[LINK] Streaming TV Tests

[Adam Todd]

At 01:52 AM 29/03/2007, stephen at melbpc.org.au wrote:
>At 10:33 PM 28/03/2007, George writes:
>
> > Thanks Rick, you've got a nice pipe there :)
>
>Ok .. Rick may have a fat pipe .. but I've got a long one :-)
>
>By that I mean, just for interest, and, perhaps of some help,
>here, being 100km NW of Bendigo with waving wheat fields,
>and via the free broadband-connect-scheme 256/64 satellite
>/wireless 'broadband' (haha) connection, (either that or dialup
>which is untimed and unlimited but one dollar a call) the audio
>for your vvr-test-3 streams well, but the video is non-existent :-(

Not uncommon for most video streaming type protocols.  I can't understand 
how most people watch them.  Because, you describe this action ...

>It's a little surprising because, on the other hand, YouTube works
>fine.. true, it's slow (trick is to hit pause, build a buffer at around a
>minute of buffer for a minute play time) but it eventually does work.

And I've watched HUNDREDS of people who haven't a clue about that!  They 
just watch it in 2 second bursts.

Maybe that's a hint to advertisers!  (Patent A.Todd 2007  - 2 second burst 
advertising)

I'm not sure what streaming technology George is using.  Last time I 
checked (was it two or three years ago?) it was not a flash protocol.

YouTube converts the films to flash.  The encoding is pretty bad with 
artifacts on fast moving parts.  Unless you actually encode your streams 
with compression markers for areas that change suddenly or whip pans, the 
blocks look really bad.

I'm getting pretty good at using common day compression tools to compress 
films now.  We've got quite a collection of them.  You can watch the three 
minute films at http://iconoclast.inoz.com/ in the light green column.

Even saying all the above, Flash is still an issue where bandwidth is on 
the lower side.  It's not so much the bandwidth to the user either.

I have HEAPS of bandwidth and when I'm streaming Flash from Filmaka or 
YouTube, then inbound streams are less than 10% of the capacity of my 
bandwidth.

If I set up four tabs (firefox) and have four streams filling buffering, 
then it's still 10-13% of my bandwidth.

However when I download an IOS image from ftp.aarnet.edu.au  I can easily 
hit 90-100% of my bandwidth.

So it's a network congestion and peering interconnect issue more than a 
source and destination bandwidth issue.  (Well I presume both Filmaka and 
YouTube have sufficient bandwidth to delivery full speed streams to more 
than 50% of the users at any one time.

>Be that as it may, it may make sense in assisting George if Linkers
>visit say, <http://www.ozspeedtest.com/bandwidth/> and include the
>results in your report, to give George a hard-data comparison basal?
>
>For example ...
>
>1. Test run on 28/03/2007 @ 10:44 PM
>
>Mirror: Telstra Bigpond
>Test Time: 2.76 secs
>Your line speed is 208 kbps (0.21 Mbps).
>Your download speed is 26 KB/s (0.03 MB/s).


Test run on 29/03/2007 @ 11:17 AM

Mirror: Telstra Bigpond
Data: 600 KB
Test Time: 19.61 secs

Your line speed is 249 kbps (0.25 Mbps).
Your download speed is 31 KB/s (0.03 MB/s).


Well that's far from the truth!  So it's not really a helpful diagnostic as 
it hasn't determined my line speed, it's determined the time it took to 
start sending and finish sending the file.

There are so many reasons that could cause this to be inaccurate.


Test run on 29/03/2007 @ 11:19 AM

Mirror: OptusNet
Data: 600 KB
Test Time: 18.41 secs

Your line speed is 265 kbps (0.27 Mbps).
Your download speed is 33 KB/s (0.03 MB/s).




Test run on 29/03/2007 @ 11:20 AM

Mirror: TPG
Data: 600 KB
Test Time: 20.22 secs

Your line speed is 242 kbps (0.24 Mbps).
Your download speed is 30 KB/s (0.03 MB/s).



I might bounce these to my ISP and see what he has to say ;)