[ 
https://issues.apache.org/jira/browse/IMPALA-7034?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Vincent Tran updated IMPALA-7034:
---------------------------------
    Description: 
Currently the practical limit for catalog topic update is in the neighborhood 
of 4GB - the fundamental limit of max thrift message size. This is an 
architectural limitation and not a resource limitation

Larger enterprise clusters with high file counts can easily surpass this with 
normal usage.
The high level ask here is for a more scalable implementation for metadata 
handling. The amount metadata that a cluster can handle should be proportional 
to the amount of hardware resource that an user is willing to allocate to it.

  was:
Currently the practical limit for catalog topic update is in the neighborhood 
of 4GB - the fundamental limit of max thrift message size. This is an 
architectural limitation and not a resource limitation

Larger enterprise clusters with high file counts can easily surpass this with 
normal usage.
The high level ask here is for a more scalable implementation for metadata 
handling. The amount metadata that a cluster can handle should be proportional 
to the amount of hardware resource that a user is willing to allocate to it.


> Increase scalability of metadata handling
> -----------------------------------------
>
>                 Key: IMPALA-7034
>                 URL: https://issues.apache.org/jira/browse/IMPALA-7034
>             Project: IMPALA
>          Issue Type: Improvement
>    Affects Versions: Impala 2.13.0
>            Reporter: Vincent Tran
>            Priority: Major
>
> Currently the practical limit for catalog topic update is in the neighborhood 
> of 4GB - the fundamental limit of max thrift message size. This is an 
> architectural limitation and not a resource limitation
> Larger enterprise clusters with high file counts can easily surpass this with 
> normal usage.
> The high level ask here is for a more scalable implementation for metadata 
> handling. The amount metadata that a cluster can handle should be 
> proportional to the amount of hardware resource that an user is willing to 
> allocate to it.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-all-unsubscr...@impala.apache.org
For additional commands, e-mail: issues-all-h...@impala.apache.org

Reply via email to