hbase-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ted Yu <yuzhih...@gmail.com>
Subject Re: VOTE: hbase-0.95RC1, the second "Development" Series Release is available [WAS -> VOTE: hbase-0.95.0RC0, the first "Developer Release" release candidate is available for download and vote]
Date Thu, 04 Apr 2013 16:48:30 GMT
One thing I don't understand is why there is noticeable difference in
performance between 0.95.0 and trunk.
I think their code should be very close to each other.


On Thu, Apr 4, 2013 at 8:20 AM, Jean-Marc Spaggiari <jean-marc@spaggiari.org
> wrote:

> I ran the performances tests against 0.95.0RC1.
>
> While filteredScan, read and writes got nice improvements (18% faster
> for the reads), scans seems to be negatively impacted (32% slower for
> scanRange100).
>
> Results are there:
> http://www.spaggiari.org/media/blogs/hbase/pdf/performances_20130404.pdf
>
> I have just restarted scanRange100 to validate the numbers.
>
> JM
>
> 2013/4/3 Andrew Purtell <apurtell@apache.org>:
> > I think that sounds good.
> >
> > On Wednesday, April 3, 2013, Stack wrote:
> >
> >> On Wed, Apr 3, 2013 at 10:55 AM, Andrew Purtell <apurtell@apache.org
> <javascript:;>>
> >> wrote:
> >>
> >> > Then what if we do not provide native binary convenience artifacts and
> >> > instead print an INFO level log message, should they be missing, which
> >> > points to a book chapter on compiling and installing them?
> >> >
> >> >
> >> Yes.  For now, I could just fill out the section in refguide on mlockall
> >> since that the only native lib we ship?
> >> St.Ack
> >>
> >
> >
> > --
> > Best regards,
> >
> >    - Andy
> >
> > Problems worthy of attack prove their worth by hitting back. - Piet Hein
> > (via Tom White)
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message