[
https://issues.apache.org/jira/browse/IOTDB-840?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17191973#comment-17191973
]
Xiangdong Huang commented on IOTDB-840:
---------------------------------------
In memory (MeasurementNode), for each time series, we have 3 Enum instances:
data type, encoding method, compression method, which at least takes 3 * 8
bytes memory.
Suppose we have 10 Million timeseries, they will take about 240MB memory.
So, the idea is, encoding these 3 enums within 4 bytes. e.g., the first byte
represents data type, the second byte represents the encoding method, and the
third byterepresents the compression method, and the last bytes is reserved.
To keep the data clear and object-oriented, we can retain the method like
`DataType getDataType()` etc..
[~wyh]
> Encode the three enum fields in MeasurementSchema to save memory
> ----------------------------------------------------------------
>
> Key: IOTDB-840
> URL: https://issues.apache.org/jira/browse/IOTDB-840
> Project: Apache IoTDB
> Issue Type: Improvement
> Components: Core/Engine
> Reporter: Jialin Qiao
> Priority: Major
> Labels: easy-fix
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
