Jialin Qiao created IOTDB-894:
---------------------------------
Summary: Optimize the MergeReader for the faked overlapped
BatchData
Key: IOTDB-894
URL: https://issues.apache.org/jira/browse/IOTDB-894
Project: Apache IoTDB
Issue Type: Improvement
Components: Core/Engine
Reporter: Jialin Qiao
If there are some pages, no matter sequence or unsequence [1, 99, 100], [2, 3,
4, 5]
the SeriesReader will treat the [1,99,100] and [2,3,4,5] as overlapped pages
and put them into PriorityMergeReader and merge data one by one.
Is there any possibility that we output [1], [2, 3, 4, 5], [99, 100], which
avoid puting the data of [2, 3, 4, 5] which actually not overlapped into the
heap.
The scenario is: we insert sequence pages [2,3,4] [6,7,8,9] [11,12,13,14]. only
little unsequence page [1, 5, 10, 15] . The unsequence page will destroy many
sequence pages.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
