Mobile Media needs Mobile CDNs
The traditional content delivery networks (CDNs) of the mid-1990s were designed to move web content close to the narrowband last mile of modems and ISDN connections. As I’ve written about here before, even today in the broadband age CDNs are booming because digital media is pushing large pieces of content online, with the promise of more to come in the form of movies and television shows.
And now, we’re seeing new CDNs focused on delivering digital media over the limited data rates of today’s cellular networks. The problem is that while cellular appears to be migrating to higher data rates, only 3G (third generation) networks promise data rates above 144 kilobits/sec that are close to the slowest terrestrial broadband data rates. So cellular nets are going to need help — from CDNs that are mobile-specific.
At slower speeds, downloading digital media can be a frustrating event (imagine using YouTube over a 56 kilobit/sec modem). As an example, the AT&T cellular network in the U.S. supports EDGE technology that can theoretically provide a maximum of 470 kilobits/sec although anecdotal evidence suggests that the average user regularly gets less than half of that data rate.
To solve the data rate problem, the cellular industry is doing a lot of work and there are a number of competing technologies (including what the industry is beginning to call 4G) that should come to market over the next few years, including mobile phones with Wi-Fi and WiMAX. Even with these efforts, it is clear that the cellular network itself will be the skinny straw between large pieces of digital media and mobile devices for the quite some time. And, I suspect that as the data rates increase so will the size of the digital media – the cellular networks are where the Internet was in the mid-1990s and needs help catching up.
That brings us back to the new CDNs that are mobile-specific. Mobile-specific CDNs are deployed close to the edge of the cellular network and provide features that are unique in the cellular world, such as shaping the data rate depending on coverage and re-formatting the content as necessary for specific device types. In theory, using a mobile-specific CDN, most of today’s digital media could be easily consumable on a mobile device.
So, if we want to download music and watch videos on mobile devices, a mobile-specific CDN focused on bringing the content as close to the edge of the cellular network makes sense. The big question is – will the market provide any pull for digital media on our mobile devices? Are you downloading music and videos to your mobile device on a regular basis now? And how is the experience?
Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.
Whether or not there is a case for mobile CDN’s is less dependent upon the technology and more on the businesses owning the network. If content-owners would be able to treat mobile IP-networks as normal networks, they would quickly develop websites and content targeted at the mobilephone and the various screensizes. Eyeballs are king and so only an idiot would not develop for the 300 by 300 pixel screen-crowd if its large enough. Bringing content close to the end-users only means bringing it over fibre (or anything with enough bandwidth) to the cell-site. In the backhaul part a mobile network is no different from a DSL network and there should be no problem moving the data over the fixed network. So there is not a need for an Akamai for the mobile crowd.
Formatting content specifically for certain platforms is probably best done by the owner of the content. There is no way one can automagically create a mobile site from a fixed site that is optimized for 1024×768. This is most hard for the look and feel and user interface. Storing a specific movie encoded for mobile devices is not that hard and if one has the license for encoding windows media or adobe it shouldn’t cost too much either. so one might conclude that there is no reason for Mobile CDN’s
However Mobile phone companies don’t like the internet. They don’t want it on their phones. Europe is hardest hit in this regard. They want to be able to bill a user for each and every youtube movie seen. And they would prefer to be able to charge different prices for different qualities etc etc. Technically there might not be a reason, because it really doesn’t cost extra to send more bits, but practically they will try to find a way. It’s a pity, because the end-user gets deprived of a whole lot of consumer surplus.
Bonus question: If we compare the roll-out of mobile internet to the roll-out of fixed internet and say that mobile internet started in 2001 and fixed internet in 1994 at what year in the evolution of fixed internet is the mobile internet now? My bet is on early 1995, still all sitting in walled gardens and almost nobody reachable by e-mail. It took them 6 years to go 12 months
This is an interesting tidbit, but I am a bit perplexed by a post that talks about a new market yet does not disclose the name of a company in it. Maybe you could update the post with some references to companies in the mobile CDN business to validate your statement: “And now, we’re seeing new CDNs focused on delivering digital media over the limited data rates of today’s cellular networks.”
I, for one, would really appreciate the addition. Thanks.
Raindeer – I absolutely agree with your first statement. We need to cellular providers or the content owners to own the job of moving the content. The technology is there to do website transformation and to play media over the cellular networks, although the experience may be less not of the best quality. However, I would contend that we need a CDN for mobile because, just like the traditional CDNs, moving content close to the limited last mile (or 3G connection in this case) can produce substantial performance benefits if done intelligently. Various folks will argue their methods are the most intelligent and I’m waiting for the market to vote.
Alex – I did not cover specific companies as I have spoken to a few, but not all of the players in this space. Also, Katie did a nice job speaking to specific companies earlier this year in this post http://gigaom.com/2007/02/19/mobile-video/.
ROTFLMAO!!!
Oh this just brought back a memory of mine with Sprint PCS. Prior to the Virgin MVNO announcement and the 1xRTT light up, in a meeting with the folks at Sprint (who always pride themselves on being the most innovative wireless data operator) I asked these guys what kinda of CDN setup were they planning, and all they could do was ask what’s a CDN. Opps. lols.
The biggest obstacle to our intuitively obvious wireless future are the carriers themselves. The sooner they conform to Googin’s Law the better we’ll be.
CDNs in the early internet solved the core network having insufficient speed to deliver to the edge.
