[
https://issues.apache.org/jira/browse/HBASE-23279?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16974596#comment-16974596
]
Lars Hofhansl edited comment on HBASE-23279 at 11/15/19 2:44 AM:
-----------------------------------------------------------------
Thanks all.
So based on the discussion here ... Is already a problem if I went and changed
the block encoding to anything other than NONE? Most block encodings (like
FAST_DIFF, etc) will decrease the size, but even there there're abnormal cases
where the size might be increased.
Or in other words any encoding (or compression) will cause the actual size of a
block to not be a constant.
Other cases are large key values. The block is extended at the end to hold the
last key value, right?
NM: Read the above again. I think we do not have to change the formula. How
much bigger the index encoded file is depends on the type of the data.
was (Author: lhofhansl):
Thanks all.
So based on the discussion here ... Is already a problem if I went and changed
the block encoding to anything other than NONE? Most block encodings (like
FAST_DIFF, etc) will decrease the size, but even there there're abnormal cases
where the size might be increased.
Or in other words any encoding (or compression) will cause the actual size of a
block to not be a constant.
Other cases are large key values. The block is extended at the end to hold the
last key value, right?
> Switch default block encoding to ROW_INDEX_V1
> ---------------------------------------------
>
> Key: HBASE-23279
> URL: https://issues.apache.org/jira/browse/HBASE-23279
> Project: HBase
> Issue Type: Wish
> Reporter: Lars Hofhansl
> Assignee: Viraj Jasani
> Priority: Minor
>
> Currently we set both block encoding and compression to NONE.
> ROW_INDEX_V1 has many advantages and (almost) no disadvantages (the hfiles
> are slightly larger about 3% or so). I think that would a better default than
> NONE.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
