Bufferbloat Disrupting IP Video Surveillance Streams?

#1, thanks for sharing.

Here is an excerpt for others:

I do not know enough here but let's see what people think.

I'd say bufferbloat affects video surveillence latency quite a bit.

Though it's different than VOIP or surfing the web, etc, in that those are interactive activities in which the latency becomes noticeable because it occurs between every interaction, e.g. clicking a link and waiting for the page, or saying "hello" and waiting for the response.

When watching a stream though, you may not mind when a second or two or three go by before your stream starts.  And do you really care (normally) that you are 3 seconds in the past when viewing from a remote location?

Two big exceptions would be PTZ control and intensive video scrubbing...

I've never heard this term before. But increasing buffers only helps if you haven't oversubscribed the line and you only have occasional spikes of traffic. Still, some caching is beneficial, and if the avg cache size has increased over time (likely) then the cumulation of caches between the source and destination will add up (if they are actually in use). But it should still be a fairly small amount of caching because TCP retransmissions would soon be sent if expected traffic was not received - negating the benefits of caching.

You should not expect (up to) 3 seconds of video sitting in switch/router buffers throughout the internet on a regular basis. But occasionally it's expected that you might see some buffering on certain segments of the route.

I feel like the issue is overstated and the majority of latency problems are due to actual overall round trip latency. I can certainly be wrong, I don't have experience with ISP/backbone level networking.