SQL - 哪个索引将加速 COUNT() 查询

Question

I have a very simple Postgres database that looks like that (it's defined in Django but this doesn't matter):

class Country:
    id = models.UUIDField(primary_key=True, default=uuid.uuid4, editable=False)
    # Not important
    # name = models.TextField(nullable=False)

class City:
    id = models.UUIDField(primary_key=True, default=uuid.uuid4, editable=False)
    # name = models.TextField(nullable=False)
    country = models.ForeignKey(Country)

class Street:
    id = models.UUIDField(primary_key=True, default=uuid.uuid4, editable=False)
    # name = models.TextField(nullable=False)
    city = models.ForeignKey(City)
    # Just some boolean field
    is_big = models.BooleanField(nullable=False)

I want to query the number of "big" streets in a country. This is done like that:

SELECT COUNT(*)
  FROM Street
 INNER JOIN City
    ON Street.city_id = City.id
 WHERE Street.is_big = true AND City.country_id = 'xxxxxxx'::uuid

There are a total of around ~20 countries, ~5000 cities and ~2 million streets in the database, which is not a lot, yet this query can take sometimes 1-2 seconds.

What index should I add to make this query fast?

Answer 1

This is your query:

SELECT COUNT(*)
FROM Street s JOIN
     City c
     ON s.city_id = c.id
WHERE s.is_big = true AND c.country_id = 'xxxxxxx'::uuid;

There are two possible approaches for this. However, I suspect that neither will work really well in all cases -- I am guessing that some countries have lots of big streets.

The first approach is to filter on the big streets: street(is_big, city_id) and city(id, country_id) . The second is to filter on the specific country: city(country_id, id) and street(city_id, is_big) .

If the database is static, then the query can use an index-only scan -- assuming the data is "visible".