This is an implementation of "probabilistic counter" as described in the article "Probalistic Counting Algorithms for Data Base Applications", published by Flajolet and Martin in 1985.
Contents of the extension
The extension provides the following elements
probabilistic_estimator data type (may be used for columns, in PL/pgSQL)
functions to work with the probabilistic_estimator data type
probabilistic_size(nbytes int, nsalts int)
probabilistic_init(nbytes int, nsalts int)
probabilistic_add_item(counter probabilistic_estimator, item anyelement)
The purpose of the functions is quite obvious from the names, alternatively consult the SQL script for more details.
probabilistic_distinct(anyelement, int, int)
where the 1-parameter version uses 4 bytes and 32 salts as default values for the two parameters. That's quite generous and it may result in unnecessarily large estimators, so if you can work with lower precision / expect less distinct values, pass the parameters explicitly.
Using the aggregate is quite straightforward - just use it like a regular aggregate function
db=# SELECT probabilistic_distinct(i, 4, 32) FROM generate_series(1,100000) s(i);
and you can use it from a PL/pgSQL (or another PL) like this:
DO LANGUAGE plpgsql $$ DECLARE v_counter probabilistic_estimator := probabilistic_init(4, 32); v_estimate real; BEGIN PERFORM probabilistic_add_item(v_counter, 1); PERFORM probabilistic_add_item(v_counter, 2); PERFORM probabilistic_add_item(v_counter, 3); SELECT probabilistic_get_estimate(v_counter) INTO v_estimate; RAISE NOTICE 'estimate = %',v_estimate; END$$;
And this can be done from a trigger (updating an estimate stored in a table).
Be careful about the implementation, as the estimators may easily occupy several kilobytes (depends on the precision etc.). Keep in mind that the PostgreSQL MVCC works so that it creates a copy of the row on update, an that may easily lead to bloat. So group the updates or something like that.
This is of course made worse by using unnecessarily large estimators, so always tune the estimator to use the lowest amount of memory.