You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@geronimo.apache.org by Matt Hogstrom <ma...@hogstrom.org> on 2007/10/23 05:51:33 UTC
Draft of 2.0.2 Performance Report
I've been noodling on this for a bit and wanted to give y'all a
gander at what I have for the performance report at this point. This
is based on 2.0.2 and uses DayTrader 2.0. There are a few numbers
that are missing. I originally had planned on not producing them but
the charts look odd with the missing numbers. It includes a
comparison of 1.1.1 and 2.0.2 using DayTrader 1.2 and 2.0. Heh, we
need to release those monsters.
This should be considered an alpha release but will move quickly to
final by the end of the week :)
Please provide your feedback on content, what's interesting, not, etc.
Thanks for taking a few minutes to look at the draft.
Look at http://people.apache.org/~hogstrom/performance/geronimo/2.0/
Geronimo2.0.2PerformanceReport-v01draft.pdf
Thanks
Re: Draft of 2.0.2 Performance Report
Posted by Matt Hogstrom <ma...@hogstrom.org>.
On Oct 23, 2007, at 10:43 AM, Christopher Blythe wrote:
> matt... just did an initial look. just a few comments for now...
>
> - were there funcational/load issues with the daytrader 1.2 numbers
> that were omitted?
Not really. The runs were clean, CPU was high and all the
fundamentals seemed to be correct. Was there something specific you
were thinking of?
> - was really surprised by the slow down in the web container
> primitives (probably has to do with the spec upgrade) and the jump
> in direct mode performance
The primary difference there is a new Tomcat version and perhaps some
changes to our integration. Later on I'd like to do some profiling
to better understand the issues but the slow down seems to be
consistent with other performance numbers I've seen. I think the
Linux Journal guys did something in this space but I'd have to go
back and look.
> - thanks for the kudos in the acknowledgements
Heh, thank you
> - yes, we need to tag 1.2 and 2.0 so we can start the next turn of
> the crank on 2.X
>
I'll start that process this week. Need to get the web pages updated
a bit as well. Lots of little stuff to do.
> chris
>
> On 10/22/07, Matt Hogstrom <ma...@hogstrom.org> wrote:
> I've been noodling on this for a bit and wanted to give y'all a
> gander at what I have for the performance report at this point. This
> is based on 2.0.2 and uses DayTrader 2.0. There are a few numbers
> that are missing. I originally had planned on not producing them but
> the charts look odd with the missing numbers. It includes a
> comparison of 1.1.1 and 2.0.2 using DayTrader 1.2 and 2.0. Heh, we
> need to release those monsters.
>
> This should be considered an alpha release but will move quickly to
> final by the end of the week :)
>
> Please provide your feedback on content, what's interesting, not, etc.
>
> Thanks for taking a few minutes to look at the draft.
>
> Look at http://people.apache.org/~hogstrom/performance/geronimo/2.0/
> Geronimo2.0.2PerformanceReport-v01draft.pdf
>
> Thanks
>
>
>
> --
> "I say never be complete, I say stop being perfect, I say let...
> lets evolve, let the chips fall where they may." - Tyler Durden
Re: Draft of 2.0.2 Performance Report
Posted by Christopher Blythe <cj...@gmail.com>.
matt... just did an initial look. just a few comments for now...
- were there funcational/load issues with the daytrader 1.2 numbers that
were omitted?
- was really surprised by the slow down in the web container primitives
(probably has to do with the spec upgrade) and the jump in direct mode
performance
- thanks for the kudos in the acknowledgements
- yes, we need to tag 1.2 and 2.0 so we can start the next turn of the crank
on 2.X
chris
On 10/22/07, Matt Hogstrom <ma...@hogstrom.org> wrote:
>
> I've been noodling on this for a bit and wanted to give y'all a
> gander at what I have for the performance report at this point. This
> is based on 2.0.2 and uses DayTrader 2.0. There are a few numbers
> that are missing. I originally had planned on not producing them but
> the charts look odd with the missing numbers. It includes a
> comparison of 1.1.1 and 2.0.2 using DayTrader 1.2 and 2.0. Heh, we
> need to release those monsters.
>
> This should be considered an alpha release but will move quickly to
> final by the end of the week :)
>
> Please provide your feedback on content, what's interesting, not, etc.
>
> Thanks for taking a few minutes to look at the draft.
