Mailing-List: contact issues-help@hbase.apache.org; run by ezmlm
Precedence: bulk
Date: Thu, 29 Dec 2011 03:56:31 +0000 (UTC)
From: "Mikhail Bautin (Commented) (JIRA)" <jira@apache.org>
To: issues@hbase.apache.org
Message-ID: 
 <1988430134.50367.1325130991725.JavaMail.tomcat@hel.zones.apache.org>
In-Reply-To: 
 <657671452.46177.1313609127186.JavaMail.tomcat@hel.zones.apache.org>
Subject: [jira] [Commented] (HBASE-4218) Data Block Encoding of KeyValues
  (aka delta encoding / prefix compression)
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/HBASE-4218?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13176976#comment-13176976 ] 

Mikhail Bautin commented on HBASE-4218:
---------------------------------------

Just a quick note from an offline conversation with Kannan: we need to support modifying data block encoding column family settings. In the most recent version of the patch (https://reviews.facebook.net/D447?vs=&id=3237&whitespace=ignore-all) there are the following user-facing column family settings:

* DATA_BLOCK_ENCODING - specifies data block encoding type or NONE
* ENCODE_IN_CACHE_ONLY - boolean (false by default). If true, data blocks are only encoded in cache but not on disk

We removed the "encoded scanner" flag, and we use encoded scanners by default any time we use data block encoding.

Given the above column family settings, we need to unit-test at least the following transitions:
# Switching from no data block encoding to a data block encoding everywhere, and vice versa
# Switching from no data block encoding to a data block encoding in cache only, and vice versa
# Flipping the "in cache only" flag but keeping the data block encoding type the same
# Switching from one data block encoding everywhere to another one
# Switching from one data block encoding in cache only to another one
# Switching to a different data block encoding and flipping the "in cache only" flag.

                
> Data Block Encoding of KeyValues  (aka delta encoding / prefix compression)
> ---------------------------------------------------------------------------
>
>                 Key: HBASE-4218
>                 URL: https://issues.apache.org/jira/browse/HBASE-4218
>             Project: HBase
>          Issue Type: Improvement
>          Components: io
>    Affects Versions: 0.94.0
>            Reporter: Jacek Migdal
>            Assignee: Mikhail Bautin
>              Labels: compression
>             Fix For: 0.94.0
>
>         Attachments: 0001-Delta-encoding-fixed-encoded-scanners.patch, 0001-Delta-encoding.patch, D447.1.patch, D447.10.patch, D447.11.patch, D447.12.patch, D447.13.patch, D447.14.patch, D447.15.patch, D447.2.patch, D447.3.patch, D447.4.patch, D447.5.patch, D447.6.patch, D447.7.patch, D447.8.patch, D447.9.patch, Data-block-encoding-2011-12-23.patch, Delta-encoding.patch-2011-12-22_11_52_07.patch, Delta_encoding_with_memstore_TS.patch, open-source.diff
>
>
> A compression for keys. Keys are sorted in HFile and they are usually very similar. Because of that, it is possible to design better compression than general purpose algorithms,
> It is an additional step designed to be used in memory. It aims to save memory in cache as well as speeding seeks within HFileBlocks. It should improve performance a lot, if key lengths are larger than value lengths. For example, it makes a lot of sense to use it when value is a counter.
> Initial tests on real data (key length = ~ 90 bytes , value length = 8 bytes) shows that I could achieve decent level of compression:
>  key compression ratio: 92%
>  total compression ratio: 85%
>  LZO on the same data: 85%
>  LZO after delta encoding: 91%
> While having much better performance (20-80% faster decompression ratio than LZO). Moreover, it should allow far more efficient seeking which should improve performance a bit.
> It seems that a simple compression algorithms are good enough. Most of the savings are due to prefix compression, int128 encoding, timestamp diffs and bitfields to avoid duplication. That way, comparisons of compressed data can be much faster than a byte comparator (thanks to prefix compression and bitfields).
> In order to implement it in HBase two important changes in design will be needed:
> -solidify interface to HFileBlock / HFileReader Scanner to provide seeking and iterating; access to uncompressed buffer in HFileBlock will have bad performance
> -extend comparators to support comparison assuming that N first bytes are equal (or some fields are equal)
> Link to a discussion about something similar:
> http://search-hadoop.com/m/5aqGXJEnaD1/hbase+windows&subj=Re+prefix+compression

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira