|
 |
| |
|
|
Another Take on Joost, BitTorrent, Move, Netli,
U-Verse & FIOS |
|
|
|
| |
|
|
|
|
|
| |
|
I am frequently asked about InterStream’s competitive positioning versus
a number of other “video distribution infrastructure” solutions that
the media industry believes have been gaining momentum. Frequently,
most businesses are advised not to even consider alternative models as
competition. However, in this case, the market is so nascent, the
business models and technologies so raw, that the emerging market differences become very apparent.
I thought it would be useful to consider the players who many of my
colleagues in Hollywood and the media industry consider to have
momentum behind their models. Let’s consider how these services
are evolving as well as the tenants upon which they’re based and where those will likely take them.
All We Need to Do is Build a Bunch of PoPs
Points of presence, or “PoPs”,
are locations on the Internet which act as place to originate or
redistribute any form of content. There really isn’t much
architectural difference in a Peer-to-Peer (P2P) network like BitTorrent, a CDN, or a “grid”.
They all rely on having a significant number of locations on the
Internet, which have relatively short latencies to reach their
destination (see my previous post on this topic).
P2P networks, however, do this much differently than other
“structured” approaches. These networks are “ad
hoc”. They rely on end-uses to download software to put up those
PoPs. In other words, PoPs can be engineered methodically into the
network like we have with a CDN today, or the media grid. They can also be built somewhat pervasively but randomly through end-user downloads like we have with P2P networks.
Joost has based its business model on the later pardigm. Although, this time around the guys who gave us P2P system KaZaa, and VoIP system Skype,
aren’t relying on giving the user something away for free.
Instead they’re focused on offering non-pirated content from major big media players. This puts a crimp in their plans to build all those PoPs. How do they offer an incentive to an end-user to download their player?
Existing and past users did this primarily to get pirated content, or
in the case of Skype, to get free phone service. How this works on a
network, which is entirely lacking in bandwidth for true Internet TV is
anyone’s guess at this point.
Pirates are our Friends
P2P offers another model. In most cases, users download these clients
to “share bandwidth” with their neighbors. In reality, the
vast majority of users simply pirate content with these systems. Highly
popular content, whether it be music or video, is easy to find and
download. The “long-tail” can
be quite a bit more difficult to locate or find sufficient other P2P
clients to “share” it. ISPs have begun degrading, or throttling P2P networks over the past couple of years. It is unclear what this really means. In the U.S. we can expect the FCC will step in
and rule whether this throttling is simply necessary for “network
management” purposes, or whether it truly violates the principles
that some net neutrality proponents espouse.
Regardless of the FCC outcome, it is clear from a media industry
standpoint that P2P has been the greatest threat to the music industry
and could likely be the most significant business model threat to the
television and motion picture industries. Piracy threatens both the
television advertising model as well as through “direct
theft” of monetizable content, itself. Music has only had a
direct sell-through revenue model. TV and movies don’t.
Therefore, the Internet distribution systems enabling and supporting
piracy could cause something much more dramatic to happen than has already happened with the recording industry.
One thing virtually now web-based video distribution system can beat is
cost. P2P systems very effectively leverage “shared uplink”
bandwidth from broadband consumers. This “free” bandwidth
is estimated to have one thirty-fifth (1/35) the cost of the lowest web-based distribution systems. Bandwidth cost is expected to be a key driver in Internet video distribution moving forward.
Thus, for legitimate distribution of non-pirated content, like a Linux
distribution, P2P really can’t be beat. However, it is because of
all that pirated content that there are even enough PoPs out there to
make the P2P system so effective. I think the best answer for both ISPs
and the media industry will be to throttle P2P while allowing and
encouraging high bandwidth distribution of legitimate content (see my previous blog).
The Best-Effort Paradigm will Always Be
Move Networks
represents another point of view. As some ex-Novell colleagues of mine,
I greatly admire what they want to accomplish. Unfortunately, I
don’t think their business model can truly scale to bring us real “Internet television”.