In mobile, the bottleneck is not the core network. It is the wireless link from the tower to the HS. Placing a CDN at the edge (ie at the tower) does not solve the problem.
Granted, transcoding, traffic shaping and other techniques to better format content for HS constraints helps, but this is not the core CDN move content to the edge argument.
Excellent discussion, and I believe tracking on the right path. Particularly important are the comments about early CDNs and the core problem they solved. Getting content to the edge is not the problem with mobile, ensuring a consistent user experience using a highly variable network is the problem. Upgrades to networks will help some with downloads, where the average bit rate, combined wtih error rate, will determine the speed of the download and in turn the user experience. However, with real-time applications like streaming, managing instantaneous network conditions is the key to ensuring high QoE. Having spoken with numerous mobile operators, I am convinced they get this, even if they don’t all have a clear plan for how to address it. Adding complexity is the proliferation of networks and devices. Mobile CDNs, if they are to add any value, must address both network/device complexity AND network fluctuations to improve the user experience and offer content providers and carriers greater efficiency. Thanks for launching this conversation Allan, it is critical for this ecosystem that we all realize that existing CDNs won’t solve the core problems associated with mobile media.
It would have been helpful to use at least one real-world name or example to help illustrate the (quite good) point you’re trying to make. Who (or what) were some of the CDN’s used in the broadband world, and who is trying to replicate something similar for cell networks.
Inquiring minds want to know…
In the broadband internet, CDNs solved the latency at the browser by caching content at the ISP points of presence (POP) as close to the end user as possible and by redirecting the user to that POP. The aggregate bandwidth of the last mile networks was several multiples of the bandwidth in the core, so this worked well – at least for read only assets like images and media downloads. I designed a few of these myself from my time at CERFnet and AT&T, and believe they are an elegant solution.
The mobile environment is different. To be sure, the basic physics of bring content to the user, rather than the user to content, to improve performance works – but that is not quite what makes a mobile application compelling.
Several companies, now in the dead pool, tried their hand at transcoding networks that would re-purpose popular internet applications and content for mobile. This only worked for the most basic of text oriented applications, got you out of the gate quickly, but did not require large bandwidths. These companies, from the days when WAP was going to solve all problems, were arguably early and mobile networks and devices are barely getting to be usable even now.
None the less – the big insight was that compelling mobile applications are not just re-purposed broadband applications – they need to be woven into the “mobile context” with a big nod to “attention management”
The business mobile user is time constrained – they want must-have information pushed to them modulated by their presence, availability, location, role, device type and event subscriptions.
The consumer mobile user is looking for online or live media content in bite-sized chunks to match the bite-sized attention slots they have available when they are mobile. (The rest you can arguably bulk load like you do on your ipod today)
Finally, the mobile environment is also more suited to the publication of user generated content (impromptu pictures, video clips). This upstream publishing is only marginally improved by edge-based CDNs.
None of this therefore screams for mobile CDNs in the near term.
I would argue that the “walled garden” (fixed decks, carrier lock on location and presence data etc), lack of content and of applications purpose built for the mobile environment are what needs to be fixed first.
its like a 1998-1999 boomtime for datacenters, cdn and asp(now called SaaS) which the 98-99 boom @ the time was ten years too early. back to the point that was brought up about solving for the last mile on wirless networks. it wont work if cdn’s go old school and put node in wireless sites(the spots where the network converts to airwaves from terrestrial) because in order to make it work each cdn would have to build a distribution node in each wirekess site and thatll be tough on way too many logistical, physical, operational fronts to make it work. if they go this route of buiklding distribution nodes in wirekess centers it could work if the cdn’s agreed to peer with each other. theoretically in doing so they ecome resellers of each others services and eliminate the need to have each cdn build their own infrastucture out in each wireledd node. that would entaik lots of cooperation amongst the cdn players and hell has a better chance of freezing over than akamai cooperatng. IMHO wifi or wimax is the ticket and can work in an almost peer to peer type architecture with all wifi nodes cacheing the data and re-serving it to its connected users. this would allow for revenue dharing amongst the node operator, hw vendor(if its free which eventually it will be free to the user even if somebody is subsidizing giving away products that, over their lifetime, will consume power valued @ more than 10X the cost of the hw in the first place. power companies or advertising companies will find the cost of giving it away is less than the revenue generated from marketing to you. that last scenario is pretty grassroots from a real high level but grassroots works and works quite well with social interaction driven svcs which wireless 2.0 will be…we’re barely graduating wireless 1.0 now
Allan…great post. Very timely discussion with the recent announcements around Sprint TV and Verizion’s mobile TV offering.
To answer Alex and Matt’s question regarding who is in the Mobile-CDN space, I wanted to point out Ortiva wireless who were covered in Katie’s recent post,
(http://gigaom.com/2007/02/19/mobile-video/)
The variable nature of mobile networks does not lend itself for content delivery. The elasticity in the network requires an elastic delivery mechanism in order to deliver a crisp experience with mobile content.
Because Ortiva’s M-CDN technology functions at the application layer, they are able to optimize mobile content with existing networks and handsets.
By monitoring the feedback loop from the handset and analyzing the data to evaluate network conidtions for each user, Ortiva’s M-CDN can make real time adjustments to the content being delivered.
Using the M-CDN, mobile carriers are able to deliver optimized content in crowded network conditions, using less bandwidth.
The ability of mobile carriers to deliver optimized content that will engage the end user will be key for widespread adoption of mobile TV/video.