[opendtv] Re: Internet TV distribution architecture
- From: Craig Birkmaier <craig@xxxxxxxxxxxxx>
- To: "opendtv@xxxxxxxxxxxxx" <opendtv@xxxxxxxxxxxxx>
- Date: Thu, 9 Jan 2014 08:52:58 -0500
> On Jan 8, 2014, at 5:10 PM, "Manfredi, Albert E"
> <albert.e.manfredi@xxxxxxxxxx> wrote:
>
> Craig Birkmaier wrote:
>
>> The actual location of the servers is largely irrelevant.
>
> Only if you try to twist arguments just to get the last word.
>
> Mirrored servers have to be close to the end user, within ISP networks, when
> the content is popular and the owner wants a good user experience. More so,
> the higher the bandwidth. The numbers are as plane as day.
The facts say otherwise. The vast majority of U.S. ISPs do not host CDN
servers, although many international ISPs do.
But there has been a great deal of activity in the area of ISP/CDN
collaboration in recent years, and we may in fact see more ISP collocation of
CDN servers as the streaming video market continues to grow. A quick search
turned up two related articles that take the pro and con view of collocation.
http://blog.streamingmedia.com/2013/01/netflix-announces-major-isps-deploying-their-cdn-caches-new-3d-streaming.html
http://blog.streamingmedia.com/2009/04/more-isps-not-letting-cdn-place-servers-inside-their-network-doing-it-themselves.html
> Therefore, local servers are needed before the masses can use Internet
> distribution of TV content. Therefore, the local broadcasters, some of them
> at least, should have an opportunity here. And as an added bonus, local
> broadcasters playing this role could continue the practice of local ad and
> news spot insertion. Which the TV networks might appreciate. Or they can let
> existing CDNs take up the slack, obviously.
The masses are already streaming a huge amount of video. Clearly, if the MVPDs
and broadcasters went away we would need a new ISP infrastructure with mirror
servers. But this is not happening...
Yet. And it is unclear if or when it will happen.
In order for broadcasters to get into this game they would need the rights to
stream the content the broadcast today, and any additional content they could
license and distribute.
A move to LTE Broadcast would essentially replicate what they are doing now;
only the transmission technology would change, although some new opportunities
would be enabled. For example, programming and ad insertion based on sub
markets.
If broadcasters were able to get the streaming rights for a market, they would
still need to have servers to deliver the streams. Where would these be
located? There are multiple ISPs in larger markets, so a broadcaster would need
to collocate with all of them or use large pipes in the Metropolitan Area
Networks to distribute the streams.
>
>> The post 911 situation was an expected anomaly given the spike in
>> usage.
>
> And so would mass migration to Internet TV viewing.
No. This could not happen without the necessary scaling of the infrastructures.
>
>>> Here is the information that was already provided to you, but which
>>> you should seek out on your own now, to make it stick. For LTE in
>>> broadcast mode, what is the relationship between b/s/Hz and tower
>>> spacing?
>>
>> It depends on what you are trying to do. If you want to cover an
>> entire market with the same IP a Multicast the number of sites can
>> be small
>
> We went through an *excrutiatingly* long period before certain people finally
> realized that the distance between towers in any SFN, had impact on the
> spectral efficiency (hint: guard interval). Originally, DVB-T was rather
> constraining, then DVB-T2, for FIXED service, relaxed that constraint. For
> mobile service, where symbol duration has to be short, the tower spacing
> constraint remains much like it was with DVB-T1, even when using DVB-T2.
>
> Now we have a 2-way medium, LTE, designed for mobility and for small cells,
> which includes a multicast-broadcast mode. Let's not go back to square one.
> Just like DVB-T, there are specific constraints to tower spacing, in an SFN,
> when LTE is used. And it should not be surprising if the LTE tower spacing
> constraints are more severe than they are for a one-way big stick broadcast
> medium like DVB-T.
>
> So I ask you again. In order to approach the spectral efficiency of ATSC,
> what is the tower spacing needed with LTE? You really need to dig up that
> number.
>
No. It is irrelevant.
Broadcasters need a new business model and the infrastructure to support it.
Spectral efficiency can be measured in many ways, based on application and
transmission infrastructure. What we DO KNOW is that in terms of spectral
efficiency and potential spectral reuse, the big stick broadcast model is
terribly inefficient..
Half the available spectrum cannot be used to protect the portion that is used.
Stop wasting our time worrying about bits per Hz in a single 6 MHz band. How
broadcasters use the several hundred MHz they operate in is the real issue.
Regards
Craig
Other related posts: