Here's a little thing i wrote... kohaprofile
Dear Community, I spent the last day of the hackfest in Marseille hacking together a "proof of concept" script for checking changes in the performance of Koha over time: https://github.com/digibib/kohaprofile The idea is to step x commits back in time, run a bunch of timed tests/benchmarks on Koha, take one step forwards in time, run the tests again etc. Hopefully this might be able to show us what the general trends are, and if there are any changes that lead to noticeable increases in response times. If you run the same benchmark a few times in a row, the response times vary quite a bit. Perhaps there is too much "noise" for this to be really useful. (To mitigate this, the script can run the tests y number of times and record the average time. Perhaps there are better ways to deal with the statistics of this, like removing the fastest and the slowest time etc.) This is a "proof of concept", so I would be really interested in hearing your thoughts on it. If the concensus is that it does not yield interesting data, I'll abandon the project. If people think it has some merit I would be more than happy to accept pull requests for improving it. :-) See the README for some ideas for improvements. Does anyone know if there are tools that can do what I'm trying to accomplish here (specifically the idea of stepping through Git history to check performance) already? Best regards, Magnus
This sounds like a good idea - nice work! I don’t feel that I am capable of really assessing all pros and cons for developers, but I’d still like to say this: speed is essential and everything that helps to ensure that feature creep doesn’t slow Koha down over time is welcome. I’m going out on a limb here, but would it also be relevant to (at some distant point in the future) hook this up with the ideas about Koha being able to ”call home” if you allow it in order to keep track of how many installations there are worldwide? I’d like to see larger data sets in order to follow the general trend and it might help in identifying hardware or OS-distros having atypical problems with a specific patch. Regards/Viktor Viktor Sarge Utvecklingsledare Regionbibliotek Halland Kultur i Halland TFN: 035-17 98 73 E-POST: Viktor.Sarge@regionhalland.se<mailto:Viktor.Sarge@regionhalland.se> BESÖKSADRESS: Södra vägen 9, 30180 Halmstad WEBB: www.regionhalland.se/regionbibliotek<http://www.regionhalland.se/regionbibliotek> 18 mar 2014 kl. 11:34 skrev Magnus Enger <magnus@enger.priv.no<mailto:magnus@enger.priv.no>>: Dear Community, I spent the last day of the hackfest in Marseille hacking together a "proof of concept" script for checking changes in the performance of Koha over time: https://github.com/digibib/kohaprofile The idea is to step x commits back in time, run a bunch of timed tests/benchmarks on Koha, take one step forwards in time, run the tests again etc. Hopefully this might be able to show us what the general trends are, and if there are any changes that lead to noticeable increases in response times. If you run the same benchmark a few times in a row, the response times vary quite a bit. Perhaps there is too much "noise" for this to be really useful. (To mitigate this, the script can run the tests y number of times and record the average time. Perhaps there are better ways to deal with the statistics of this, like removing the fastest and the slowest time etc.) This is a "proof of concept", so I would be really interested in hearing your thoughts on it. If the concensus is that it does not yield interesting data, I'll abandon the project. If people think it has some merit I would be more than happy to accept pull requests for improving it. :-) See the README for some ideas for improvements. Does anyone know if there are tools that can do what I'm trying to accomplish here (specifically the idea of stepping through Git history to check performance) already? Best regards, Magnus _______________________________________________ Koha-devel mailing list Koha-devel@lists.koha-community.org http://lists.koha-community.org/cgi-bin/mailman/listinfo/koha-devel website : http://www.koha-community.org/ git : http://git.koha-community.org/ bugs : http://bugs.koha-community.org/
Le 18/03/2014 11:34, Magnus Enger a écrit :
Dear Community,
I spent the last day of the hackfest in Marseille hacking together a "proof of concept" script for checking changes in the performance of Koha over time:
https://github.com/digibib/kohaprofile
The idea is to step x commits back in time, run a bunch of timed tests/benchmarks on Koha, take one step forwards in time, run the tests again etc. Hopefully this might be able to show us what the general trends are, and if there are any changes that lead to noticeable increases in response times. This idea is wonderful. When I remember Koha 1.x, I remember a *very fast* Koha. I would be curious to know if the speed lost is only because we added many many things, or if there is a pitfall we fall in (in this case, I would bet it's related to Unicode management...)
-- Paul POULAIN - BibLibre http://www.biblibre.com Free & Open Source Softwares for libraries Koha, Drupal, Piwik, Jasper
On Tue, Mar 18, 2014 at 11:07 AM, Paul Poulain <paul.poulain@biblibre.com> wrote:
Le 18/03/2014 11:34, Magnus Enger a écrit :
The idea is to step x commits back in time, run a bunch of timed tests/benchmarks on Koha, take one step forwards in time, run the tests again etc. Hopefully this might be able to show us what the general trends are, and if there are any changes that lead to noticeable increases in response times. This idea is wonderful. When I remember Koha 1.x, I remember a *very fast* Koha. I would be curious to know if the speed lost is only because we added many many things, or if there is a pitfall we fall in (in this case, I would bet it's related to Unicode management...)
It would be really cool to set this up somewhere and run it back to the first commit in master and let it rip. Kind Regards, Chris
I'm not a developer, but I lurk on this list anyway. I manage a very large Koha consortia in Kansas that has faced significant speed issues over the last year with our system, as some of you may be aware. I welcome any and all avenues of trying to pinpoint potential speed bottlenecks in the code! Targeting checkout and checkin processes and code, especially, for this implementation would be fantastic. Thanks for pursuing this. Heather Braum NExpress Coordinator Resource Sharing Librarian Northeast Kansas Library System hbraum@nekls.org "The illiterate of the 21st century will not be those cannot read and write, but those who cannot learn, unlearn, and relearn." ~Alvin Toffler, *Rethinking the Future* On Tue, Mar 18, 2014 at 5:34 AM, Magnus Enger <magnus@enger.priv.no> wrote:
Dear Community,
I spent the last day of the hackfest in Marseille hacking together a "proof of concept" script for checking changes in the performance of Koha over time:
https://github.com/digibib/kohaprofile
The idea is to step x commits back in time, run a bunch of timed tests/benchmarks on Koha, take one step forwards in time, run the tests again etc. Hopefully this might be able to show us what the general trends are, and if there are any changes that lead to noticeable increases in response times.
If you run the same benchmark a few times in a row, the response times vary quite a bit. Perhaps there is too much "noise" for this to be really useful. (To mitigate this, the script can run the tests y number of times and record the average time. Perhaps there are better ways to deal with the statistics of this, like removing the fastest and the slowest time etc.)
This is a "proof of concept", so I would be really interested in hearing your thoughts on it. If the concensus is that it does not yield interesting data, I'll abandon the project. If people think it has some merit I would be more than happy to accept pull requests for improving it. :-)
See the README for some ideas for improvements.
Does anyone know if there are tools that can do what I'm trying to accomplish here (specifically the idea of stepping through Git history to check performance) already?
Best regards, Magnus _______________________________________________ Koha-devel mailing list Koha-devel@lists.koha-community.org http://lists.koha-community.org/cgi-bin/mailman/listinfo/koha-devel website : http://www.koha-community.org/ git : http://git.koha-community.org/ bugs : http://bugs.koha-community.org/
participants (5)
-
Christopher Nighswonger -
Heather Braum (NEKLS) -
Magnus Enger -
Paul Poulain -
Viktor.Sarge＠regionhalland.se