Ismael, I pushed the benchmark code I used, with some updates (iteration: 20 -> 1000). I also updated the KIP page with the updated benchmark results. Please take a review when you are free. The attached screenshot shows how to run the benchmarker.
Thanks, Dongjin On Tue, Jan 10, 2017 at 8:03 PM, Dongjin Lee <dong...@apache.org> wrote: > Ismael, > > I see. Then, I will share the benchmark code I used by tomorrow. Thanks > for your guidance. > > Best, > Dongjin > > ----- > > Dongjin Lee > > Software developer in Line+. > So interested in massive-scale machine learning. > > facebook: www.facebook.com/dongjin.lee.kr > linkedin: kr.linkedin.com/in/dongjinleekr > github: github.com/dongjinleekr > twitter: www.twitter.com/dongjinleekr > > > > > On Tue, Jan 10, 2017 at 7:24 PM +0900, "Ismael Juma" <ism...@juma.me.uk> > wrote: > > Dongjin, >> >> The KIP states: >> >> "I compared the compressed size and compression time of 3 1kb-sized >> messages (3102 bytes in total), with the Draft-implementation of ZStandard >> Compression Codec and all currently available CompressionCodecs. All >> elapsed times are the average of 20 trials." >> >> But doesn't give any details of how this was implemented. Is the source >> code available somewhere? Micro-benchmarking in the JVM is pretty tricky so >> it needs verification before numbers can be trusted. A performance test >> with kafka-producer-perf-test.sh would be nice to have as well, if possible. >> >> Thanks, >> Ismael >> >> On Tue, Jan 10, 2017 at 7:44 AM, Dongjin Lee wrote: >> >> > Ismael, >> > >> > 1. Is the benchmark in the KIP page not enough? You mean we need a whole >> > performance test using kafka-producer-perf-test.sh? >> > >> > 2. It seems like no major project is relying on it currently. However, >> > after reviewing the code, I concluded that at least this project has a good >> > test coverage. And for the problem of upstream tracking - although there is >> > no significant update on ZStandard to judge this problem, it seems not bad. >> > If required, I can take responsibility of the tracking for this library. >> > >> > Thanks, >> > Dongjin >> > >> > On Tue, Jan 10, 2017 at 7:09 AM, Ismael Juma wrote: >> > >> > > Thanks for posting the KIP, ZStandard looks like a nice improvement over >> > > the existing compression algorithms. A couple of questions: >> > > >> > > 1. Can you please elaborate on the details of the benchmark? >> > > 2. About https://github.com/luben/zstd-jni, can we rely on it? A few >> > > things >> > > to consider: are there other projects using it, does it have good test >> > > coverage, are there performance tests, does it track upstream closely? >> > > >> > > Thanks, >> > > Ismael >> > > >> > > On Fri, Jan 6, 2017 at 2:40 AM, Dongjin Lee wrote: >> > > >> > > > Hi all, >> > > > >> > > > I've just posted a new KIP "KIP-110: Add Codec for ZStandard >> > Compression" >> > > > for >> > > > discussion: >> > > > >> > > > https://cwiki.apache.org/confluence/display/KAFKA/KIP- >> > > > 110%3A+Add+Codec+for+ZStandard+Compression >> > > > >> > > > Please have a look when you are free. >> > > > >> > > > Best, >> > > > Dongjin >> > > > >> > > > -- >> > > > *Dongjin Lee* >> > > > >> > > > >> > > > *Software developer in Line+.So interested in massive-scale machine >> > > > learning.facebook: www.facebook.com/dongjin.lee.kr >> > > > linkedin: >> > > > kr.linkedin.com/in/dongjinleekr >> > > > github: >> > > > github.com/dongjinleekr >> > > > twitter: www.twitter.com/dongjinleekr >> > > > * >> > > > >> > > >> > >> > >> > >> > -- >> > *Dongjin Lee* >> > >> > >> > *Software developer in Line+.So interested in massive-scale machine >> > learning.facebook: www.facebook.com/dongjin.lee.kr >> > linkedin: >> > kr.linkedin.com/in/dongjinleekr >> > github: >> > github.com/dongjinleekr >> > twitter: www.twitter.com/dongjinleekr >> > * >> > >> >> -- *Dongjin Lee* *Software developer in Line+.So interested in massive-scale machine learning.facebook: www.facebook.com/dongjin.lee.kr <http://www.facebook.com/dongjin.lee.kr>linkedin: kr.linkedin.com/in/dongjinleekr <http://kr.linkedin.com/in/dongjinleekr>github: <http://goog_969573159/>github.com/dongjinleekr <http://github.com/dongjinleekr>twitter: www.twitter.com/dongjinleekr <http://www.twitter.com/dongjinleekr>*
