Sample Scan

Description

Sample scan is a method of scanning tables. It can randomly extract some data from a table for query instead of scanning the entire table. Sample scan is used together with the TABLESAMPLE keyword in [ TABLESAMPLE sampling_method ( argument [, ...] ) [ REPEATABLE ( seed ) ] ] mode. This technology is usually used in a query scenario in which a large table or random sampling is required, to improve query efficiency.

Typical Scenarios

When the tablesample syntax is used for query, sampling_method is set to system or bernoulli.

Examples

Example: Query the sampling data.

-- Prepare data.
gaussdb=# CREATE TABLE t1 (c1 number, c2 number, c3 number); 
CREATE TABLE 
gaussdb=# INSERT INTO t1 VALUES (generate_series(1,100), 2, 3); 
INSERT 0 100

-- Execution result.
gaussdb=# EXPLAIN SELECT c1 FROM t1 tablesample system(10); 
                     QUERY PLAN                      
---------------------------------------------------- 
 Sample Scan on t1  (cost=0.00..272.00 rows=10000 width=6)
   Sampling: system (10::real)
(2 rows)

-- Drop.
gaussdb=# DROP TABLE t1;

In the preceding example, the output of the sample scan operator is as follows.