cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Benedict (JIRA)" <>
Subject [jira] [Commented] (CASSANDRA-8630) Faster sequential IO (on compaction, streaming, etc)
Date Wed, 09 Sep 2015 23:33:47 GMT


Benedict commented on CASSANDRA-8630:

So, looking at the code as I was about to commit, I think we should take another think about
buffer sizes. It looks like CASSANDRA-8894 had a quality I hadn't noticed, i.e. that it can
allocate > 64Kb buffers for reading. I'm pretty sure this is a bad thing (even reading
that much in one go is probably rarely a good idea; the goal was only to reduce this number
where we could safely do so). Secondly, that this patch introduces only a bound for this to
throttled readers. My understanding was for throttled readers we would _always_ use a 64Kb
buffer (since this makes quite a lot of sense, given it's sequential access of the whole file).

Can we impose the limit we have for throttled readers to _unthrottled_ readers, and special
case throttled to always return something large (64Kb being most sensible since right now
that's our largest possible cached size)? 

> Faster sequential IO (on compaction, streaming, etc)
> ----------------------------------------------------
>                 Key: CASSANDRA-8630
>                 URL:
>             Project: Cassandra
>          Issue Type: Improvement
>          Components: Core, Tools
>            Reporter: Oleg Anastasyev
>            Assignee: Stefania
>              Labels: compaction, performance
>             Fix For: 3.x
>         Attachments: 8630-FasterSequencialReadsAndWrites.txt, cpu_load.png, flight_recorder_001_files.tar.gz,
flight_recorder_002_files.tar.gz, mmaped_uncomp_hotspot.png
> When node is doing a lot of sequencial IO (streaming, compacting, etc) a lot of CPU is
lost in calls to RAF's int read() and DataOutputStream's write(int).
> This is because default implementations of readShort,readLong, etc as well as their matching
write* are implemented with numerous calls of byte by byte read and write. 
> This makes a lot of syscalls as well.
> A quick microbench shows than just reimplementation of these methods in either way gives
8x speed increase.
> A patch attached implements<Type> and SequencialWriter.write<Type>
methods in more efficient way.
> I also eliminated some extra byte copies in CompositeType.split and ColumnNameHelper.maxComponents,
which were on my profiler's hotspot method list during tests.
> A stress tests on my laptop show that this patch makes compaction 25-30% faster  on uncompressed
sstables and 15% faster for compressed ones.
> A deployment to production shows much less CPU load for compaction. 
> (I attached a cpu load graph from one of our production, orange is niced CPU load - i.e.
compaction; yellow is user - i.e. not compaction related tasks)

This message was sent by Atlassian JIRA

View raw message