being in Ottawa means all of that cost is reduced, at the related cost of no other venues sharing in the fun.
I don’t know PGCon’s headcount this year, but so far it seems far smaller than the smallest PyCon I ever attended. (Which was about 500 headcount.) So for PGCon it works.
Yesterday’s tutorial day was a 7/10. The morning session was a newbie introduction to ZFS. I don’t see a use for it by my present employer, but in the future who knows? If I were setting up a bare metal server today I’d surely consider using it. The afternoon session was a deep-dive into postgressql.conf. These are of course exposed in RDS in Parameter Groups. Super interesting and I picked up some information nuggets.
For years I’ve wondered about the magnitude of Postgres’ settings. In version 10 there were 269 settings. In version 14, 365. The era of 1K+ settings will soon be here if this continues. Even if you don’t need to modify a setting, there’s still a cognitive load from seeing it, understanding it, and deciding to not modify it. And the danger of changing it to a setting that’s harmful.
I don’t know the solution, except that I hope someone in the core development team is thinking about this. Auto-tuning scripts and tools exist; hopefully there will be more work in this area.
Seek Nuance 