[
https://issues.apache.org/jira/browse/HBASE-1249?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12680601#action_12680601
]
Erik Holstad commented on HBASE-1249:
-------------------------------------
Looks good!
A couple of things that I've been thinking about are:
1. For deletes I think it would make sense to not build the delete set
alongside but instead delete the column in the asked for column list. So when
going though memCache and the stores, when coming to a delete, just remove it
from the get list.
2. I think that the setting of the timestamp should not be exposed to the user
but, be set by HBase at all times. This limitation simplifies a lot when
reasoning about the order across storefiles and client formats, to make it as
fast, early outs and simple, list instead of map, as possible.
3. The client return type RowResult seems to be a pretty good general purpose
return format, but think we should have an option to get the list<ByteBuffer>
instead so that when you have knowledge about what kind of data that you have
in your system you can make the format much easier, for example if you are
asking for 1 row, 1 family, all columns and 1 version you only need map<qf,
value> or a list of tuples.
Looking at the implementation for the KeyValue there are a couple of things
that I want to bring up.
The Type has some unused bits that can be used in the future for different
flags, just to keep in mind.
I've added a new constructor family on my branch in KeyValue that deals with
separating the column into family and qualifier so that we don't need to look
for : and then add it back after we are done. If you still want the : for
printing purposes or what not you can put it in the family.
> Rearchitecting of server, client, API, key format, etc for 0.20
> ---------------------------------------------------------------
>
> Key: HBASE-1249
> URL: https://issues.apache.org/jira/browse/HBASE-1249
> Project: Hadoop HBase
> Issue Type: Improvement
> Reporter: Jonathan Gray
> Priority: Blocker
> Fix For: 0.20.0
>
>
> To discuss all the new and potential issues coming out of the change in key
> format (HBASE-1234): zero-copy reads, client binary protocol, update of API
> (HBASE-880), server optimizations, etc...
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
