Modern web applications or apis often handle heavy load tasks, requiring intense disk i/o or complex database queries. I will demonstrate how and - more importantly - why such operations should be processed asynchronously. We will be comparing a few messaging/queuing libraries and cover some of the most common pitfalls and obstacles developers face when implementing asynchronous processing and face long running php processes for the first time.


Comments are closed.

Great talk with good slides.
But I wish there was an example to get feedback to the client after processing a message.

Ike Devolder at 16:43 on 23 Sep 2018

Good talk on ways to offload "heavy processing" to workers outside of the main request.

I would really look into the signal handling mentioned because you can still have message loss if your system gets shutdown accidentally and your worker cannot handle this gracefully. Also using supervisord might be outdated on the current line of production servers running linux. There you can just use systemd to manage your processes without the need of other 'process managers'

Marat at 18:12 on 24 Sep 2018

Very interesting information about different queue realization and offloading the real work to daemon workers.
One suggestion to improve would be to see the real case where it might help (e.g. cache reload, huge data export/import)