Charlie Savage [EMAIL PROTECTED] writes:
1. Postgresql estimates the index scan will be 50 times more costly
than the seq scan (112870376 vs 2229858) yet in fact it only takes 3
times longer to execute (2312426 s vs. 768403 s). My understanding is
that postgresql assumes, via the
Hi Tom,
From pg_stats:
schema = tiger;
tablename = completechain;
attname = tlid;
null_frac = 0;
avg_width = 4;
n_distinct = -1;
most_common_vals = ;
most_common_freqs = ;
correlation = 0.155914;
Note that I have default_statistics_target set to 100. Here is the
first few values from
Following up with some additional information.
The machine has 1Gb physical RAM. When I run the query (with sort and
seqscan enabled), top reports (numbers are fairly consistent):
Mem: 1,032,972k total, 1,019,516k used, 13,412k free, 17,132k buffers
Swap: 2,032,140k total, 17,592k used,
