Question for anyone...
I have to queries. One runs in about 2 seconds. The other takes upwards
of 2 minutes. I have a temp table that is created with 2 columns. This
table is joined with the larger database of call detail records.
However, these 2 queries are handled very differently.
The queries:
First----
calldetail=> EXPLAIN SELECT current.* FROM current JOIN anitmp ON
current.destnum=anitmp.ani AND istf=true;
QUERY PLAN
--------------------------------------------------------------------------------------
Nested Loop (cost=0.00..2026113.09 rows=500908 width=108)
-> Seq Scan on anitmp (cost=0.00..33.62 rows=945 width=8)
Filter: (istf = true)
-> Index Scan using i_destnum on current (cost=0.00..2137.36
rows=531 width=108)
Index Cond: (current.destnum = "outer".ani)
(5 rows)
Second----
calldetail=> EXPLAIN SELECT current.* FROM current JOIN anitmp ON
current.orignum=anitmp.ani AND istf=false;
QUERY PLAN
---------------------------------------------------------------------------
Hash Join (cost=35.99..3402035.53 rows=5381529 width=108)
Hash Cond: ("outer".orignum = "inner".ani)
-> Seq Scan on current (cost=0.00..907191.05 rows=10170805 width=108)
-> Hash (cost=33.62..33.62 rows=945 width=8)
-> Seq Scan on anitmp (cost=0.00..33.62 rows=945 width=8)
Filter: (istf = false)
(6 rows)
The tables:
Table "public.current"
Column | Type | Modifiers
----------+-----------------------------+-----------
datetime | timestamp without time zone |
orignum | bigint |
destnum | bigint |
billto | bigint |
cost | numeric(6,4) |
duration | numeric(8,1) |
origcity | character(12) |
destcity | character(12) |
file | character varying(30) |
linenum | integer |
carrier | character(1) |
Indexes:
"i_destnum" btree (destnum)
"i_orignum" btree (orignum)
Table "public.anitmp"
Column | Type | Modifiers
--------+---------+-----------
ani | bigint |
istf | boolean |
Anyone have any ideas for me? I have indexes on each of the necessary
columns.
Rob
