[jira] [Commented] (IOTDB-1224) The query is pushed down to the DataGroup node.

["Chao Wang (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/IOTDB-1224?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17300082#comment-17300082 ] 

Chao Wang commented on IOTDB-1224:
----------------------------------

1. all push. When all sequences in a query statement are in a data group, the execution plan of the statement can be pushed down to the data group for processing.
2. Partial push. When the sequence in the query statement is distributed on different datagroups,
2.1 If the query is performed without a value, you can divide plans by data group, deliver operations in the same data group together, and combine the operations at the coordinator.
For example, select s1, s2 from root where time > 123
The execution plan can be split into select s1 from where time > 123.
The other is the execution plan of select s2 from where time > 123.
The stitching is then done at the coordinator.
Aggregate Query Same as above. select count(s1), count(s2) from root where time > 123
The execution plan AggPlan can be split into select count(s1) from where time > 123 and pushed to dataGroup.
The other is to push the execution plan of select count(s2) from where time > 123 to dataGroup.
The results are then spliced at the coordinator.
2.2 If the query contains value filtering, the existing optimization logic is used.

> The query is pushed down to the DataGroup node.
> -----------------------------------------------
>
>                 Key: IOTDB-1224
>                 URL: https://issues.apache.org/jira/browse/IOTDB-1224
>             Project: Apache IoTDB
>          Issue Type: Improvement
>          Components: Cluster
>            Reporter: Chao Wang
>            Assignee: Chao Wang
>            Priority: Major
>              Labels: pull-request-available
>             Fix For: 0.12.0
>
>
> Currently, in the aggregation query, the coordinator queries data from the data group in which the sequence is located, and then collects statistics. This increases the number of RPC times. We can push the statistics down to the data group for processing.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)