Mailing-List: contact dev-help@hbase.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@hbase.apache.org
Received-SPF: pass (athena.apache.org: local policy)
DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws;
  s=s1024; d=yahoo.com;
  h=X-YMail-OSG:Received:X-Rocket-MIMEInfo:X-Mailer:References:Message-ID:Date:From:Reply-To:Subject:To:In-Reply-To:MIME-Version:Content-Type;
  b=2a/ZqSE1EuRmiPyzWjUp688Xbj8uKS6Ls7lXHSqFZSDlwS5lVO1jKOo4Ohia8Q4rZnlozOoLs7PSU5wO3HWRgPNZ6guEFISx9BrFxE5s410v1DzZnkVinfIw1M0R7xdKcErl7kuzE3O9s7n3iGtNcx0zJu12/sRPOb9OVtc4UpQ=;
References: 
 <CAPcDmSt9mp5kE8_DXvMUmONX3ipOF-9Zggv0xZnQ_Co1+5PseA@mail.gmail.com>
 <1354041666.83689.YahooMailNeo@web140606.mail.bf1.yahoo.com>
 <CAPcDmSswA0deS_CwuZyC2HLM2F0RnSXBwm_A168FdDg5Wx6nVg@mail.gmail.com>
 <1354140148.62975.YahooMailNeo@web140605.mail.bf1.yahoo.com>
 <CAPcDmSs+YQnK8G_fnvGqpw3C7B4HrcsgEh0++zvvT32-g7dcrg@mail.gmail.com>
 <CAPcDmSt7VmqTj6e=5eWTHPZ6txp95GZOv7cUXZOz65QKz5URyA@mail.gmail.com>
Message-ID: <1354218713.36190.YahooMailNeo@web140606.mail.bf1.yahoo.com>
Date: Thu, 29 Nov 2012 11:51:53 -0800 (PST)
From: lars hofhansl <lhofhansl@yahoo.com>
Reply-To: lars hofhansl <lhofhansl@yahoo.com>
Subject: Re: small perf degradation in 0.94 trunk vs. older versions
To: "dev@hbase.apache.org" <dev@hbase.apache.org>
In-Reply-To: 
 <CAPcDmSt7VmqTj6e=5eWTHPZ6txp95GZOv7cUXZOz65QKz5URyA@mail.gmail.com>
MIME-Version: 1.0
Content-Type: multipart/alternative;
 boundary="1905101558-1917084383-1354218713=:36190"

--1905101558-1917084383-1354218713=:36190
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: quoted-printable

Thanks N!=0A(I didn't know it was so easy to reproduce)=0A=0A=0A=0A________=
________________________=0A From: Nicolas Liochon <nkeywal@gmail.com>=0ATo:=
 lars hofhansl <lhofhansl@yahoo.com> =0ACc: dev@hbase.apache.org =0ASent: T=
hursday, November 29, 2012 8:01 AM=0ASubject: Re: small perf degradation in=
 0.94 trunk vs. older versions=0A =0AHere are some tests on a standalone hb=
ase. I just create a table with 3000=0Aregions from the shell: create 't1',=
 'f1', {NUMREGIONS =3D> 3000, SPLITALGO=0A=3D> 'HexStringSplit'}=0A=0AIt se=
ems that's it's between august 20th and september 1st, but there could=0Abe=
 some small other stuff along the line as well...=0A=0Acommit eb36a3e410998=
e72d76edcc5c181dfa54bba1c39=0ADate:=A0  Tue Oct 2 20:39:14 2012 +0000=0A0 r=
ow(s) in 184.4040 seconds=0A0 row(s) in 191.4830 seconds=0A0 row(s) in 189.=
6370 seconds=0A0 row(s) in 198.9080 seconds=0A0 row(s) in 202.8180 seconds=
=0A=0Acommit 8c93fcfca2eb12162b99b8e1e327bab872bba6b7=0ADate:=A0  Wed May 3=
0 17:06:52 2012 +0000=0Acreate 't1', 'f1', {NUMREGIONS =3D> 3000, SPLITALGO=
 =3D> 'HexStringSplit'}=0A0 row(s) in 165.2160 seconds=0A0 row(s) in 142.78=
90 seconds=0A0 row(s) in 137.7260 seconds=0A=0Acommit 01f0b3c77ea30b1c22a0e=
96929e97bb9f5faab2bgit=0ADate:=A0  Thu Jul 19 22:19:49 2012 +0000=0A0 row(s=
) in 173.9160 seconds=0A0 row(s) in 155.0360 seconds=0A0 row(s) in 141.5210=
 seconds=0A0 row(s) in 140.6480 seconds=0A=0Acommit bc51a7267d2647630224f64=
6b62b80458591414a=0ADate:=A0  Sat Sep 1 04:35:29 2012 +0000=0A0 row(s) in 1=
61.1990 seconds=0A0 row(s) in 171.0010 seconds=0A0 row(s) in 167.2430 secon=
ds=0A0 row(s) in 189.9960 seconds=0A0 row(s) in 205.0460 seconds=0A0 row(s)=
 in 194.4270 seconds=0A0 row(s) in 191.9020 seconds=0A=0Acommit 5ea73b79928=
19ec6e24ba7c2c5beee306f7d12da=0ADate:=A0  Thu Aug 9 21:52:21 2012 +0000=0A0=
 row(s) in 164.1430 seconds=0A0 row(s) in 154.3410 seconds=0A0 row(s) in 15=
