cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Paul Loy (JIRA)" <>
Subject [jira] [Commented] (CASSANDRA-2820) Re-introduce FastByteArrayInputStream (and Output equivalent)
Date Thu, 07 Jul 2011 08:11:16 GMT


Paul Loy commented on CASSANDRA-2820:

> I'm actually not too much of a fan of extending ByteArray*Stream but redeclaring all
the fields, giving "fake" argument to the super constructor. It's a bit ugly and not too optimal
in memory size...

Haha, me too!

> it may be simpler/cleaner to make FastByteArray*Stream extends Input/OutputStream directly

I prefer this too.

> my earlier comment was more about avoiding to use FastByteArray*Stream in more places
than strictly necessary

Sure. I guess this was lost in translation :S

I'll update the patch.

> Re-introduce FastByteArrayInputStream (and Output equivalent)
> -------------------------------------------------------------
>                 Key: CASSANDRA-2820
>                 URL:
>             Project: Cassandra
>          Issue Type: Improvement
>          Components: Core
>    Affects Versions: 0.8.0
>         Environment: n/a
>            Reporter: Paul Loy
>            Priority: Minor
>              Labels: bytearrayinputstream, bytearrayoutputstream, license, synchronized
>             Fix For: 1.0
>         Attachments: fast_bytearray_iostreams_harmony-patch-2.txt, fast_bytearray_iostreams_harmony-patch-3.txt
> In FastByteArrayInputStream and FastByteArrayOutputStream
were removed due to being code copied from the JDK and then subsequently modified. The JDK
license is incompatible with Apache 2 license so the code had to go.
> I have since had a look at the performance of the JDK ByteArrayInputStream and a FastByteArrayInputStream
(i.e. one with synchronized methods made un-synchronized) and seen the difference is significant.
> After a warmup-period of >10000 loops I get the following for 10000 loops through
a 128000 byte array:
> bais : 3513ms
> fbais: 72ms
> This varies depending on the OS, machine and Java version, but it's always in favour
of the FastByteArrayInputStream as you might expect.
> Then, at Jonathan Ellis' suggestion, I tried this using a modified Apache Harmony ByteArrayInputStream
- i.e. one whose license is compatible - and the results were the same. A significant boost.
> I will attach a patch with changes for the 0.8.0 tag.

This message is automatically generated by JIRA.
For more information on JIRA, see:


View raw message