>
> Look at http://people.apache.org/~hogstrom/performance/geronimo/2.0/
> Geronimo2.0.2PerformanceReport-v01draft.pdf
>
> Thanks
>
--
"I say never be complete, I say stop being perfect, I say let... lets
evolve, let the chips fall where they may." - Tyler Durden
Re: Draft of 2.0.2 Performance Report
Posted by Matt Hogstrom <ma...@hogstrom.org>.
Thanks. Amazing what you can read when you know the answer.
On Oct 23, 2007, at 1:25 PM, Piyush Agarwal wrote:
> Hi Matt,
>
> Page 9 mentions -
> SO F T WA R E
> Operating System:
> SuSE Enterprise Linux Enterprise SP1
> 2.6.16.46-0.4-smp #1 SMP Mon Apr 2 17:59:08 UTC
> 2007 x86_64 x86_64 x86_64 GNU/Linux
> The OS version is missing "SuSE Enterprise Linux Enterprise 10 SP1"
>
> As for JIBE, If you have the .log files which go with the xml files
> its easy as they have the complete JIBE output captured verbatim
> which you can copy paste in the report. I dont know any automated
> way of getting the numbers from the XML... seems it does summarize
> the run at the end of then file under the <HTTP_summary> tag.
>
>
> On 10/23/07, Matt Hogstrom <ma...@hogstrom.org> wrote:
>
> On Oct 23, 2007, at 10:53 AM, Piyush Agarwal wrote:
>
>> Hi Matt,
>>
>> This is a great report .. thanks for taking the time to create it.
>>
>> Here is my feedback on it-
>> Run spellcheck :-p there were a few typos here and there like
>> "enough" spelled "ebnough"
> what is a spell checker ?
>>
>> The Introduction on page 3 takes about PT, that was a metric you
>> used in the last report, but in this report you are comparing
>> against G1.1.1. The PT is not being used or mentioned throughout
>> the remaining of the report.
> thanks for catching this. PT is left over from the initial
> report. I'll fix that.
>
>>
>> Please mention if you need RAIDs or Ramdisks for the DB
> K
>
>> I think you forgot the mention the OS version... I assume its SLES
>> "10" SP1
> Page 9
>> Quite surprised that we go from 70M unzipped to 111M after initial
>> startup (whats the big hitter here.. the logs??)
>
>>
>> The URL at the end of page 8 (more info about daytrader) seems to
>> be incorrect and the link doesnt match the URL
>> the slowdown in the web primitives is a bit surprising... and yet
>> the jump in direct mode numbers is pretty cool
> The mysteries of life ... it wierded me out as well. I think it
> needs to be looked into but it is an external measurement so I
> figure putting out the info as it is with all the caveats is better
> than waiting another 6 months :)
>>
>> would have loved to see the Trade Session2JDBC numbers as well
> could due that
>> pg 14, 2nd para, typo "due" --> "do"
> doh
>> can we have an appendix at the end of the report with individual
>> run results like last time?
> Maybe...I have the XML files from JIBE but there doesn't seem to be
> a way to regerenate a nice summary...do you know how to do this ?
>>
>> HTH,
>> Piyush Agarwal
>>
>> On 10/22/07, Matt Hogstrom < matt@hogstrom.org > wrote:
>> I've been noodling on this for a bit and wanted to give y'all a
>> gander at what I have for the performance report at this point. This
>> is based on 2.0.2 and uses DayTrader 2.0. There are a few numbers
>> that are missing. I originally had planned on not producing them but
>> the charts look odd with the missing numbers. It includes a
>> comparison of 1.1.1 and 2.0.2 using DayTrader 1.2 and 2.0. Heh, we
>> need to release those monsters.
>>
>> This should be considered an alpha release but will move quickly to
>> final by the end of the week :)
>>
>> Please provide your feedback on content, what's interesting, not,
>> etc.
>>
>> Thanks for taking a few minutes to look at the draft.
>>
>> Look at http://people.apache.org/~hogstrom/performance/geronimo/2.0/
>> Geronimo2.0.2PerformanceReport-v01draft.pdf
>>
>> Thanks
>>
>
>
Re: Draft of 2.0.2 Performance Report
Posted by Forrest Xia <fo...@gmail.com>.
I think that article details the method about how to perform an out-of-box
performance testing on JEE 5 compliant java application servers, including G
2.2. User can follow it to do a simple and quick benchmark, and make choice
accordingly.
