First, I must confess…this blog has much less to do with video than the title insinuates. Rather, it’s really more of a math problem that is somewhat related to video streaming.

Within the last couple weeks, I came across an article that talked about Google’s new “experimental fiber network,”and mentioned some of the efforts that the city of Kalamazoo is doing to become a prototype community for the venture. It’s a pretty short and to-the-point article. But one thing jumped out at me, and it was this:

Michael Cross, city information technology operations manager, said that technology could be a major bonus in attracting and retaining business. Cross said a typical cable modem Internet connection would usually take just over two hours to stream a full-length, high-definition feature film. With Google Fiber, streaming the same file would take less than two minutes.

Now, admittedly, this is a pretty harmless statement. But, due to some construct in my mind that sometimes refuses to accept facts in the absence of a number trail to follow, I decided to spend a few minutes and verify those numbers (”2 hours” and “2 minutes”) for myself. Although I often do this with stories I read these days, I was also just plain curios as to how long it really does take to download a full-length high-def. feature film, on average, in the Kalamazoo and Portage area.

WARNING: LOTS OF (admittedly basic) MATH TO FOLLOW!

Average Feature Film Length:

110 minutes or 6600 seconds (This assumes that the "average" person is downloading a "popular" film, which is skewed towards longer movies)

http://www.slashfilm.com/2009/07/16/by-the-numbers-the-length-of-feature-films/
http://wiki.answers.com/Q/What_is_the_average_length_of_a_movie

Average YouTube HD Bitrates:

Video = 2Mbps
Audio = 178Kbps = 0.178Mbps
Video+Audio = 2.178Mbps

http://webvideotechniques.com/123/bigger-and-better-encoding-for-youtube-hd
http://techvideoblog.com/reviews/online-video-sites-hd-quality-comparison/

Average NetFlix HD Bitrates:

2.600 - 3.800 Mbps (or let’s assume average of 3.2Mbps).
(This assumes that this number includes audio, which it may not.)

http://blog.netflix.com/2008/11/encoding-for-streaming.html
http://gizmodo.com/5093323/the-tech-specs-of-hd-netflix-streaming

Average CHARTER cable modem download speed in Kalamazoo:

8.53Mbps

http://speedtest.net/global.php#0,1,1,58
http://www.charter.com/Visitors/Products.aspx?MenuItem=20

Google’s Proposed Fiber Network Speed:

1Gbps = 1024 Mbps

http://www.google.com/appserve/fiberrfi/

Bit/Byte Conversions:
http://www.unitconversion.org/data-storage/megabits-to-megabytes-conversion.html
http://www.unitconversion.org/data-storage/megabits-to-gigabits-conversion.html

Formula for calculating size of h.264 encoded video:
http://stackoverflow.com/questions/701991/h-264-file-size-for-1-hr-of-hd-video

Total File Size of HD Feature Length Video:
YouTube:

6600 seconds * 2.178mbits = 6600 seconds * 0.272MB = 1795.2MB

NetFlix:

6600 seconds * 3.2mbits = 6600 seconds * 0.4MB = 2640MB

Download Speeds using Cable Modem:
Typical cable modem in Kalamazoo: 8.53Mbps = 1.07MBps
YouTube HD

1795.2MB / 1.07MBps = 1677.76 seconds = 27.96 minutes = 0.47 hours

NetFlix HD

2640.0MB / 1.07MBps = 2467.29 seconds = 41.12 minutes = 0.69 hours

Download Speeds using Google Fiber:
1024Mbps = 128MBps
YouTube HD

1795.2MB / 128MBps = 14.02 seconds = 0.23 minutes

NetFlix HD

2640.0MB / 128MBps = 20.62 seconds = 0.34 minutes

END OF MATH

Certainly, the addition of 1080p HD from YouTube will change downloads for the larger, as will set-top streaming boxes…but right now I tend to think that 1080p is not the average.

I’m not picking on Mr. Cross. These are more than likely not even his numbers, and came from yet another source. And after all, does it really matter
if the difference in download time is 30 minutes vs. 15 seconds or 2 hours vs. 2 minutes? No, not to the average user. The point is that Google’s fiber is really really fast. In fact, I would not have been compelled to write this blog post had the article stuck with such characterizations as “ridiculously fast” and “warp-speed.”

Preferably, and in theory, the ubiquity of the web should make citations a cinch. Just cite the source (and make others cite their source for you) with a link and that’s it; you’re done. That source, in turn, should have any references marked in it, and eventually the user will get to the root source of information…even if 9 out of 10 “trails” end at Wikipedia, at which time the user can decide the extent to which they find that source credible.

In other words, I in no way claim that my work above is the holy grail of video streaming calculations! There are many assumptions which may have lead my conclusions astray. My only claim is that you can at least see a thought process, and verify for yourself (at least to the extent that I have done so), the sources and thought processes employed. In my opinion, part of the fun of putting it out there is to see how you may have gone wrong.

This is a bit of a hobby for me. I pick a story that has volunteered information and presented it as fact. Then, I spend a half-hour or so trying to verify
that information. I’d like to find a different hobby though…like winter climbing, wingsuit flying, or maybe even extreme ironing.

For now, I suppose I’ll just stick with this, extreme video stream calculating…or perhaps I should coin my own term: “Exstream Calculation.”