> And if you run?
> select count(*) from articles where article_id between %s and %s
> ie without reading json, is your buffers hit count increasing?
Tried this. This is somewhat interesting, too... Even index-only scan is faster for the "fast" range. The results are consistently fast in it, with small and constant numbers of hits and reads. For the big one, in contrary, it shows huge number of hits (why? how it manages to do the same with lesser blocks access in "fast" range?) and the duration is "jumping" with higher values in average.
"Fast":
https://i.stack.imgur.com/63I9k.png"Slow":
https://i.stack.imgur.com/QzI3N.pngNote that results on the charts are averaged by 1M, but particular values in "slow" range reached 4 s, while maximum execution time for the "fast" range was only 0.3 s.
Regards,
Vlad
