Trouble finding a cloud client/server technology stack that fits

manpreet Best Answer 2 years ago

I've been trying to learn and play with some cloud computing client/server interactions. I was hoping to get suggestions on technology stacks that fit this criteria:

1. Clients are web browsers that communicate asynchronously using websockets, long-polling, and server-sent events.
2. Server is scalable and should support multiple instances should the need arise (for my case of playing/learning I obviously won't run into a scalability problem, but I want to plan for it correctly and will be testing using multiple nodes)
3. Server state should to be persisted using some database technology with ORM
4. Server state should be able to be cached
5. Server should be able to support a scenario where Alice connected to Instance A does something that Bob connected to Instance B needs to be notified of
6. No development cost (assuming I'm using my own hardware)

I thought I had this figured out. I was thinking:

JavaEE/Jetty + Maven for server stuff and deployment

Hibernate for ORM and database agnosticism

Redis for 2nd level cache and multiple instance support

My problem was the client/server communication using the various browser protocols which are different enough and non-trivial to support seamlessly. I looked around and found the Atmosphere framework:

Atmosphere for client/server communication using websockets, long-polling, and server-sent events. Also it apparently supports Redis, but...

This all looked good until after some research I realized that Atmosphere doesn't support Node-based broadcasting (criteria #5). Atmosphere's Redis support is shallow (localhost only out-of-the-box) and I would need to maintain some sort of Node/broadcaster map using Redis to satisfy criteria #5. AtmospherePro apparently does all of this well, but that has a cost which I can't justify (criteria #6).

Can anyone suggest an appropriate technology stack for my criteria? Perhaps an alternative to Atmosphere that is cloud-ready?