On Mon, Jul 8, 2019 at 12:19 PM Peter Geoghegan <pg@xxxxxxx> wrote: > Well, you're still running autovacuum very aggressively here. It'll > easily keep up when run on a relatively small table such as this. Also, an exactly equal number of insertions and deletions is rather likely to result in bloated indexes in a way that probably isn't representative of many workloads. Even if the number of insertions was only slightly greater than the number of deletions, then the overall pattern would be one of continual growth, which is generally considered much more interesting. For far far more information on the topic than you want, see the paper "B-Trees with Inserts and Deletes: Why Free-at-Empty Is Better Than Merge-at-Half": https://www.sciencedirect.com/science/article/pii/002200009390020W The salient point made by the paper is that good space utilization rests on the assumption that there are fewer deletes than inserts, though maybe only slightly fewer: "The tendency of the utilization to remain near 69% can be explained by the following arguments: If there are even just a few more inserts than deletes, the B-tree will grow at the net insert rate (the rate of inserts minus the rate of deletes)." If the volume of data never grows past a certain point, then it's unlikely that the space utilization is very important. This may even be premature optimization. -- Peter Geoghegan
