[
https://issues.apache.org/jira/browse/OMID-131?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16817232#comment-16817232
]
Ohad Shacham commented on OMID-131:
-----------------------------------
The reason is that now you have the same shadow cell for different columns.
When you do a get/scan in Omid, you ask to get the shadow cells for the rows
you are looking for. In the new scheme we will ask to get the row or row/family
shadow cell for the cells we are looking for and these might be in version
higher than the cell we are looking for.
Lets take the following example:
For row R with columns c1 and c2, assume we write R with c1 at version 1 and R
with c2 with version 2.
With the current scheme we will have:
R c1 c1_shadow - version 1
R c2 c2_shadow - version 2.
Reading c1 will return c1 and c1_shadow from version 1.
Changing the scheme will results in:
R c1 R_shadow - version 1
R c2 R_shadow - version 2.
Reading c1 will return c1 from version 1 and R_shadow from version 2 and we
will need to reread R_shadow again to find R_shadow from version 1.
> Enhance row level conlict detection to reduce client memory footprint.
> ----------------------------------------------------------------------
>
> Key: OMID-131
> URL: https://issues.apache.org/jira/browse/OMID-131
> Project: Apache Omid
> Issue Type: Improvement
> Affects Versions: 1.0.1
> Reporter: Yonatan Gottesman
> Assignee: Yonatan Gottesman
> Priority: Major
> Time Spent: 20m
> Remaining Estimate: 0h
>
> Currently even if omid is working in row level conflict detection, all the
> cells are kept by the client until after commit for writing shadow cells or
> removing them after abort.
> Instead the client should store only the rows or only the family.
> If we choose to store only the row, the client memory footprint will be
> smaller, but if a region server wants to query a shadow cell for a cell it
> might have to query a different region.
> On the other hand client can store the famelies of a writeset. This will have
> a bigger footprint if we have lots of families but if a region queries a
> shadow cell it will be on the same region.
>
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
