Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

共进行三种查询:带值过滤的原始数据查询;带值过滤的聚合查询(Count和Max_Value两种),带值过滤的group by查询(Count和Max_Value两种+10-1000不同粒度)

以在集种分布下选择率为0的查询为例:

select s0 from root.sg1.d1 where s0<0
select s0 from root.sg1.d1 where s0<0
select max_value(s0) from root.sg1.d1 where s0<0
select count(s0) from root.sg1.d1 where s0<0 group by ([1, 10000000), 10ms)
select max_value(s0) from root.sg1.d1 where s0<0 group by ([1, 10000000), 10ms)
select count(s0) from root.sg1.d1 where s0<0 group by ([1, 10000000), 100ms)
select max_value(s0) from root.sg1.d1 where s0<0 group by ([1, 10000000), 100ms)
select count(s0) from root.sg1.d1 where s0<0 group by ([1, 10000000), 1000ms)
select max_value(s0) from root.sg1.d1 where s0<0 group by ([1, 10000000), 1000ms)

实验结果

如上表所示,可以发现,在进行原始数据查询和聚合查询时,当选择率较低的场景下,查询执行时间有明显的提升,选择率大于50%以后与原有时间持平。

...