Afrihost Capped ADSL Feedback (MTN) - New beginning

Status
Not open for further replies.
Razotter said:
Its an experimental protocol that is currently making approximately 50% of all calls to google servers via Chrome. http://blog.chromium.org/2015/04/a-quic-update-on-googles-experimental.html I'd say its well in the wild and good to go.

And call it whatever you want Afriman, when a stream hits a steady 25KB/s with no fluctuation up or down, I call it throttling.
Click to expand...
Been reading the issues with QUIC with interest and I'm whether something I've noted could be linked.
I've been testing a couple of VDSL routers by pushing through UDP and TCP streams using iPerf.
These worked perfectly well with one of the routers. The other had a peculiar problem. The wheels regularly fell of the UDP stream tests. TCP worked perfectly well. A reboot was needed to rectify.
So I'm wondering - could this be in some way linked to specific routers / firmware?
A quick test would be the following. When the wheels fall off Chrome reboot your router and immediately resume the Chrome session.
Anyway thought I'd share. Was an extremely interesting experiment with the routers. Not sure whether it's linked but worth looking into.
 
steel_dog said:
Been reading the issues with QUIC with interest and I'm whether something I've noted could be linked.
I've been testing a couple of VDSL routers by pushing through UDP and TCP streams using iPerf.
These worked perfectly well with one of the routers. The other had a peculiar problem. The wheels regularly fell of the UDP stream tests. TCP worked perfectly well. A reboot was needed to rectify.
So I'm wondering - could this be in some way linked to specific routers / firmware?
A quick test would be the following. When the wheels fall off Chrome reboot your router and immediately resume the Chrome session.
Anyway thought I'd share. Was an extremely interesting experiment with the routers. Not sure whether it's linked but worth looking into.
Click to expand...

That is an interesting point! Keep me posted if you run some more tests :)
 
AfriGuy said:
That is an interesting point! Keep me posted if you run some more tests :)
Click to expand...
Don't use Chrome so don't have any QUIC experience however a quick test right now of gmail and youtube revealed no issues. In fact - it was super responsive... subjectively tho ;)
If anyone is having a specific issue with a particular URL drop that in a PM to me and I'll check it out.
PS: I'm not online always but will certainly check it out when I am :)

EDIT: Found something really interesting whilst testing on YouTube.
I was able to mess with the playback quality even though the video was already buffered.
It might be the manner in which QUIC is implemented. So here's what I did.
1. Played a QUIC video (a long one).
2. Allowed the buffer to run ahead.
3. Introduced some latency - I just performed an upload test whilst playback was happening.
4. Audio was disrupted during the period of the test. Lots of stuttering.
5. Remove the latency - i.e.: Wait for the upload test to complete.
6. Replay the stuttering (from 4) segment again and note - crystal clear.

So buffering QUIC a video doesn't necessarily guarantee smooth playback. Realtime latency might play a role. Will continue investigating.
 
Last edited:
steel_dog said:
Don't use Chrome so don't have any QUIC experience however a quick test right now of gmail and youtube revealed no issues. In fact - it was super responsive... subjectively tho ;)
If anyone is having a specific issue with a particular URL drop that in a PM to me and I'll check it out.
PS: I'm not online always but will certainly check it out when I am :)
Click to expand...

You're awesome, thanks for offering to get this checked out in detail!
I've seen similar tests personally. Sometimes having QUIC enabled showed better results, while others it was worse - couldn't link it back to any IPC trends though.
 
AfriGuy said:
You're awesome, thanks for offering to get this checked out in detail!
I've seen similar tests personally. Sometimes having QUIC enabled showed better results, while others it was worse - couldn't link it back to any IPC trends though.
Click to expand...
Check the EDIT on my prior post. Seeing some odd things with QUIC enabled.
Would be cool if someone else could reproduce.
 
steel_dog said:
Check the EDIT on my prior post. Seeing some odd things with QUIC enabled.
Would be cool if someone else could reproduce.
Click to expand...

steel_dog said:
/snip

I was able to mess with the playback quality even though the video was already buffered.
It might be the manner in which QUIC is implemented. So here's what I did.
1. Played a QUIC video (a long one).
2. Allowed the buffer to run ahead.
3. Introduced some latency - I just performed an upload test whilst playback was happening.
4. Audio was disrupted during the period of the test. Lots of stuttering.
5. Remove the latency - i.e.: Wait for the upload test to complete.
6. Replay the stuttering (from 4) segment again and note - crystal clear.

So buffering QUIC a video doesn't necessarily guarantee smooth playback. Realtime latency might play a role. Will continue investigating.
Click to expand...

Also very interesting! I'll see if I can get this replicated and marked against any trends or network status from our end, thanks :)
 
AfriGuy said:
Also very interesting! I'll see if I can get this replicated and marked against any trends or network status from our end, thanks :)
Click to expand...
Reproduced with a completely loaded video.
Turn on Stats for Nerds and notice that, if latency is high enough, frame drops will start to increase - which is VERY unexpected for a fully loaded video.
 
steel_dog said:
Reproduced with a completely loaded video.
Turn on Stats for Nerds and notice that, if latency is high enough, frame drops will start to increase - which is VERY unexpected for a fully loaded video.
Click to expand...

That probably means that even if a video is fully buffered there is some kind of uplink taking place with the Youtube servers to make sure the video cannot be downloaded from the cache of the browser... Very interesting indeed!
 
AfriGuy said:
I think we can probably look into this a little more from our end, but the time we've spent here should have yielded something to tweak from our end.
I'm afraid for now, disabling QUIC within Chrome fixes the issue.
Click to expand...

Well just out of interest sake I enabled the Experimental QUIC protocol and hey presto! I can stream 720p Youtube vids with no buffers whatsoever on my 2mb line tonight!

Just a couple of minutes before enabling all the streams buffered and would automatically set itself to the lowest quality... So it works... knock on wood...
 
steel_dog said:
Reproduced with a completely loaded video.
Turn on Stats for Nerds and notice that, if latency is high enough, frame drops will start to increase - which is VERY unexpected for a fully loaded video.
Click to expand...

What browser are you using? Stuttering on Youtube videos occur on FireFox, most other browsers you can go offline and the video still plays.
 
whatwhat said:
What browser are you using? Stuttering on Youtube videos occur on FireFox, most other browsers you can go offline and the video still plays.
Click to expand...
Latest version of Chrome - 43.0.2357.81 m
Checked and it works fine if it's offline.
 
Last edited:
Nikkel said:
Well just out of interest sake I enabled the Experimental QUIC protocol and hey presto! I can stream 720p Youtube vids with no buffers whatsoever on my 2mb line tonight!

Just a couple of minutes before enabling all the streams buffered and would automatically set itself to the lowest quality... So it works... knock on wood...
Click to expand...

QUIC is actually a very cool concept, in that it basically strips out all unneccesary traffic that is usually involved with routine TCP traffic, and just focuses on getting the stream to you as quickly as possible. So it is great to see that it is working!
 
Hi Afripeeps

I am looking to swap around my package but my decision will be based on the whole double data deal. What exactly is going to happen after 30 June? Will double data be extended or will some other policy come into effect?
 
Last edited:
Status
Not open for further replies.
Back
Top
Sign up to the MyBroadband newsletter
X