As someone just getting into Node.js development, I had been looking into a good (free, for now) PaaS for a hobby project. I had settled on Heroku, and I like it so far for its simplicity. However, reading this was the first time I had considered OpenShift for Node.js. Can anyone comment on the merits of OpenShift vs. Heroku? I've found surprisingly little information comparing the two while searching around.
From what I was seeing, OpenShift looked like it may offer more capacity with its free plan, but I may be missing something here...
Comparing the free Web Dyno on Heroku vs free Cartridge on OpenShift, the latter starts up faster for me. On the other hand, the client tools for OpenShift are of slightly lesser quality.
And one more thing, with OpenShift you get Node.js 0.6 while with Heroku you can specify the engine version.
Not sure if that is a factor, but being in the UK I got way too many 404s in the free tier. In comparison (the same app), Heroku or OpenShift work like a charm.
Our current approach is to create our own machines with dedicated & VMs we rent cheap and setup ourselves. Always interested in recipes to do that.
Nevertheless we are currently evaluating managed solutions like joyent, nodejitsu or Redhat's Openshift to take away the extra workload in managing the iron.
I may actually just start with that. From what I can tell, you get a much much larger database than Heroku, but with a lower bandwidth cap (although 50GB vs 2TB for my purposes is indistinguishable). AppFog also claims it has the "Fastest servers available on any infrastructure" (for what it's worth).
Unless I'm missing something, it sounds like a much better deal. Then again, I'm talking about two free services, but it would be nice to be able to avoid upgrading to paid for as long as possible.
If that is your criteria for using a tool, you're never going to move into anything new. For web development many are happy with Express. Sure they tend to use Mongo or some no-sql type storage, but that is due to novelty and the fact that the individuals writing the async drivers are into that right now. MySQL has support and it's increasing.
Node's an interesting idea and should be given some room to grow. When done correctly it seems to scale well, especially on one box.
To go back to the topic, look at Java. There are so many tools out there to do the same thing in the MVC space. Would you say that you won't use Java because people can figure out which tools to use? To me the benefit of Node and Java is that you are not tied to a particular engineering group's view of the world. You can plug in what you want. You can experiment and learn yourself. A monolithic stack might be good from the it just work's perspective (though I've never seen that to be true and I've programmed in C#), but it does tie you to a release schedule of the vendor.
I hope this should help some one to get started on Node quickly - https://github.com/joyent/node/wiki/Resources
And try to stay in control with NodeJS mailing list :)