Re: Large tables, ORDER BY and sequence/index scans

>>>>> "PS" == Pavel Stehule  writes:

    PS> Have you original values random_page_cost and seq_page_cost in
    PS> postgres.conf?

Yes.  To be sure I uncommented the values in postgresql.conf

  seq_page_cost = 1.0			# measured on an arbitrary scale
  random_page_cost = 4.0		# same scale as above

and restarted PostgreSQL.  The result looks basically the same:

  explain analyze declare c cursor for select * from foo2 order by value;
                                                         QUERY PLAN                                                        
  -------------------------------------------------------------------------------------------------------------------------
   Sort  (cost=1829429.20..1854429.20 rows=9999999 width=10) (actual time=43709.313..49265.244 rows=9999999 loops=1)
     Sort Key: value
     Sort Method:  external merge  Disk: 204208kB
     ->  Seq Scan on foo2  (cost=0.00..154049.99 rows=9999999 width=10) (actual time=0.072..1760.585 rows=9999999 loops=1)
   Total runtime: 54399.967 ms
  (5 rows)