Forrest
Re: Draft of 2.0.2 Performance Report
Posted by frapien <fr...@gmx.de>.
are there any plans to update the Performance Report for Geronimo 2.2?
IBM has writen a performance benchmark article for the DayTrader Apps
comparing JBoss with Geronimo
http://www.ibm.com/developerworks/opensource/library/os-perfbenchmk/index.htm
--
View this message in context: http://old.nabble.com/Draft-of-2.0.2-Performance-Report-tp13357025s134p26525262.html
Sent from the Apache Geronimo - Dev mailing list archive at Nabble.com.
Re: Draft of 2.0.2 Performance Report
Posted by Piyush Agarwal <pv...@gmail.com>.
Unfortunately I dont have any such script for parsing the xml... if you have
used a sync-engine for the 2 jibe machines then its log files might have
merged data.
On 10/23/07, Matt Hogstrom <ma...@hogstrom.org> wrote:
>
>
> >
> > As for JIBE, If you have the .log files which go with the xml files
> > its easy as they have the complete JIBE output captured verbatim
> > which you can copy paste in the report. I dont know any automated
> > way of getting the numbers from the XML... seems it does summarize
> > the run at the end of then file under the <HTTP_summary> tag.
> >
>
> I have the log files but used two drivers so they are individuals. I
> have the XML files which were merged. If you have a script to
> transform the XML doc into a readable format I'd love to use it and
> include the info.
>
>
>
Re: Draft of 2.0.2 Performance Report
Posted by Matt Hogstrom <ma...@hogstrom.org>.
>
> As for JIBE, If you have the .log files which go with the xml files
> its easy as they have the complete JIBE output captured verbatim
> which you can copy paste in the report. I dont know any automated
> way of getting the numbers from the XML... seems it does summarize
> the run at the end of then file under the <HTTP_summary> tag.
>
I have the log files but used two drivers so they are individuals. I
have the XML files which were merged. If you have a script to
transform the XML doc into a readable format I'd love to use it and
include the info.
Re: Draft of 2.0.2 Performance Report
Posted by Piyush Agarwal <pv...@gmail.com>.
Hi Matt,
Page 9 mentions -
SO F T WA R E
Operating System:
SuSE Enterprise Linux Enterprise SP1
2.6.16.46-0.4-smp #1 SMP Mon Apr 2 17:59:08 UTC
2007 x86_64 x86_64 x86_64 GNU/Linux
The OS version is missing "SuSE Enterprise Linux Enterprise 10 SP1"
As for JIBE, If you have the .log files which go with the xml files its easy
as they have the complete JIBE output captured verbatim which you can copy
paste in the report. I dont know any automated way of getting the numbers
from the XML... seems it does summarize the run at the end of then file
under the <HTTP_summary> tag.
On 10/23/07, Matt Hogstrom <ma...@hogstrom.org> wrote:
>
>
> On Oct 23, 2007, at 10:53 AM, Piyush Agarwal wrote:
>
> Hi Matt,
>
> This is a great report .. thanks for taking the time to create it.
>
> Here is my feedback on it-
>
> - Run spellcheck :-p there were a few typos here and there like
> "enough" spelled "ebnough"
>
> what is a spell checker ?
>
>
> -
> - The Introduction on page 3 takes about PT, that was a metric you
> used in the last report, but in this report you are comparing against
> G1.1.1. The PT is not being used or mentioned throughout the
> remaining of the report.
>
> thanks for catching this. PT is left over from the initial report. I'll
> fix that.
>
>
> -
> - Please mention if you need RAIDs or Ramdisks for the DB
>
> K
>
>
> - I think you forgot the mention the OS version... I assume its SLES
> "10" SP1
>
> Page 9
>
>
> - Quite surprised that we go from 70M unzipped to 111M after initial
> startup (whats the big hitter here.. the logs??)
>
>
>
> -
> - The URL at the end of page 8 (more info about daytrader) seems to
> be incorrect and the link doesnt match the URL
> - the slowdown in the web primitives is a bit surprising... and yet
> the jump in direct mode numbers is pretty cool
>
> The mysteries of life ... it wierded me out as well. I think it needs to
> be looked into but it is an external measurement so I figure putting out the
> info as it is with all the caveats is better than waiting another 6 months
> :)
>
>
> -
> - would have loved to see the Trade Session2JDBC numbers as well
>
> could due that
>
>
> - pg 14, 2nd para, typo "due" --> "do"
>
> doh
>
>
> - can we have an appendix at the end of the report with individual
> run results like last time?
