This question is more of a theoretical one, than a programming one.
I am currently in the process of writing a single page javascript application with Knockout.js. I choose Knockout because I love the way it handles DOM updates, but it does require you to write your own REST backend for data storage and loading.
My application is an enterprise application, where the server and client will be installed on it’s own server instance, and not shared with any other users. It does not require authentication, and only will only need to save and load JavaScript objects.
My original thought was to use a node.js backend with Mongo for storage, but this seems a little overkill for me. Would it be considered a bad practice to save the raw json as a flat file, and save/load it as needed? The only problem I see is if multiple users were using the application and try to save data at the same time. Is there an easy way to handle a merge of JSON data?
If there are concurrent access to your application that can modify this data I would definitely advise to keep it on the server side. It depends on your application, but merging the JSON will most likely become a nightmare. I believe it will be better to manage concurrent access from the backend. Mongo DB is a good option, but take a look at CouchDB (which provides REST interface) and Redis as well.
On the other hand, if concurrency was not an issue, you may want to check HTML5 local storage, which basically lets you store key/value pairs on the client side. It’s a great alternative to cookies since it is not sent on each request and allows you to store more data in a cleaner way.