back
2 comments
I'm badly aching for this feature. Running log aggregation are such a pain when a statistical understanding of what is slow is as useful and extremely cheap (in terms of compute power) to gather. Adding some of the detailed information to that log would also be quite expensive (such as detailed timing of queries in a handful of dimensions), but are cheap to aggregate on the fly.

Even though it is a mere contrib and as such not included in the top of the release notes, I predict pg_stat_statements will be a huge leap in productivity in terms of tuning database indexes, which nobody gets around to getting time to do because it's a total pain to do the reporting about what is actually slow. Now it'll be pretty easy by comparison and I hope people will write even more reporting tools to make it trivial to do a reasonably good job.

An index suggestion tool may be more credibly in the cards of the moderately distant future in the future now that this critical work is done.

This sounds like a fantastic feature! So much easier than checking logs and running explain manually. Plus it also looks like something you can trivially run in production.

Is there any close equivalent for mysql, which we use at $work? The slow query log is the closest I know about, but doesn't seem nearly as useful (or detailed).