Hi!

I have two tables with some indices on them:

CREATE TABLE subscriber
(
  id serial NOT NULL,
  anumber character varying(32) NOT NULL,
  CONSTRAINT subscriber_pk PRIMARY KEY (id)
)

CREATE INDEX anumber_idx_numeric
  ON subscriber
  USING btree
  (anumber::numeric);

CREATE TABLE output_message_log
(
  id serial NOT NULL,
  subscriber_id integer NOT NULL,
  crd timestamp without time zone NOT NULL DEFAULT now(),
  CONSTRAINT output_message_log_pk PRIMARY KEY (id),
  CONSTRAINT subscriber_fk FOREIGN KEY (subscriber_id)
      REFERENCES subscriber (id) MATCH SIMPLE
      ON UPDATE NO ACTION ON DELETE NO ACTION,
)

CREATE INDEX crd_idx
  ON output_message_log
  USING btree
  (crd);

CREATE INDEX subscriber_id_idx
  ON output_message_log
  USING btree
  (subscriber_id);

I would like to run a query like this one:

select l.id
from output_message_log l join subscriber s on l.subscriber_id = s.id
where s.anumber::numeric = 5555555555
order by l.crd desc
limit 41
offset 20

The thing I do not understand is why postgresql wants to use crd_idx:

"Limit  (cost=4848.58..14788.18 rows=41 width=12) (actual time=7277.115..8583.814 rows=41 loops=1)"
"  ->  Nested Loop  (cost=0.00..1195418.42 rows=4931 width=12) (actual time=92.083..8583.713 rows=61 loops=1)"
"        ->  Index Scan Backward using crd_idx on output_message_log l  (cost=0.00..17463.80 rows=388646 width=16) (actual time=0.029..975.095 rows=271447 loops=1)"
"        ->  Index Scan using subscriber_pk on subscriber s  (cost=0.00..3.02 rows=1 width=4) (actual time=0.026..0.026 rows=0 loops=271447)"
"              Index Cond: ("outer".subscriber_id = s.id)"
"              Filter: ((anumber)::numeric = 36308504669::numeric)"
"Total runtime: 8584.016 ms"

I would like postgresql to use subscriber_id_idx which resulst in a far less execution time on this database.

I tried to lower random_page_cost, but that didn't help as an index is already used, just not the "good" one.

Could you please comment on this issue and suggest some possible soulutions?

Thanks,

Zizi