测试表和索引(PostgreSQL的9.5.3):与单项指标的使用ANY(ARRAY [...])
CREATE TABLE public.t (id serial, a integer, b integer);
INSERT INTO t(a, b)
SELECT round(random()*1000), round(random()*1000)
FROM generate_series(1, 1000000);
CREATE INDEX "i_1" ON public.t USING btree (a, b);
CREATE INDEX "i_2" ON public.t USING btree (b);
如果“A = 50”在第一个查询,一切都确定,适当的索引 “I_1” 用于:
SELECT * FROM t WHERE a = 50 ORDER BY b LIMIT 1
"Limit (cost=0.42..4.03 rows=1 width=12) (actual time=0.085..0.085 rows=1 loops=1)"
" Buffers: shared hit=1 read=3"
" -> Index Scan using i_1 on t (cost=0.42..4683.12 rows=1300 width=12) (actual time=0.084..0.084 rows=1 loops=1)"
" Index Cond: (a = 50)"
" Buffers: shared hit=1 read=3"
"Planning time: 0.637 ms"
"Execution time: 0.114 ms"
随着“一个IN(50)”结果是相同的:
SELECT * FROM t WHERE a IN (50) ORDER BY b LIMIT 1
"Limit (cost=0.42..4.03 rows=1 width=12) (actual time=0.058..0.058 rows=1 loops=1)"
" Buffers: shared hit=4"
" -> Index Scan using i_1 on t (cost=0.42..4683.12 rows=1300 width=12) (actual time=0.056..0.056 rows=1 loops=1)"
" Index Cond: (a = 50)"
" Buffers: shared hit=4"
"Planning time: 0.287 ms"
"Execution time: 0.105 ms"
问题是当我尝试使用“a = ANY(ARRAY [50])”。索引错误“I_2”代替“I_1”和执行时间变得更长X25:
SELECT * FROM t WHERE a = ANY(ARRAY[50]) ORDER BY b LIMIT 1
"Limit (cost=0.42..38.00 rows=1 width=12) (actual time=2.591..2.591 rows=1 loops=1)"
" Buffers: shared hit=491 read=4"
" -> Index Scan using i_2 on t (cost=0.42..48853.65 rows=1300 width=12) (actual time=2.588..2.588 rows=1 loops=1)"
" Filter: (a = ANY ('{50}'::integer[]))"
" Rows Removed by Filter: 520"
" Buffers: shared hit=491 read=4"
"Planning time: 0.251 ms"
"Execution time: 2.627 ms"
你可以说:“如果你使用任何(ARRAY [])PostgreSQL的不能使用索引”,但实际上它可以。如果我删除 “ORDER BY” 它再次工作:
SELECT * FROM t WHERE a = ANY(ARRAY[50]) LIMIT 1
"Limit (cost=0.42..4.03 rows=1 width=12) (actual time=0.034..0.034 rows=1 loops=1)"
" Buffers: shared hit=4"
" -> Index Scan using i_1 on t (cost=0.42..4683.12 rows=1300 width=12) (actual time=0.033..0.033 rows=1 loops=1)"
" Index Cond: (a = ANY ('{50}'::integer[]))"
" Buffers: shared hit=4"
"Planning time: 0.182 ms"
"Execution time: 0.090 ms"
我的问题:
如果PostgreSQL是足够聪明地用 “IN”,什么是与任何问题很好地工作(ARRAY [])?
如果我删除了“ORDER BY”子句,它为什么能与ANY(ARRAY [])一起使用?
为什么使用“IN”?如果这是更优化的,那么使用“IN”而不是“ANY(ARRAY [])”是否是好习惯? “结果不会按b排序” - 您能提供一些参考文件来证实这一点吗? –
我已经扩展了解决这些问题的答案。 –
这只是关于一个有1个元素的数组,在任何情况下,如果没有真正的原因,为什么IN的工作方式不同(在这种情况下更好)比ANY(ARRAY []),那么这很混乱。 –