我想了解phoenix的查询执行计划。我正在执行下面的命令以获取查询执行计划。
命令:-
explain select "col1","col2", count(*) as "Count" from "tab_name" group by "col1","col2";
说明计划:-
+----------------------------------------------------------------------------------------------------------+
| PLAN |
+----------------------------------------------------------------------------------------------------------+
|CLIENT 85-CHUNK PARALLEL 85-WAY FULL SCAN OVER tab_name |
| SERVER AGGREGATE INTO DISTINCT ROWS BY [d."col1", d."col2"] |
| CLIENT MERGE SORT |
+----------------------------------------------------------------------------------------------------------+
有谁能帮我理解Phoenix查询执行计划。
先谢谢你。
1条答案
按热度按时间ie3xauqp1#
您正在对85个单独的进程进行完全扫描(读取每一行)。在服务器端(每个节点),将在您选择的col1、col2行上运行一个聚合(distinct)
然后phoenix客户端将获取这些结果,应用合并排序以获得输出