I don't quite understand what you mean by "server". If you mean "server" is the server app, then it is unique to the browser side. But if you mean "server" is the backend, there is only one server app for a user to interact with the backend.
So let's say for example Facebook decided to adopt this paradigm tomorrow. All the calculations that were previously done by the client are now going to be done by one of their servers. Should they expect to have to make more server instances assuming they're in the cloud or have more physical servers to handle the increased load.
Ok, I see what you're asking. If tomorrow FB decided to become Shared App Web, then this would mean that they would no longer have to worry about how many redundant requests each user would make (just assuming there is no optimization whatsoever).
For example, if a user visits a page A, that page creates a WebSocket connection, and a new window opens with page A still creating a new WebSocket connection. Obviously, as the number of window pages increases, it will use more server resources. But this paradigm will only have one WebSocket connection, no matter how many browser windows there are.
That's the problem.
How much less server size can this paradigm use? It depends on how many windows all users are used to, and whether there are any optimizations made to support multiple browser windows. If an application doesn't have any multiple browser window optimization, then such a model can at least significantly reduce server usage (I would guess maybe 10%-50%)
I don't quite understand what you mean by "server". If you mean "server" is the server app, then it is unique to the browser side. But if you mean "server" is the backend, there is only one server app for a user to interact with the backend.
So let's say for example Facebook decided to adopt this paradigm tomorrow. All the calculations that were previously done by the client are now going to be done by one of their servers. Should they expect to have to make more server instances assuming they're in the cloud or have more physical servers to handle the increased load.
I want to get an idea of what this paradigm looks like at scale
Ok, I see what you're asking. If tomorrow FB decided to become Shared App Web, then this would mean that they would no longer have to worry about how many redundant requests each user would make (just assuming there is no optimization whatsoever).
For example, if a user visits a page A, that page creates a WebSocket connection, and a new window opens with page A still creating a new WebSocket connection. Obviously, as the number of window pages increases, it will use more server resources. But this paradigm will only have one WebSocket connection, no matter how many browser windows there are.
That's the problem.
How much less server size can this paradigm use? It depends on how many windows all users are used to, and whether there are any optimizations made to support multiple browser windows. If an application doesn't have any multiple browser window optimization, then such a model can at least significantly reduce server usage (I would guess maybe 10%-50%)
Thank you, that's really neat.