>
> Maybe...I have the XML files from JIBE but there doesn't seem to be a way
> to regerenate a nice summary...do you know how to do this ?
>
>
> -
>
> HTH,
> Piyush Agarwal
>
> On 10/22/07, Matt Hogstrom < matt@hogstrom.org > wrote:
> >
> > I've been noodling on this for a bit and wanted to give y'all a
> > gander at what I have for the performance report at this point. This
> > is based on 2.0.2 and uses DayTrader 2.0. There are a few numbers
> > that are missing. I originally had planned on not producing them but
> > the charts look odd with the missing numbers. It includes a
> > comparison of 1.1.1 and 2.0.2 using DayTrader 1.2 and 2.0. Heh, we
> > need to release those monsters.
> >
> > This should be considered an alpha release but will move quickly to
> > final by the end of the week :)
> >
> > Please provide your feedback on content, what's interesting, not, etc.
> >
> > Thanks for taking a few minutes to look at the draft.
> >
> > Look at http://people.apache.org/~hogstrom/performance/geronimo/2.0/<http://people.apache.org/%7Ehogstrom/performance/geronimo/2.0/>
> > Geronimo2.0.2PerformanceReport-v01draft.pdf
> >
> > Thanks
> >
>
>
>
Re: Draft of 2.0.2 Performance Report
Posted by Matt Hogstrom <ma...@hogstrom.org>.
On Oct 23, 2007, at 10:53 AM, Piyush Agarwal wrote:
> Hi Matt,
>
> This is a great report .. thanks for taking the time to create it.
>
> Here is my feedback on it-
> Run spellcheck :-p there were a few typos here and there like
> "enough" spelled "ebnough"
what is a spell checker ?
> The Introduction on page 3 takes about PT, that was a metric you
> used in the last report, but in this report you are comparing
> against G1.1.1. The PT is not being used or mentioned throughout
> the remaining of the report.
thanks for catching this. PT is left over from the initial report.
I'll fix that.
> Please mention if you need RAIDs or Ramdisks for the DB
K
> I think you forgot the mention the OS version... I assume its SLES
> "10" SP1
Page 9
> Quite surprised that we go from 70M unzipped to 111M after initial
> startup (whats the big hitter here.. the logs??)
> The URL at the end of page 8 (more info about daytrader) seems to
> be incorrect and the link doesnt match the URL
> the slowdown in the web primitives is a bit surprising... and yet
> the jump in direct mode numbers is pretty cool
The mysteries of life ... it wierded me out as well. I think it
needs to be looked into but it is an external measurement so I figure
putting out the info as it is with all the caveats is better than
waiting another 6 months :)
> would have loved to see the Trade Session2JDBC numbers as well
could due that
> pg 14, 2nd para, typo "due" --> "do"
doh
> can we have an appendix at the end of the report with individual
> run results like last time?
Maybe...I have the XML files from JIBE but there doesn't seem to be a
way to regerenate a nice summary...do you know how to do this ?
> HTH,
> Piyush Agarwal
>
> On 10/22/07, Matt Hogstrom < matt@hogstrom.org > wrote:
> I've been noodling on this for a bit and wanted to give y'all a
> gander at what I have for the performance report at this point. This
> is based on 2.0.2 and uses DayTrader 2.0. There are a few numbers
> that are missing. I originally had planned on not producing them but
> the charts look odd with the missing numbers. It includes a
> comparison of 1.1.1 and 2.0.2 using DayTrader 1.2 and 2.0. Heh, we
> need to release those monsters.
>
> This should be considered an alpha release but will move quickly to
> final by the end of the week :)
>
> Please provide your feedback on content, what's interesting, not, etc.
>
> Thanks for taking a few minutes to look at the draft.
>
> Look at http://people.apache.org/~hogstrom/performance/geronimo/2.0/
> Geronimo2.0.2PerformanceReport-v01draft.pdf
>
> Thanks
>
Re: Draft of 2.0.2 Performance Report
Posted by Christopher Blythe <cj...@gmail.com>.
ah... piyush raises a good point regarding the session 2 direct mode. this
is provided by both daytrader 1.2 and 2.0 and is one of the more common
patterns we see out in the j2ee community.
chris
On 10/23/07, Piyush Agarwal <pv...@gmail.com> wrote:
>
> Hi Matt,
>
> This is a great report .. thanks for taking the time to create it.
