What is the best solution to improve OroCRM performance?

Hello,

I’d like to know how to improve the performance on server side to make orocrm works faster. Today I use most of the time the latest technologies (PHP 5.5, MariaDB 5.5.42, Apache 2.4), the server works quite well with other web applications (even Magento :-) – http://demomagento.diglin.com) but OroCRM interface could be faster.

Do you have any tricks which can help to improve the responsiveness of the interface ?

Thank you for your reply

Regards
Sylvain

Topic

[RodolfoParticipant]

Did you try Varnish?

One thing that really speeded up my server was disabling realtime tracking processing.. Having less CPU consumption from database gave a much better performance from the interface. (But this only helps if you’re using oro-tracking.)

What’s the specification of your server? Let me know if you find something else helpful.
Cheers.

[AlexanderModerator]

Hi, @diglin.

Actually we do not have any tricks.
But I can suggest to check that OpCache is enabled, also check xdebug, disabling it can improve performance a bit.
And additionally you can configure Redis to cache annotations, metadata, oro_data caches, etc.

Also you can significantly improve performance by configuring web farm nodes.
The documentation on how to do that is coming.

[diglinParticipant]

Hi

Thanks for your reply. Varnish won’t help as everything is dynamic and need actual information from the server.

Regarding OpCache, that’s already enabled (I use PHP 5.5), xdebug was also deactivated. Redis could be a candidate. NewRelic shows me the followings, it’s no so detailed because I use a limited version of new relic but you still see which controller needs some performance improvements:

[Ivan KlymenkoSpectator]

Thanks, diglin.
We will check what we could do with them. Actually, performance is the hottest topic for us, so we have plans to improve it in next releases. We started with optimizing the base requests, than will go thru features and check what could be improved, so any feedback will be appreciated.

[richardh87Participant]

@Diglin, did you try with Redis? How did this improve?

Right now it is painfully slow for me, and I have to look for an alternative.

[Author]

Replies