We have just left Parse.com due to restrictions on the number of API requests and limitations on joining tables. Having now read extensively about denormalising data we have learnt about dealing with large datasets and not using joins. We are trying to decide whether to move to a noSQL DB or use mySQL which is what we all know (and love).
So the question is: If building an app with an anticipated large growth rate should we start out with noSQL and go through a steep learning curve now or build it faster on mySQL and load test to prove it will cope?
Advise from anyone who has been through this experience gratefully received.
My two cents: ship as fast as you can to validate that people want what you’re building. Learn as quickly as possible.
Therefore, I’d advise building it faster on what you know…in your case MySQL. Load test on MySQL to prove it will scale.
You can always switch to NoSQL later and DreamFactory actually makes that really easy. You can just set up a NoSQL database later and swap out the REST API call names in your client app. The REST API is structurally the same…basically the NoSQL API is a subset of the SQL API.
Thanks Ben, that’s what we figured, just wanted to hear it from someone else. Unless anyone can provide solid justification for starting out on noSQL it seems like mySQL will scale comfortably to our first million users.
