[
https://issues.apache.org/jira/browse/CASSANDRA-5863?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13922619#comment-13922619
]
Benedict edited comment on CASSANDRA-5863 at 3/6/14 3:09 PM:
-------------------------------------------------------------
I wonder if storing decompressed chunks on SSD really makes much sense? It's
likely the decompression is still faster than IO to even a fast flash drive.
Moving hot compressed chunks onto an SSD makes a lot of sense, but think that
maybe these are two different tickets? One to bring the page/buffer cache in
process, and store it uncompressed, the other to track hot file regions and
store them on a cache drive.
One possible neat thing to try doing as well might be to have a mixed
compressed/uncompressed in-memory page cache. i.e. have a smaller uncompressed
page cache, into which pages are moved from a larger compressed page cache
(being removed from there at the same time), and when evicted the data is
recompressed and moved back. This is probably only helpful for non-SSD boxes,
though, as reading may be faster than recompressing.
was (Author: benedict):
I wonder if storing decompressed chunks on SSD really makes much sense? It's
likely the decompression is still faster than IO to even a fast flash drive.
Moving hot compressed chunks onto an SSD makes a lot of sense, but think that
maybe these are two different tickets? One to bring the page/buffer cache in
process, and store it uncompressed, the other to track hot file regions and
store them on a cache drive.
One possible neat thing to try doing as well might be to have a mixed
compressed/uncompressed in-memory page cache. i.e. have a smaller uncompressed
page cache, into which pages are moved from a larger compressed page cache
(being removed from their at the same time), and when evicted the data is
recompressed and moved back. This is probably only helpful for non-SSD boxes,
though, as reading may be faster than recompressing.
> Create a Decompressed Chunk [block] Cache
> -----------------------------------------
>
> Key: CASSANDRA-5863
> URL: https://issues.apache.org/jira/browse/CASSANDRA-5863
> Project: Cassandra
> Issue Type: New Feature
> Components: Core
> Reporter: T Jake Luciani
> Assignee: Pavel Yaskevich
> Labels: performance
> Fix For: 2.1 beta2
>
>
> Currently, for every read, the CRAR reads each compressed chunk into a
> byte[], sends it to ICompressor, gets back another byte[] and verifies a
> checksum.
> This process is where the majority of time is spent in a read request.
> Before compression, we would have zero-copy of data and could respond
> directly from the page-cache.
> It would be useful to have some kind of Chunk cache that could speed up this
> process for hot data. Initially this could be a off heap cache but it would
> be great to put these decompressed chunks onto a SSD so the hot data lives on
> a fast disk similar to https://github.com/facebook/flashcache.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
