You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hbase.apache.org by "Anoop Sam John (JIRA)" <ji...@apache.org> on 2016/10/20 08:57:58 UTC

[jira] [Comment Edited] (HBASE-16438) Create a cell type so that chunk id is embedded in it

    [ https://issues.apache.org/jira/browse/HBASE-16438?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15591255#comment-15591255 ] 

Anoop Sam John edited comment on HBASE-16438 at 10/20/16 8:57 AM:
------------------------------------------------------------------

It might be better if this is a new Cell type with a new interface implemented.  Adding a tag is having more overhead wrt memory usage.
This is needed in another place as well.
When we upsert cell into Memstore, we will remove the cells which is newly added cells mask out.  And as part of that we will adjust the memstore size (region level as well as global).  Here we just reduce the size of the removed cell now.  But we cannot do that blindly.  When this removed cell is having its data bytes in MSLAB, we are not really freeing up those MSLAB chunk bytes.  Means the cell data size part is still not reduced!  We just got rid of one Java object. So only that object's overhead part got reduced.  Right now there is no way to make this diff.  When the cell can be removed, we can not know whether it is having its data in MSLAB chunk or not.   Once this jira is solved, we can do that also.



was (Author: anoop.hbase):
It might be better if this is a new Cell type with a new interface implemented. 
This is needed in another place as well.
When we upsert cell into Memstore, we will remove the cells which is newly added cells mask out.  And as part of that we will adjust the memstore size (region level as well as global).  Here we just reduce the size of the removed cell now.  But we cannot do that blindly.  When this removed cell is having its data bytes in MSLAB, we are not really freeing up those MSLAB chunk bytes.  Means the cell data size part is still not reduced!  We just got rid of one Java object. So only that object's overhead part got reduced.  Right now there is no way to make this diff.  When the cell can be removed, we can not know whether it is having its data in MSLAB chunk or not.   Once this jira is solved, we can do that also.


> Create a cell type so that chunk id is embedded in it
> -----------------------------------------------------
>
>                 Key: HBASE-16438
>                 URL: https://issues.apache.org/jira/browse/HBASE-16438
>             Project: HBase
>          Issue Type: Sub-task
>    Affects Versions: 2.0.0
>            Reporter: ramkrishna.s.vasudevan
>            Assignee: ramkrishna.s.vasudevan
>
> For CellChunkMap we may need a cell such that the chunk out of which it was created, the id of the chunk be embedded in it so that when doing flattening we can use the chunk id as a meta data. More details will follow once the initial tasks are completed. 



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)