Use the following steps to cluster a table:

1. Before using the CLUSTER statement, check the time at which the query used in the previous recipe was executed by executing the following commands:

      SELECT schoolid FROM caschools sc JOIN sfpoly sf 
      ON ST_Intersects(sf.geom, ST_Transform(sc.geom, 3310));

2. We get the following performance numbers for three query runs:

      Time: 80.746 ms
      Time: 80.172 ms
      Time: 80.004 ms

3. Cluster the caschools table using the caschools_geom_3310_idx index as follows:

      CLUSTER caschools USING caschools_geom_3310_idx;

4. Rerun the query from the first step three times for the following performance timings:

      Time: 57.880 ms
      Time: 55.939 ms
      Time: 53.107 ms

The performance improvements were not significant.