Parse closes down: Is Backend as a Service profitable?

Parse starting sending messages last week that it would be winding down it’s services. This came as a shock for many people who have come to rely on their services as a way to reduce the complexity of having to maintain, deploy and scale your own backend solution. For those of you who haven’t seen it, you can take a look at the announcement below.

Parse closes

It is clear that Facebook is not joking around and not keeping services and tools that are not as profitable as they need. This is a shock for companies and developers as it demonstrates that in the end trusting your backend to an external system can have nefarious consequences for you. From the moment you don’t control all the variables, the risk of this type of problems always exists.

Parse are going a give a full year until the sun sets on the platform, they are even open sourcing the Parse server (which is build on node.js and requires a Mongo Database). But mind you, self hosting a Parse server is by no means easy and never substitutes the ease of the original Parse panel, so other alternatives like setting up your own APIs and creating your own infrastructure, or another Backend as a Service should be considered.

A few years ago, I published a post named “Is Backend as a Service the new Gold rush?”. In this post I analysed if systems like Parse that made developing Apps with a server side were the new money making scheme of the tech world companies. However, after the Parse announcement, one can’t but help thinking that perhaps it is not such a profitable endeavour.

If you stop to think about it, big companies usually create their own Backends and host them on systems like Amazon Web Services, and those are the ones that pay the big bills. Small developers that use Parse, tend to have the free version (which has a ridiculously high limit usage before it starts charging) or pay very little money.

So, is backend as a service really profitable? Will it become an open source self hosted alternative? Or will companies use this as a launch pad for their new and reinforced Parse alternatives?

Let me know your thoughts!


Is Backend as a Service the new Gold rush?

cloudkit logoOne of the worst things about learning to code mobile Apps is that sooner or later you realise that you have just done half of the job. With your incredible new skills you will be able to develop amazing Apps, the problem is that these Apps will be standalone apps. This will be OK for some Apps, but in the world we live in interconnectivity has become the new rule, so you are severely limited if you stick to local.

This ultimately translates in the horrifying fact that you need to develop a backend for your App. For some people this can be quite daunting, and a real turn off. That is why many companies employ different programmers for backend and frontend. In 2011, Parse (the company Facebook bought recently) went live with an extremely ambitious and amazing plan, to enable App developers to set up their own backend very easily. This meant they didn’t have to learn PHP with SQL and other programming languages, they just had to use their intuitive API that creates objects and sends them off to a well-constructed backend infrastructure. It even handled push notifications easily.

So, where the trick? The fact if Parse charges you per request, it has a free tier, and then prices skyrocket. They offer rates of approximately 30 requests per second that can go up to 200 (for a whopping 1700$). This means that prorated hourly, your app can do up to 30 requests per second, effectively it can handle:


30*60*60*24=2592000 per day.

Think of a Chat App, a message would take a request to send, and a request to receive. Which means, per day you can send up to 1296000 messages. It isn’t bad, but the steep money curve means you might want to control those numbers very closely.

Microsoft also decided it wanted a BAAS (Backend as a Service) and pushed out Windows Azure.

Not surprisingly, Apple in its last keynote showcased CloudKit, their answer to this dilemma. As you might imagine, it only works on OSX and iOS and comes in a “free with limits” package. However those limits are extremely high. This is a great strategy to tie developers to the iOS universe, taking them away for the pain of designing robust backends and offering free solutions with intuitive “Apple like” systems.


What do you think? Do you see BAAS as the future? Or do you believe that these are just patches to avoid what really should be done, a good robust old fashioned server?