Timing overhead

Assume you're executing a simple query to count all the customers in the database, and want to time how long this takes:

dellstore2=# \timing
Timing is on.
dellstore2=# SELECT count(*) FROM customers;
 count 
-------
 20000
Time: 7.994 ms

You might then be curious what query plan was used to get this result. The amount of time taken to determine that, may shock you:

dellstore2=# EXPLAIN ANALYZE SELECT count(*) FROM customers;
QUERY PLAN                                                     
----------
Aggregate  (cost=726.00..726.01 rows=1 width=0) (actual time=68.836..68.838 rows=1 loops=1)
   ->  Seq Scan on customers  (cost=0.00..676.00 rows=20000 width=0) (actual time=0.012..33.609 rows=20000 loops=1)
 Total runtime: 68.934 ms
Time: 69.837 ms

This fairly trivial query was picked because it demonstrates something close to a worst-case here, where instrumenting the query causes the result to slow dramatically, to almost 10X as long. Using EXPLAIN ANALYZE...