4.3370 seconds=0A0 row(s) in 142.6060 seconds=0A0 row(s) in 138.8110 second=
s=0A0 row(s) in 144.4190 seconds=0A=0Acommit 7179970b4a00ce630004d72e8e45e3=
0fa9f4881b=0ADate:=A0  Mon Aug 20 23:38:50 2012 +0000=0A0 row(s) in 146.455=
0 seconds=0A0 row(s) in 150.3020 seconds=0A0 row(s) in 143.6050 seconds=0A0=
 row(s) in 148.4000 seconds=0A0 row(s) in 164.8320 seconds=0A=0Acommit b4e5=
c3ae45935a37e1ab8651998c6a8131180eec=0ADate:=A0  Mon Aug 13 19:32:19 2012 +=
0000=0A0 row(s) in 151.0200 seconds=0A0 row(s) in 186.2860 seconds=0A0 row(=
s) in 141.6130 seconds=0A0 row(s) in 141.1880 seconds=0A0 row(s) in 140.932=
0 seconds=0A0 row(s) in 162.9540 seconds=0A0 row(s) in 139.5530 seconds=0A=
=0A=0AOn Thu, Nov 29, 2012 at 10:07 AM, Nicolas Liochon <nkeywal@gmail.com>=
 wrote:=0A=0A> I'm going to try with some random commits. Hopefully I will =
reproduce it a=0A> on standalone instance as well...=0A> Stay tuned :-)=0A>=
=0A>=0A>=0A> On Wed, Nov 28, 2012 at 11:02 PM, lars hofhansl <lhofhansl@yah=
oo.com>wrote:=0A>=0A>> Did a quick scan through the changes committed since=
 0.94.1:=0A>> HBASE-6608=0A>> HBASE-6364=0A>> HBASE-6587=0A>> HBASE-6537=0A=
>> HBASE-6713=0A>> HBASE-6438=0A>> HBASE-6299=0A>> HBASE-7018=0A>> HBASE-70=
38=0A>> HBASE-7060=0A>>=0A>> Any chance you can do this again with 0.94.2 (=
or even do a binary search=0A>> through the commits to pinpoint the change)=
?=0A>>=0A>> -- Lars=0A>>=0A>>=A0  ------------------------------=0A>> *From=
:* Nicolas Liochon <nkeywal@gmail.com>=0A>> *To:* dev@hbase.apache.org; lar=
s hofhansl <lhofhansl@yahoo.com>=0A>> *Sent:* Tuesday, November 27, 2012 10=
:42 AM=0A>> *Subject:* Re: small perf degradation in 0.94 trunk vs. older v=
ersions=0A>>=0A>> It was Version 0.94.3, r38dbd22c99debd9010e9e5f4fbabeeaf3=
c4e1ddd=0A>>=0A>>=0A>> On Tue, Nov 27, 2012 at 7:41 PM, lars hofhansl <lhof=
hansl@yahoo.com>wrote:=0A>>=0A>> Interesting. Thanks N.=0A>>=0A>> I'll look=
 through the 0.94 commits since 0.94.1 to see what's causing=0A>> this.=0A>=
> With 0.94 trunk you mean the 0.94.3RC?=0A>>=0A>>=0A>> -- Lars=0A>>=0A>>=
=0A>>=0A>> ________________________________=0A>>=A0 From: Nicolas Liochon <=
nkeywal@gmail.com>=0A>> To: dev@hbase.apache.org=0A>> Sent: Tuesday, Novemb=
er 27, 2012 5:07 AM=0A>> Subject: small perf degradation in 0.94 trunk vs. =
older versions=0A>>=0A>> Hi,=0A>>=0A>> On a create table / reassignment, I =
feel we lost some performances=0A>> recently=0A>> on 0.94. It's not huge (5=
% to 10%), but I would prefer them to be the=0A>> other=0A>> way around.=0A=
>> Here are some tests I've done on test cluster, all the time with Hadoop=
=0A>> 1.1.0;=0A>>=0A>> scenario: 2 RS. Create a table with 3000 regions.=0A=
>> Start a third RS. Kill -15 the second one: there are 1500 regions to=0A>=
> reassign.=0A>>=0A>> I've made multiple measures, there are all there:=0A>=
> It's in seconds.=0A>>=0A>> Creating a table with 3000 regions:=0A>> 0.92:=
 261s; 260s=0A>> 0.94.0: 260s; 260s=0A>> 0.94.1: 261s; 260s=0A>> 0.94 trunk=
: 292s; 281s; 282s;=0A>> 0.96 trunk: 173s; 178s=0A>>=0A>> Reassign after th=
e kill=0A>> 0.92: 107s, 110s=0A>> 0.94.0: 105s; 105s=0A>> 0.94.1: 107s; 107=
s=0A>> 0.94 trunk: 122s; 105s; 116s=0A>> 0.96 trunk: 50s; 50s;=0A>>=0A>> I =
don't know if there is a reason the the 0.94 trunk results...=0A>> The resu=
lts for 0.96s are actually not that great, my tests (on a single=0A>> machi=
ne) were much better a few months ago. I will look at that.=0A>>=0A>> Cheer=
s,=0A>>=0A>> N.=0A>>=0A>>=0A>>=0A>>=0A>>=0A>
--1905101558-1917084383-1354218713=:36190--