[ 
https://issues.apache.org/jira/browse/HBASE-16438?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15610669#comment-15610669
 ] 

Anoop Sam John commented on HBASE-16438:
----------------------------------------

Describing diff possibilities at that time.  We keep ref and offset means we 
can not keep whole of the index meta in off heap chunk.  chunkid+ offset we can 
easily keep as both are ints and we can keep ints in offheap BB.  But the ref 
means we need to have a ref array which is on heap.  And we will need another 
data structure (chunk only) to keep offset + length.   Yes we can avoid length 
when there are no tags.  But when tags are there, we need to keep length :-(. 
This is because tags parts comes after KL, VL, Key and Value.   Also missed one 
more thing at that time of discuss.  ie. we need to keep the seqId 8 bytes).. 
When cells in CSLM, it is a long state on the object.  Other way would be that 
when we copy Cell to MSLAB (or to a temp byte[]) at the time of addition to 
Memstore, we need to keep the seqId not as a state in obj. But that should be 
put as the last 8 bytes of the cell data bytes. (After key, value and tags).  
When MSLAB is off heap this will help us  to keep more data off heap.  We need 
a diff version of cell which can read seqId correctly from last 8 bytes. We 
will need decode it.  Will need lots of PoC work around diff ideas.  All these 
goes to other jira which says abt ChunkMap variant.

> 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. 
> Why we need to embed the chunkid in the Cell is described by [~anastas] in 
> this remark over in parent issue 
> https://issues.apache.org/jira/browse/HBASE-14921?focusedCommentId=15244119&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-15244119



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

Reply via email to