I hear so much about how well an app scales but there is so much conflicting information. On the one hand every web host claims to make scaling a breeze by adding more servers when the load increases, which seems so simple. But then other services come out which claim to do the scaling of your app for you on an already auto-scaled web host (like heroscale on heroku). If there is a need for these services then the web hosts themselves can’t have been 100% scalable. And you still hear in the news about apps that blow up overnight and end up crashing or nearly crashing and ruining all of their traffic because they didn’t scale well enough. Clearly the developers of these big time apps have heard about popular “scalable” web hosts like heroku, so if using them was all they needed to do, then there wouldn’t be a problem. But there still is one.
Are there any summaries, guides, ect. out there which explain why all of this happens, the current solutions, and how scalable you will really be with those solutions? Where should I start looking?
An app must be able to utilize the extra servers / resources, which it gets. And that sounds easier than it is. A common problem is e.g. the database. Let’s say you have one server with the database and at one point in time you see too much load / too many connections on it. Now you want to start a second db-server but your app doesn’t support multiple db-servers, so scaling will fail.
Here is a presentation from wooga (social game developer) about the topic:
http://www.slideshare.net/wooga/20121130scalable-game-servers