Liya Fan created ARROW-13733:
--------------------------------
Summary: [Java] Allow JDBC adapters to reuse vector schema roots
Key: ARROW-13733
URL: https://issues.apache.org/jira/browse/ARROW-13733
Project: Apache Arrow
Issue Type: New Feature
Components: Java
Reporter: Liya Fan
Assignee: Liya Fan
According to the current design of the JDBC adapter, it is not possible to
reuse the vector schema roots. That is, a new vector schema root is created and
released for each batch.
This can cause performance problems, because in many scenarios, the client code
only reads data in vector schema root. So the vector schema roots can be reused
in the following cycle: populate data -> client use data -> populate data -> ...
The current design has another problem. For most times, it has two alternating
vector schema roots in memory, causing a large waste of memory, especially for
large batches.
We solve both problems by providing a flag in the config, which allows the user
to reuse the vector shema roots.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