They operate a CDN (or P2P?)-like
service, which assumes that we will always have a best-effort paradigm
for bandwidth. Inasmuch as this is true today, this is really not true tomorrow. Tiered services exist today on the network, and I contend will exist in much more significant way in the future.
In essence, Move changes the quality of the viewing experience based on
the available bandwidth. While this creates smoother playback than the
“buffering” and “stopped” messages we
frequently get today, it doesn’t solve the problem generally. It
also cannot provide a consumer with a true Internet TV experience
because there simply is not enough bandwidth in the bottlenecks and
backbone of the network to enable even good standard definition
television in most locations during much of the day. Of course, in
places like Korea and Japan, they have much more bandwidth but due to
the best-effort nature of their networks, they still can’t offer,
on average, a truly seamless viewing experience like we have with
television today.
I will grant that the system works fairly well during
“off-peak” broadband usage periods. It certainly has a
place in the overall Internet Video ecosystem. I just wonder if once
tiered bandwidth services are broadly implemented for a small fraction
of the cost higher than best-effort services today if their model will
provide any true differentiation in cost or the viewing experience. Consider the picture below:
Let’s Forget about Being Polite on the Network
Another approach about which I’ve previously blogged is to “game the network”. Netli, which was originally started to improve the performance of HTTP over "their TCP", was acquired by Akamai
a few months ago, effectively does this through exploitation of a
performance defect in the bottlenecks of the network. I wrote a
detailed explanation of this in my previous blog.
While this does give CDN services like theirs a leg-up, it really
doesn’t put in those “diamond lanes” necessary for
Internet TV. In fact, I hope Akamai can act as a partner in offering
InterStream services by participating in the OSPA.
Despite the overly aggressive, or "impolite" behavior of protocols like Netli's, it proves on key point. ISPs cannot not mediate
their networks. I use the double negative for emphasis. If the ISP
chooses to do nothing, they're allowing Akamai and others to mediate
the network for them. If they choose to sign the OSPA, then then they choose to manage their bandwidth allocation. By
deploying the mediation technology
or by simply participating in the OSPA, these same ISPs will be
creating those diamond lanes on their network. Why would any broadband
ISP allow only the CDN to be making money off those diamond lane
services? To level the playing field, we beleive, most ISPs will choose
to sign the OSPA and deploy the counter-measure technology on their
networks.
A Walled Garden is What the Customer Really Wants?
AT&T and Verizon, plus a few other international ISPs have a
different model – “IPTV”. With IPTV the broadband
ISPs compete with existing cable and satellite services. Whether they
actually have enough bandwidth on their networks to compete is another matter.
IPTV is not an “open internet” approach, however. That same
bandwidth that may be prioritized to deliver TV by their service may
not necessarily be used for general Internet access. Effectively, IPTV
services are the only ones who get those “diamond
lanes”. Therefore, whether it be a YouTube, Joost, Hulu, or
some other offering, they don’t get the same level of access unless they somehow negotiate a contract directly with each broadband ISP (some estimate close to 10,000 of them in the world).
Of course, this approach can’t scale generally across the
Internet. There is no specific protocol or set of technical standards
that can be easily used by the whole ISP industry.
Instead, it either represents a set of “one-off” agreements
and technology or the current cable TV paradigm. This doesn’t
allow them to do “video snacking”
in high quality or watch advertising sponsored long-form
content. The consumer has spoken. They want Internet TV, not IPTV, and
they want it broadly across the Internet like we have with our Voice
over IP (e.g. Skype/Vonage) services today.
Website-based and P2P-downloaded Video is where all this is heading…
So, where are we going with all of these approaches? In one sense,
distribution cost has a direct bearing on the quality of experience
that can be offered with video on the Internet. P2P networks will never
be able to effectively compete with the other methods when it comes to
that quality. Direct, web-based distribution will have a higher cost
and also offer a much better experience in the end to the user. P2P
will continue to have a major position unless ISPs are allowed to mediate
it. None of us really expects that the Internet will be 100% mediated,
however. Therefore, it will be important to take a pragmatic, hybrid
approach to solve the problem. The InterStream wiki documents technically and business-wise our approach.
Jeff Turner
|
|
|
|
