[
https://issues.apache.org/jira/browse/IMPALA-13019?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Wenzhe Zhou updated IMPALA-13019:
---------------------------------
Description:
IMPALA-12910 adds a new class DataSourceObjectCache to save DBCP DataSource
objects. DBCP DataSource objects hold JDBC driver. By sharing DBCP DataSource
objects across multiple JDBC requests, we can avoid to load JDBC driver.
Current code checks reference count for each DBCP DataSource object, and
removes the cached object once reference count reaches 0.
DBCP DataSource objects should be kept in cache even reference count reaches 0.
We could add a query option to enable it. This need a working thread to
periodically clean up objects which are idle for long time. Maybe separate maps
to save active objects and idle objects. Query options could be passed to JDBC
code in TOpenParams.
was:
IMPALA-12910 adds a new class DataSourceObjectCache to save DBCP DataSource
objects. DBCP DataSource objects hold JDBC driver. By sharing DBCP DataSource
objects across multiple JDBC requests, we can avoid to load JDBC driver.
Current code checks reference count for each DBCP DataSource object, and
removes the cached object once reference count reaches 0.
DBCP DataSource objects should be kept in cache even reference count reaches 0.
We could add a query option to enable it. This need a working thread to
periodically clean up objects which are idle for long time. Maybe separate maps
to save active objects and idle objects.
> Add query option to keep DBCP DataSource objects in cache for longer time
> -------------------------------------------------------------------------
>
> Key: IMPALA-13019
> URL: https://issues.apache.org/jira/browse/IMPALA-13019
> Project: IMPALA
> Issue Type: Sub-task
> Components: Frontend
> Reporter: Wenzhe Zhou
> Assignee: Pranav Yogi Lodha
> Priority: Major
>
> IMPALA-12910 adds a new class DataSourceObjectCache to save DBCP DataSource
> objects. DBCP DataSource objects hold JDBC driver. By sharing DBCP
> DataSource objects across multiple JDBC requests, we can avoid to load JDBC
> driver. Current code checks reference count for each DBCP DataSource object,
> and removes the cached object once reference count reaches 0.
> DBCP DataSource objects should be kept in cache even reference count reaches
> 0. We could add a query option to enable it. This need a working thread to
> periodically clean up objects which are idle for long time. Maybe separate
> maps to save active objects and idle objects. Query options could be passed
> to JDBC code in TOpenParams.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
