You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@iotdb.apache.org by "Yanzhe An (Jira)" <ji...@apache.org> on 2020/03/17 08:32:00 UTC

[jira] [Created] (IOTDB-559) Optimize the working mechanism of MemTable

Yanzhe An created IOTDB-559:
-------------------------------

             Summary: Optimize the working mechanism of MemTable
                 Key: IOTDB-559
                 URL: https://issues.apache.org/jira/browse/IOTDB-559
             Project: Apache IoTDB
          Issue Type: Improvement
            Reporter: Yanzhe An


At present, when a MemTable is full, it will be submitted to wait for flushing and the latest flushed time will be updated. This approach will probably generate some unsequence data caused by network transmission. For example, we set the size of MemTable as 5 and we insert a series of data <1,1> <2,2> <3,3> <4,4> <5,5> <6,6> <7,7> into IoTDB. Unfortunately, because of transmission delay, the sequence of data that IoTDB receives is <1,1> <2,2> <3,3> <7,7> <5,5> <6,6> <4,4>. According to the current procedure of inserting, one MemTable will be filled with <1,1> <2,2> <3,3> <7,7> <5,5> and the latest flushed time will be updated to 7 so that <6,6> and <4,4> will be regarded as unsequence data. Actually we can totally avoid to produce this kind of data by optimizing the working mechanism of MemTable.



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