SELECT 查询中的索引扫描速度较慢

Index scan slow in a SELECT query

请问有没有人看到,尽管有索引,为什么 PostgreSQL 10.6 中的以下 SELECT 查询需要更长的时间。

我正在尝试查找在过去一个月(和 display them in a PHP script)中至少执行过一次移动的所有用户:

SELECT
        u.uid,
        u.elo,
        TO_CHAR(avg_time, 'HH24:MI') AS avg_time,
        ROUND(avg_score::numeric, 1) AS avg_score,
        s.given,
        s.photo
FROM words_users u
-- take the most recent record from words_social
LEFT JOIN LATERAL (SELECT * FROM words_social s WHERE s.uid = u.uid ORDER BY s.stamp DESC LIMIT 1) AS s ON TRUE
-- only show players who where active in the last month
WHERE EXISTS (SELECT 1
            FROM words_moves m
            WHERE u.uid = m.uid
            AND m.action = 'play'
            AND m.played > CURRENT_TIMESTAMP - INTERVAL '1 month')
ORDER BY u.elo DESC;

结果如下 EXPLAIN ANALYZE output

                                                                          QUERY PLAN
-----------------------------------------------------------------------------------------------------------------------------------------------------------------
 Sort  (cost=20075.91..20078.03 rows=848 width=160) (actual time=433.461..433.477 rows=219 loops=1)
   Sort Key: u.elo DESC
   Sort Method: quicksort  Memory: 58kB
   ->  Nested Loop Left Join  (cost=0.71..20034.67 rows=848 width=160) (actual time=0.065..433.315 rows=219 loops=1)
         ->  Nested Loop Semi Join  (cost=0.43..12970.83 rows=848 width=32) (actual time=0.049..431.592 rows=219 loops=1)
               ->  Seq Scan on words_users u  (cost=0.00..391.02 rows=10702 width=32) (actual time=0.005..2.351 rows=10702 loops=1)
               ->  Index Scan using words_moves_uid_idx on words_moves m  (cost=0.43..80.04 rows=115 width=4) (actual time=0.040..0.040 rows=0 loops=10702)
                     Index Cond: (uid = u.uid)
                     Filter: ((action = 'play'::text) AND (played > (CURRENT_TIMESTAMP - '1 mon'::interval)))
                     Rows Removed by Filter: 35
         ->  Limit  (cost=0.29..8.30 rows=1 width=180) (actual time=0.004..0.005 rows=1 loops=219)
               ->  Index Scan using words_social_uid_stamp_idx on words_social s  (cost=0.29..8.30 rows=1 width=180) (actual time=0.004..0.004 rows=1 loops=219)
                     Index Cond: (uid = u.uid)
 Planning time: 0.289 ms
 Execution time: 433.526 ms
(15 rows)

为什么 INDEX SCAN 需要更长的时间?

下面是用到的表和索引 -

words_moves:

                                      Table "public.words_moves"
 Column  |           Type           | Collation | Nullable |                 Default
---------+--------------------------+-----------+----------+------------------------------------------
 mid     | bigint                   |           | not null | nextval('words_moves_mid_seq'::regclass)
 action  | text                     |           | not null |
 gid     | integer                  |           | not null |
 uid     | integer                  |           | not null |
 played  | timestamp with time zone |           | not null |
 tiles   | jsonb                    |           |          |
 score   | integer                  |           |          |
 letters | text                     |           |          |
 hand    | text                     |           |          |
 puzzle  | boolean                  |           | not null | false
Indexes:
    "words_moves_pkey" PRIMARY KEY, btree (mid)
    "words_moves_gid_played_idx" btree (gid, played DESC)
    "words_moves_uid_idx" btree (uid)
Check constraints:
    "words_moves_score_check" CHECK (score >= 0)
Foreign-key constraints:
    "words_moves_gid_fkey" FOREIGN KEY (gid) REFERENCES words_games(gid) ON DELETE CASCADE
    "words_moves_uid_fkey" FOREIGN KEY (uid) REFERENCES words_users(uid) ON DELETE CASCADE
Referenced by:
    TABLE "words_scores" CONSTRAINT "words_scores_mid_fkey" FOREIGN KEY (mid) REFERENCES words_moves(mid) ON DELETE CASCADE

words_users:

