Table sampling has long been the real strength of commercial database vendors. Traditional database systems have provided sampling for many years. However, the monopoly has been broken. Since PostgreSQL 9.5, we have also had a solution to the problem of sampling.
Here is how it works:
test=# CREATE TABLE t_test (id int); CREATE TABLE test=# INSERT INTO t_test SELECT * FROM generate_series(1, 1000000); INSERT 0 1000000
First, a table containing 1 million rows is created. Then tests can be executed:
test=# SELECT count(*), avg(id) FROM t_test TABLESAMPLE BERNOULLI (1); count | avg --------+--------------------- 9802 | 502453.220873291165 (1 row) test=# SELECT count(*), avg(id) FROM t_test TABLESAMPLE BERNOULLI ...
