openerp-community team mailing list archive

[Alberto Barrionuevo <abarrio@xxxxxxxxxx>]

Hi Fréderic, Nhomar,

On 08/10/14 09:05, Frédéric Clementi wrote:
> Even with these tunings Odoo has no chance to handle such a volume of
> data. We are already struggling at 3000 SO/day.ç

   Doing changes in the OpenERP v6.0 code to optimize and remove 
serious performance problems that it has, and using a middleware in 
front of OpenERP (v60), we have been able to manage +55.000 orders each 
12 hours with this server architecture [attached systems hardware diagram].

   Since our architecture is not "as big", we are sure that we can 
multiply that number using more application servers, since, once 
optimized. With improvements we get the initial software limit moved 
from OpenERP to PostgreSQL, that is much more powerful in terms of 
scalability than the standard OpenERP.

   The attached video shows the parallel creation of 32 SO in 25 
seconds using the hardware architecture finally used in productive. It 
means 0,78 seconds per SO.

   Indeed, the sales orders being created in this demo are heavier than 
the standard ones, since they include some custom business processes to 
be executed in OpenERP before the SO is confirmed in the sales flow. 
Orders originally come mainly from a web trading platform from the final 
customer, but also from other sources, all them passing through our 
middleware for OpenERP done with Django. In the video we are simulating 
the source with our own launching script used in QA tests.

   Disclaimer: This kind of solutions are on demand for each actual 
case. They cannot be generalized. Some of the used optimization 
techniques may be more generally applied. Additionally, Odoo v7.0 and 
8.0 would need less work to escalate than the v6.0 that we "suffered".

   If anybody is really interested, with some days work in advance to 
install again the whole solution, we can prepare and show a working 
demo. But it would use our demo servers that are less powerful than the 
attached productive architecture.

Best regards,
--
Alberto Barrionuevo, Director
OPENTIA Group
Odoo/OpenERP Silver Partner
T (+34) 918 38 38 58
F (+34) 944 34 00 77

        OPENTIA
Open Tech for Open Minds
     www.opentia.com
> *camptocamp*
>
> INNOVATIVE SOLUTIONS
> BY OPEN SOURCE EXPERTS
>
> *Frédéric Clementi*
> Project Manager
> Business Solutions
>
> +41 21 619 10 41
> *
> *
> www.camptocamp.com <http://www.camptocamp.com/>
>
>
> 2014-09-29 13:54 GMT+02:00 Nhomar Hernández <nhomar@xxxxxxxxx
> <mailto:nhomar@xxxxxxxxx>>:
>
>
>     2014-09-29 4:26 GMT-04:30 lin.yu <lin.yu@xxxxxxxxxxxxxx
>     <mailto:lin.yu@xxxxxxxxxxxxxx>>:
>
>
>         You can find more test results and details here
>
>         http://www.jianshu.com/p/eda222886d7a
>
>         Your comments are welcome.
>
>
>     Hello.
>
>     I couldn't find some basic things mentioned in your post:
>
>     1.- are your running with what # of Workers?
>     2.- # of Cores?
>     3.- Load balancing?
>     4.- # Number of instances runing?
>     5.- Is it postgres optimized?
>
>     I have one customer like this, and my tests given different results
>     (not perfect but better than yours).
>
>     did you follow this advices?
>
>     http://www.slideshare.net/openobject/performance2014-35689113
>
>
>
>     --
>     --------------------
>     Saludos Cordiales
>     --
>     Nhomar Hernandez
>     http://about.me/nhomar
>
>     <http://about.me/nhomar?promo=email_sig>
>
>
>     _______________________________________________
>     Mailing list: https://launchpad.net/~openerp-community
>     Post to     : openerp-community@xxxxxxxxxxxxxxxxxxx
>     <mailto:openerp-community@xxxxxxxxxxxxxxxxxxx>
>     Unsubscribe : https://launchpad.net/~openerp-community
>     More help   : https://help.launchpad.net/ListHelp
>
>
>
>
> _______________________________________________
> Mailing list: https://launchpad.net/~openerp-community
> Post to     : openerp-community@xxxxxxxxxxxxxxxxxxx
> Unsubscribe : https://launchpad.net/~openerp-community
> More help   : https://help.launchpad.net/ListHelp