>
> Here is my feedback on it-
>
> - Run spellcheck :-p there were a few typos here and there like
> "enough" spelled "ebnough"
> - The Introduction on page 3 takes about PT, that was a metric you
> used in the last report, but in this report you are comparing against
> G1.1.1. The PT is not being used or mentioned throughout the
> remaining of the report.
> - Please mention if you need RAIDs or Ramdisks for the DB
> - I think you forgot the mention the OS version... I assume its SLES
> "10" SP1
> - Quite surprised that we go from 70M unzipped to 111M after initial
> startup (whats the big hitter here.. the logs??)
> - The URL at the end of page 8 (more info about daytrader) seems to
> be incorrect and the link doesnt match the URL
> - the slowdown in the web primitives is a bit surprising... and yet
> the jump in direct mode numbers is pretty cool
> - would have loved to see the Trade Session2JDBC numbers as well
> - pg 14, 2nd para, typo "due" --> "do"
> - can we have an appendix at the end of the report with individual
> run results like last time?
>
> HTH,
> Piyush Agarwal
>
> On 10/22/07, Matt Hogstrom < matt@hogstrom.org > wrote:
> >
> > I've been noodling on this for a bit and wanted to give y'all a
> > gander at what I have for the performance report at this point. This
> > is based on 2.0.2 and uses DayTrader 2.0. There are a few numbers
> > that are missing. I originally had planned on not producing them but
> > the charts look odd with the missing numbers. It includes a
> > comparison of 1.1.1 and 2.0.2 using DayTrader 1.2 and 2.0. Heh, we
> > need to release those monsters.
> >
> > This should be considered an alpha release but will move quickly to
> > final by the end of the week :)
> >
> > Please provide your feedback on content, what's interesting, not, etc.
> >
> > Thanks for taking a few minutes to look at the draft.
> >
> > Look at http://people.apache.org/~hogstrom/performance/geronimo/2.0/<http://people.apache.org/%7Ehogstrom/performance/geronimo/2.0/>
> > Geronimo2.0.2PerformanceReport-v01draft.pdf
> >
> > Thanks
> >
>
>
--
"I say never be complete, I say stop being perfect, I say let... lets
evolve, let the chips fall where they may." - Tyler Durden
Re: Draft of 2.0.2 Performance Report
Posted by Piyush Agarwal <pv...@gmail.com>.
Hi Matt,
This is a great report .. thanks for taking the time to create it.
Here is my feedback on it-
- Run spellcheck :-p there were a few typos here and there like
"enough" spelled "ebnough"
- The Introduction on page 3 takes about PT, that was a metric you
used in the last report, but in this report you are comparing against
G1.1.1. The PT is not being used or mentioned throughout the remaining
of the report.
- Please mention if you need RAIDs or Ramdisks for the DB
- I think you forgot the mention the OS version... I assume its SLES
"10" SP1
- Quite surprised that we go from 70M unzipped to 111M after initial
startup (whats the big hitter here.. the logs??)
- The URL at the end of page 8 (more info about daytrader) seems to be
incorrect and the link doesnt match the URL
- the slowdown in the web primitives is a bit surprising... and yet
the jump in direct mode numbers is pretty cool
- would have loved to see the Trade Session2JDBC numbers as well
- pg 14, 2nd para, typo "due" --> "do"
- can we have an appendix at the end of the report with individual run
results like last time?
HTH,
Piyush Agarwal
On 10/22/07, Matt Hogstrom <matt@hogstrom.org > wrote:
>
> I've been noodling on this for a bit and wanted to give y'all a
> gander at what I have for the performance report at this point. This
> is based on 2.0.2 and uses DayTrader 2.0. There are a few numbers
> that are missing. I originally had planned on not producing them but
> the charts look odd with the missing numbers. It includes a
> comparison of 1.1.1 and 2.0.2 using DayTrader 1.2 and 2.0. Heh, we
> need to release those monsters.
>
> This should be considered an alpha release but will move quickly to
> final by the end of the week :)
>
> Please provide your feedback on content, what's interesting, not, etc.
>
> Thanks for taking a few minutes to look at the draft.
>
> Look at http://people.apache.org/~hogstrom/performance/geronimo/2.0/<http://people.apache.org/%7Ehogstrom/performance/geronimo/2.0/>
> Geronimo2.0.2PerformanceReport-v01draft.pdf
>
> Thanks
>