Hi, here's the relevant section from the Twitter documentation:
To connect to the Streaming API, form a HTTP request and consume the resulting stream for as long as is practical. Our servers will hold the connection open indefinitely, barring server-side error, excessive client-side lag, network hiccups, routine server maintenance or duplicate logins.
Not sure what you mean about it being "immune". The issue is that it doesn't stop sending data until something goes wrong or the client terminates it. Why Twitter decided this was a good way to serve up the data I don't know.