CUBE是pg内置的插件，实现了一种数据类型cube来表示多维立方体。CUBE是一个多维数据类型，支持两种多维类型：多维POINT、区间（左下+右上）。以及这些几何对象的几何特性搜索和计算（方位搜索、距离计算），这些搜索都支持GiST索引。
例子：
求多维点的距离：
–欧几里得距离
bill@bill=>select '(1,2,3,4)'::cube <-> '(2,2,3,10)'::cube ;
?column?
-------------------
6.082762530298219
(1 row)
–切比雪夫距离(分别每个坐标的距离，取最大值)
bill@bill=>select '(1,2,3,4)'::cube <=> '(2,2,3,10)'::cube ;
?column?
----------
6
(1 row)
–直线距离(分别每个坐标的距离，求和)
bill@bill=>select '(1,2,3,4)'::cube <#> '(2,2,3,10)'::cube ;
?column?
----------
7
(1 row)
使用cube类型进行索引查询：
对于cube类型我们可以使用gist空间索引来进行索引查询。
创建测试表：
bill@bill=>create table test2 (
bill(# id int primary key,
bill(# cb cube
bill(# );
CREATE TABLE
插入数据：
bill@bill=>insert into test2 select generate_series(1,100000),cube(array[random()*10,random()*100,random()*1000],array[random()*10,random()*100,random()*1000]);
INSERT 0 100000
bill@bill=>\dt+ test2
List of relations
Schema | Name | Type | Owner | Size | Description
--------+-------+-------+-------+---------+-------------
bill | test2 | table | bill | 9120 kB |
(1 row)
创建gist索引：
需要注意gist索引很大，我们的表实际只有1MB不到，但是索引竟然有464MB！
bill@bill=>create index idx_t2 on test2 using gist(cb);
CREATE INDEX
bill@bill=>\di+ idx_t2
List of relations
Schema | Name | Type | Owner | Table | Size | Description
--------+--------+-------+-------+-------+--------+-------------
bill | idx_t2 | index | bill | test2 | 464 MB |
(1 row)
包含操作的查询性能：
bill@bill=>explain (analyze,verbose,timing,costs,buffers) select * from test2 where cb @> cube '[(7,67,548), (3,72,389)]';
QUERY PLAN
---------------------------------------------------------------------------------------------------------------------
Bitmap Heap Scan on bill.test2 (cost=79.18..196.86 rows=100 width=60) (actual time=5.892..8.220 rows=2399 loops=1)
Output: id, cb
Recheck Cond: (test2.cb @> '(7, 67, 548),(3, 72, 389)'::cube)
Heap Blocks: exact=1005
Buffers: shared hit=2822
-> Bitmap Index Scan on idx_t2 (cost=0.00..79.16 rows=100 width=0) (actual time=5.720..5.721 rows=2399 loops=1)
Index Cond: (test2.cb @> '(7, 67, 548),(3, 72, 389)'::cube)
Buffers: shared hit=1817
Planning Time: 0.303 ms
Execution Time: 8.572 ms
(10 rows)
相交查询性能：
bill@bill=>explain (analyze,verbose,timing,costs,buffers) select * from test2 where cb && cube '[(7,67,918), (9,32,399)]';
QUERY PLAN
-------------------------------------------------------------------------------------------------------------------------
Bitmap Heap Scan on bill.test2 (cost=391.68..857.07 rows=500 width=60) (actual time=31.379..37.760 rows=32931 loops=1)
Output: id, cb
Recheck Cond: (test2.cb && '(7, 67, 918),(9, 32, 399)'::cube)
Heap Blocks: exact=1137
Buffers: shared hit=24210
-> Bitmap Index Scan on idx_t2 (cost=0.00..391.56 rows=500 width=0) (actual time=31.224..31.224 rows=32931 loops=1)
Index Cond: (test2.cb && '(7, 67, 918),(9, 32, 399)'::cube)
Buffers: shared hit=23073
Planning Time: 0.107 ms
Execution Time: 39.620 ms
(10 rows)
包含点的查询：
bill@bill=>explain (analyze,verbose,timing,costs,buffers) select * from test2 where cb @> cube '(5,23,908)';
QUERY PLAN
---------------------------------------------------------------------------------------------------------------------
Bitmap Heap Scan on bill.test2 (cost=79.18..196.86 rows=100 width=60) (actual time=7.135..8.711 rows=2944 loops=1)
Output: id, cb
Recheck Cond: (test2.cb @> '(5, 23, 908)'::cube)
Heap Blocks: exact=1044
Buffers: shared hit=4650
-> Bitmap Index Scan on idx_t2 (cost=0.00..79.16 rows=100 width=0) (actual time=6.986..6.986 rows=2944 loops=1)
Index Cond: (test2.cb @> '(5, 23, 908)'::cube)
Buffers: shared hit=3606
Planning Time: 0.074 ms
Execution Time: 8.921 ms
(10 rows)
参考链接：https://www.postgresql.org/docs/12/cube.html
