Question

假设我有长的独特哈希（长度为250个字符）。我将它们存储为primary key。查找某一行会比主键基于int？

慢得多

Answer 1

我通过准备表格来制作简单的基准：

create table test_int(id integer primary key);
create table test_text(id text primary key);

insert into test_int select * from generate_series(1,1000000);
insert into test_text select md5(random()::text)||md5(random()::text)||md5(random()::text)||md5(random()::text)||md5(random()::text)||md5(random()::text)||md5(random()::text)||md5(random()::text)
    from generate_series(1,1000000);

analyze;

explain analyze
select * from test_int where id=<somevalue>;

Index Only Scan using test_int_pkey on test_int  (cost=0.42..8.44 rows=1 width=4) (actual time=0.023..0.024 rows=1 loops=1)
  Index Cond: (id = 58921)
  Heap Fetches: 1
Planning time: 0.129 ms
Execution time: 0.058 ms

explain analyze
select * from test_text where id=<somevalue>;

Index Only Scan using test_text_pkey on test_text  (cost=0.68..8.69 rows=1 width=260) (actual time=0.105..0.106 rows=1 loops=1)
  Index Cond: (id = '3e74346a6a060a208b3916c9b289d1a5020fa9ff3175fdb78022001a2e5f0857a70bddfbd0f148b11ec8928edd9ce008835c9b8624f6afb7b05e2d05ad9cc049418b046ea5e0d4814ffc34fc7fad476996c634353189140702148ba6c1f8fb055ca8ff18747bb3e62f3ca7189fb5165dff850c699e6 (...)
  Heap Fetches: 1
Planning time: 0.122 ms
Execution time: 0.140 ms

在我的机器上，通过数字ID查找的速度大约是文本ID的两倍;

更有趣的是PK索引大小：

select indexname, pg_size_pretty(pg_relation_size(indexname::text)) from pg_indexes  where tablename in ('test_int', 'test_text');

"test_int_pkey"   "21 MB"
"test_text_pkey"  "420 MB"

因此250个字符长文本肯定比整数ID慢得多。

SQL - 索引SELECT，它是一个长哈希

1 个答案: