[
https://issues.apache.org/jira/browse/KUDU-1836?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15830686#comment-15830686
]
Todd Lipcon commented on KUDU-1836:
-----------------------------------
A couple numbers from a ~10GB tablet I analyzed from an internal use case which
has lots of UPSERTs. In particular, this workload has a lot of cases where rows
get upserted but the changed value is no different than the previous contents
of the row (so multiple deltas in a row are basically dupes and highly
compressible). This is obviously close to a best-case, but it's also not a
contrived use case (this is a real app).
{code}
Codec Total size of deltas
----------------------
NONE 10458MB
LZ4 413MB (25x)
GZIP 296MB (35x)
{code}
> Enable compression of delta files
> ---------------------------------
>
> Key: KUDU-1836
> URL: https://issues.apache.org/jira/browse/KUDU-1836
> Project: Kudu
> Issue Type: Improvement
> Components: perf, tablet
> Affects Versions: 1.3.0
> Reporter: Todd Lipcon
> Assignee: Todd Lipcon
>
> deltafiles are currently not compressed. This can be somewhat problematic for
> workloads that are heavy on updates -- the deltafiles can be much larger than
> the underlying data in many cases, especially if the user has a dataset which
> is highly column-compressible. We should consider enabling LZ compression of
> deltafiles by default.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
