Il 23/02/15 11:16, Zahir Lalani ha scritto:
Hi
We are building an app which will be required to scale easily as it is expected to grow in use. From my previous job we had a good model (java based) for scaling where we could add n nodes easily as the number of users grew. I am now thinking about how I can achieve the same here, and want to get input from you guys.
I am currently looking at either memcached/redis as the session stores. We previously used couchbase and that worked very well. Has anyone had any experience of these session stores with Dancer?
The next item is the actual web server mechanism. Starman seems to be mentioned a lot, but I will be honest and say I do not know much about the Plack ecosystem. I am very used to using Apache (driving a Java container) and would like to continue with Apache as it allows us to do some wonderful tricks. Raw cgi seems a waste as its performance will not be good. I have seen articles that use starman behind apache but was wondering if any of you guys had direct experience of this, or what your recommendations would be.
Ultimately we will need to be able to deploy n+1 nodes of the app layer behind a load balancer.
Your experience in this would be appreciated
Hello, I typically use uWSGI, a very fast application server that supports PSGI [1] and integrates nicely with Apache via mod_proxy_uwsgi [2]. It can even work as a load-balancer, sitting between your apache and the various uWSGI instances [3] [1] http://uwsgi-docs.readthedocs.org/en/latest/PSGIquickstart.html [2] http://uwsgi-docs.readthedocs.org/en/latest/Apache.html [3] http://uwsgi-docs.readthedocs.org/en/latest/Fastrouter.html -- Mirko Iannella