                                         Table "public.words_users"
    Column     |           Type           | Collation | Nullable |                 Default
---------------+--------------------------+-----------+----------+------------------------------------------
 uid           | integer                  |           | not null | nextval('words_users_uid_seq'::regclass)
 created       | timestamp with time zone |           | not null |
 visited       | timestamp with time zone |           | not null |
 ip            | inet                     |           | not null |
 fcm           | text                     |           |          |
 apns          | text                     |           |          |
 adm           | text                     |           |          |
 motto         | text                     |           |          |
 vip_until     | timestamp with time zone |           |          |
 grand_until   | timestamp with time zone |           |          |
 banned_until  | timestamp with time zone |           |          |
 banned_reason | text                     |           |          |
 elo           | integer                  |           | not null |
 medals        | integer                  |           | not null |
 coins         | integer                  |           | not null |
 avg_score     | double precision         |           |          |
 avg_time      | interval                 |           |          |
Indexes:
    "words_users_pkey" PRIMARY KEY, btree (uid)
Check constraints:
    "words_users_banned_reason_check" CHECK (length(banned_reason) > 0)
    "words_users_elo_check" CHECK (elo >= 0)
    "words_users_medals_check" CHECK (medals >= 0)
Referenced by:
    TABLE "words_chat" CONSTRAINT "words_chat_uid_fkey" FOREIGN KEY (uid) REFERENCES words_users(uid) ON DELETE CASCADE
    TABLE "words_games" CONSTRAINT "words_games_player1_fkey" FOREIGN KEY (player1) REFERENCES words_users(uid) ON DELETE CASCADE
    TABLE "words_games" CONSTRAINT "words_games_player2_fkey" FOREIGN KEY (player2) REFERENCES words_users(uid) ON DELETE CASCADE
    TABLE "words_moves" CONSTRAINT "words_moves_uid_fkey" FOREIGN KEY (uid) REFERENCES words_users(uid) ON DELETE CASCADE
    TABLE "words_reviews" CONSTRAINT "words_reviews_author_fkey" FOREIGN KEY (author) REFERENCES words_users(uid) ON DELETE CASCADE
    TABLE "words_reviews" CONSTRAINT "words_reviews_uid_fkey" FOREIGN KEY (uid) REFERENCES words_users(uid) ON DELETE CASCADE
    TABLE "words_scores" CONSTRAINT "words_scores_uid_fkey" FOREIGN KEY (uid) REFERENCES words_users(uid) ON DELETE CASCADE
    TABLE "words_social" CONSTRAINT "words_social_uid_fkey" FOREIGN KEY (uid) REFERENCES words_users(uid) ON DELETE CASCADE
    TABLE "words_stats" CONSTRAINT "words_stats_uid_fkey" FOREIGN KEY (uid) REFERENCES words_users(uid) ON DELETE CASCADE

words_social:

                Table "public.words_social"
 Column |       Type       | Collation | Nullable | Default
--------+------------------+-----------+----------+---------
 sid    | text             |           | not null |
 social | integer          |           | not null |
 given  | text             |           | not null |
 family | text             |           |          |
 photo  | text             |           |          |
 lat    | double precision |           |          |
 lng    | double precision |           |          |
 stamp  | integer          |           | not null |
 uid    | integer          |           | not null |
Indexes:
    "words_social_pkey" PRIMARY KEY, btree (sid, social)
    "words_social_uid_stamp_idx" btree (uid, stamp DESC)
Check constraints:
    "words_social_given_check" CHECK (given ~ '\S'::text)
    "words_social_photo_check" CHECK (photo ~* '^https?://...'::text)
    "words_social_social_check" CHECK (0 < social AND social <= 64)
Foreign-key constraints:
    "words_social_uid_fkey" FOREIGN KEY (uid) REFERENCES words_users(uid) ON DELETE CASCADE
Referenced by:
    TABLE "words_payments" CONSTRAINT "words_payments_sid_fkey" FOREIGN KEY (sid, social) REFERENCES words_social(sid, social) ON DELETE CASCADE

这是我创建的索引(除了 PK):

CREATE INDEX ON words_moves (gid, played);
CREATE INDEX ON words_moves (uid);

更新: 按照 Laurenz 的建议添加另一个索引有帮助 -

CREATE INDEX ON words_moves (uid, action, played);

这里是新的 EXPLAIN ANALYZE output:

                                                                                  QUERY PLAN
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
 Sort  (cost=14203.82..14205.94 rows=848 width=160) (actual time=19.766..19.780 rows=219 loops=1)
   Sort Key: u.elo DESC
   Sort Method: quicksort  Memory: 58kB
   ->  Nested Loop Left Join  (cost=0.72..14162.57 rows=848 width=160) (actual time=0.105..19.695 rows=219 loops=1)
         ->  Nested Loop Semi Join  (cost=0.43..7098.73 rows=848 width=32) (actual time=0.082..18.644 rows=219 loops=1)
               ->  Seq Scan on words_users u  (cost=0.00..391.02 rows=10702 width=32) (actual time=0.011..1.920 rows=10703 loops=1)
               ->  Index Only Scan using words_moves_uid_action_played_idx on words_moves m  (cost=0.43..20.82 rows=114 width=4) (actual time=0.001..0.001 rows=0 loops=10703)
                     Index Cond: ((uid = u.uid) AND (action = 'play'::text) AND (played > (CURRENT_TIMESTAMP - '1 mon'::interval)))
                     Heap Fetches: 219
         ->  Limit  (cost=0.29..8.30 rows=1 width=180) (actual time=0.003..0.003 rows=1 loops=219)
               ->  Index Scan using words_social_uid_stamp_idx on words_social s  (cost=0.29..8.30 rows=1 width=180) (actual time=0.003..0.003 rows=1 loops=219)
                     Index Cond: (uid = u.uid)
 Planning time: 0.456 ms
 Execution time: 19.813 ms
(14 rows)

您看到昂贵的索引扫描中的过滤器平均删除了 35 行,留下的很少。

您可以创建一个索引来避免这种情况:

CREATE INDEX ON words_moves (uid, action, played);