[Koha-devel] Proposal of a new development for Marseille hackfest, feedback needed

Paul Poulain paul.poulain at biblibre.com
Mon Feb 24 16:26:33 CET 2014


Le 11/02/2014 20:09, Chris Cormack a écrit :
> On 12 February 2014 06:30, Paul Poulain <paul.poulain at biblibre.com> wrote:
>> **** 1st idea : collecting -anonymous- data on staff interface usage
>> We sometimes are wondering whether librarians are running an old version
>> of IE, or if they have a large screen, or if the page X is often run or not.
>> There's a tool that could let us know: piwik (BibLibre already has an
>> offer on Piwik for the OPAC)
>> The idea here would be to add piwik for staff interface, and collect all
>> datas on a single piwik instance.
>> We made something (quick) for one of our current supported library, just
>> to see what kind result we could get. We're quite happy with the
>> performances, datas are interesting (which browser is used, which screen
>> size, which pages. [ For example, the library runs an average of 40
>> times "reports/catalogue-stats.pl" every day (yes, that's crazy ;-) ).
>> This page is worth being optimized !!! ]
>>
>> The only caveat we've seen for now is that it can be tricky to have many
>> websites pointing to the same piwik database.
>> We also are not completely peaceful with having 10 000 librarians
>> calling the server. We could face a heavy load...

Hi Matthieu & Thomas,
(From France, Marseille ;-) )

I'm not sure anymore my idea is a good one. I feel we will face 2 problems:
 * I wanted *any* Koha setup to be able to call a given piwik
installation. It mean that piwik whould not check for the caller URL. It
can be mylibrary.org or anotherfrenchlibrary.fr or kiwilibrary.nz
 * The volumes will be *huge*. We don't know exactly how many libraries
are running Koha, but that may be 5000+, some of them being very very
large (the largest french one is 50+ library branches and 200+
librarians). So I feel that 'piwiking' all libraries, all all branches
and all librarians staff use will be more than huge, so a bad idea.

Your opinion, any hint, welcomed !

PS for koha-devel => the other idea about collecting nominative datas is
still a good one, at least I haven't found any technical problem with it
yet.


-- 
Paul Poulain - Associé-gérant
Tel : (33) 4 91 81 35 08
http://www.biblibre.com
Logiciels Libres pour les bibliothèques et les centres de documentation


More information about the Koha-devel mailing list