Massively & Linearly Scalable

Glavin001 commented 10 years ago

Worth reading:

http://stackoverflow.com/a/4710617/2578205 http://stackoverflow.com/a/4737648/2578205 http://stackoverflow.com/a/16289251/2578205 http://stackoverflow.com/a/17031967/2578205

Interesting module:

https://github.com/dotcloud/hipache https://github.com/makesites/drone (right idea, but maybe too small / not polished enough)

Reason:

If the teams decides to go commercial, they could host this in an Amazon EC2 Cloud and scale to handle literally every university / organization they wanted. I'm talking thousands of concurrent users making more than static web content requests. Let's make this system bad-ass!

Glavin001 commented 10 years ago

Scaling Node.js Applications: http://cjihrig.com/blog/scaling-node-js-applications/

Glavin001 commented 10 years ago

Solution: separate the services, such as Redis and Database (MongoDB).

Run Redis on 1-n server(s) (http://redis.io/topics/partitioning),
run MongoDB on another 1-n server(s) (http://docs.mongodb.org/manual/core/sharding/), and
then run your servers in multi-core more on separate 1-n servers that are configured to connect to the external Redis and MongoDB server(s).

In fact, this should already work with the current source code! :+1:

[ ] Test this.

Glavin001 commented 10 years ago

Socket.io stress test: http://weblog.bocoup.com/node-stress-test-analysis/

Glavin001 / SMMApp2