HTTP pull ttl is implemented from client side so that the server does not create lots of open connections which can lead to out of memory. But this is the server limitation that it can create a fixed amount of open connections only.
So if a client decides not to implement ttl and start sending a lot of requests then the server will result in out of memory.
So is there no control from server side itself ?
Course: Web Application and Software Architecture 101 - Learn Interactively
Lesson: HTTP Push - Web Application and Software Architecture 101