[ 
https://issues.apache.org/jira/browse/PHOENIX-7464?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17898841#comment-17898841
 ] 

Istvan Toth commented on PHOENIX-7464:
--------------------------------------

Thanks.

We need to cache, as optimizing HBaseConfiguration would happen in Hbase 3 at 
the earliest.

I have looked at ConnectionInfo, and we're modifying the Configuration when 
logging in into Kerberos,
so caching is not entirely straightforward. Will need to revisit the Kerberos 
parts.

> Performance Regression in Connection Initialization due to Configuration 
> Handling in ConnInfo
> ---------------------------------------------------------------------------------------------
>
>                 Key: PHOENIX-7464
>                 URL: https://issues.apache.org/jira/browse/PHOENIX-7464
>             Project: Phoenix
>          Issue Type: Bug
>            Reporter: Lokesh Khurana
>            Priority: Major
>
> The recent changes to {{ConnectionInfo}} (PHOENIX-6523) to resolve HBase 
> connection parameters earlier have introduced a performance regression
>  
> The root cause appears to be the heavy operation of creating an 
> {{HBaseConnection}} object. In contrast, Hadoop's implementation optimizes 
> configuration handling through efficient caching, making {{Configuration}} 
> object creation nearly cost-free.
> Possible solutions include:
>  # Pre-loading and caching the default {{HBaseConfiguration}} object in 
> {{PhoenixDriver}} or {{{}ConnectionInfo{}}}.
>  # Investigating why HBase does not use a similar caching mechanism as Hadoop 
> and implementing improvements accordingly.
>  



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to