Jialin Qiao created IOTDB-706:
---------------------------------
Summary: Introduce virtual memtable for larger Chunk
Key: IOTDB-706
URL: https://issues.apache.org/jira/browse/IOTDB-706
Project: Apache IoTDB
Issue Type: Improvement
Components: Core/Engine
Reporter: Jialin Qiao
When managing too many time series on a memory limited machine, the chunk of
each time series may be very small, such as 32 points, which impacts the query
performance.
Here is an idea:
Introducing virtual memtable(an unclosed TsFile on disk) for each TsFile. When
flushing a memtable, If the average chunk size of memtable + the average Chunk
size of virtual memtable does not reach the threshold, we flush the memtable
into the virtual memtable (if the virtual memtable exists, we merge the virtual
memtable with the memtable first).
Otherwise, we merging the virtual memtable with the flushing memtable into the
origin target tsfile.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
