OnePlaylist Stack Overview – ruby elasticsearch mongodb redis HA-Proxy varnish memcache….

I’m working now full time on my new project OnePlaylist since 4 months. This is the best opportunity i had to setup a complete architecture with my own choices.

So what is the stack for now ?

Ruby on Rails

The backend engine: super fast dev. Lot of ruby packages from the community.

Capistrano

Deploy the backend, restart the workers, gen the google site map. Great tool.

Ruby event-machine

Why use node ? i known ruby. This broker talk to websocket and redis. All asynchronous off course.

websocket

Real time in the browser, and so amazing possible features .

redis

Cache, key value storage, pub sub for websocket…

MongoDB

With mongoig wrapper, really cool. All dataset in ram for a while i think.

elasticsearch

yeah, cool stuff. Search engine but also as key value json storage for me. A key element in the architecture.

HA-proxy

great tcp level load balancer. Amazed with the websocket on same domain and subpath capacity.

varnish cache

For an eventually consistent architecture, whole page in cache. really helpfull for my public part of the site. already 120 req /s with  no configuration tuning at this moment.

memcache

integration with ruby on rails is great !

resque

My application is “heavy background processing” with lot of api call. like a lot the real time push to websocket when a job is done..

coffeescript

Lot of work on the browser side, and if i had the choice, i’ll always code in coffee and never in js.

HTML5

websocket, local storage….

d3.js

For data viz on pucblic playlists. More incoming on collection part.

So here is the overview of the techno i used. i’ve decided to share more what i do each day. so stay tuned !

0 replies

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

Your email address will not be published. Required fields are marked *