We have never been excited about CDNs in the past, but suddenly all the Over the Top video activity in Europe had a number of players at the IP&TV World forum excited about the concept and that’s enough to get Faultline following suit.
The concept came up in at least three presentations ‘ from VoD storage player Edgeware (‘it’s not about the storage, but about the software’) from codec supplier Broadcast International and from a stealth style CDN supplier to US cable firms, Verivue, and their stories were all more or less the same and they rang a bell.
The logic was very much like the message that Faultline got from Akamai over the past few years at our annual get together at Amsterdam’s International Broadcaster Conference (IBC) each September.
To put it simply, Akamai probably the largest content delivery network (CDN) in the world, had a problem a few years ago. It wanted to enable its clients to send HD video files all over the internet. The method is a tried and tested one and it involves moving content first from a central hub to perhaps a regional hub, and then, based on usage, moving the content (in this case HD Video files) to devices, usually self-contained ‘appliance’ servers, at the edge of the telco access network, close to users.
The logic is simple, popular video files, be they YouTube shorts downloaded millions of times or immense 3D HD films with racy special effects, are moved to the edge of the network once, and then stored there. The stay stored there, and are used to serve clients on multiple occasions, without ever being deleted unless there are more popular files that succeed them. As popularity moves on, this type of hierarchical feeder network shifts content that has lost its appeal back to the core and fetches the new popular content.
Of course such a network idea goes all the way back to pre-fetch instructions on disk drives in the 50s and 60s ‘ fetch data that you think the application, or in this case the customer or user, is likely to want, based on some form of prediction of behavior, or if you can’t predict it then simple fill a cache with data that has already been requested. If your strategy works then the customer won’t have to wait for a video file to be close enough to begin streaming.
We’ve seen this in the very first IPTV VoD network at Fastweb, where BitBand emerged from the process as a company, simply because a traditionally trained IT man thought the same problem through for the first European IPTV network. We’ve seen it before at Akamai and all that it has done is upgraded its topology over the past few years to identify and accelerate HD video, rather than any other files.
Storage and telecommunication speed have always been critical in the architectural development of all forms of data processing, and the mass shipment of HD files is no different, in fact DVD and CD physical logistics is no different. You keep music CDs and video DVDs in central silos near manufacturing sites, then shift them to distribution centers and then to local stores and then on to consumers. The digital is merely mimicking the physical here. But with a couple of core differences.
In a store we have a global standard for video ‘ the DVD, and so we only have to ship one type of copy of the video in its physical format. In the online format we have to a world interested in multiple forms of video formats ‘ Adobe Flash Video, as well as adaptive streams in HTML 5.0 based HTTP Live Streaming, Google WebM VP8 format and also in Microsoft IIS, and a few other formats.
If we are going to have cloud based content digestion, we have to be able not only to get HD video content to each home or each portable device, but also to support multiple concurrent formats ‘ mostly streaming, but potentially download for a little longer as well.
Flash VoD server company Edgeware talks about its new business model being a combination of a wholesale CDN, built around its 20 Gbps IPTV video storage farms, small enough to leave lying about a DSL network. Three years ago it reached out to adaptive steaming innovator Move Networks and has since mastered the adaptive streaming side of the business itself, but we figure there is room for perhaps refinement in the Edgeware model.
Step up Broadcast International, with its Codecsys system. Codecsys used to be married to the IBM servers built around the Cell chip. This was because it was super powerful and could run multiple codecs in parallel and the BI system selected the least bandwidth intensive output from a choice of 5 or 6 codecs ‘ and by swapping between them, it led to extremely low bandwidth video. It seemed like a good idea, but partnership with IBM is a tough master and in the end not enough business came out of it and the company switched to implementing the architecture on the Xeon multiprocessor Intel Nehalem chips.
There has been a sudden, perhaps unexpected, benefit from the architecture, which is that being software loaded, but fast enough to run ahead of real time, Broadcast International (BI) has realized that it can offer the system to softload a codec to support one adaptive streaming output, and then load another to support a different one. ‘We can change H.264 to WebM in real time to create a local multi-format, multi-portal head end, within the network,’ said Steve Jones General Manager of the Codecsys division of BI.
Jones talks about this being his ‘cloud based’ offering and we can even imagine a stream being viewed on one device, and the consumer shifting to view it on another, which supports a different streaming protocol. A distributed CDN with this distributed transcoding capability, with multiple styles of streaming output could cope with streaming one minute on a TV and then another on a Tablet, with all the concomitant changes in resolution and format and software stack.
‘We can then push this video to a CDN in an organized sequential workflow with all of it happening at the network edge,’ said Jones.
What’s stunning about this idea is that he already has a client which he says will use 2,000 such appliance transcoders in the network and which he expects will use 200 of them at any point in time. While Jones refused to say which telco this is, BI and Australia’s Telstra have been put together in news stories more than once and this could be its first major tier one account. Whoever it is, that size of deal will finally put BI’s Codecsys on the map and with the pressure to offer a multi-device video cloud service in Europe, that could lead to a potful of contracts.
Verivue was unknown to us before the show and although it’s been around for 4 years, and has 100 people and has taken $85 million in funds, it’s been as close to stealth as you can while working for some of the biggest names in US cable.
This is strange because CEO Jim Dolce is not the retiring type, and has a great grasp of the turf wars in an around pay TV, ‘VoD for US cablecos used to be a few satellite streams, feeding a few hundred locations with a few hundred video assets, but now Comcast and Verizon have decided to up this to 25,000 assets and have set their sights on 100,000, so it was no longer feasible to distribute all that by satellite,’ And that’s the rationale for his business.
Verivue helps operators create a scalable, hierarchical content delivery network, a bit like the ones we have been talking about earlier, with regional caches and edge cache. Since then it has been a flat out turf war up against Cisco, to nail down this business in the US and is now ready for Europe. ‘We are the de facto incumbent at UPC in Hungary and Germany and will soon be launched in their Switzerland and Ireland operations, ‘ adds Dolce.
And at the show he announced deals with Cogeco in Canada and Telekomunikacja Polska (TPSA), the Orange subsidiary in Poland. ‘TPSA will both support its own video on its CDN, but also the video of other pay TV and broadcast rivals,’ he said.
Here we are at the heart of our sudden fascination with CDNs ‘ Telcos HAVE to build them, because they HAVE to carry the massive broadcast video feeds if they are the leading broadband supplier in a given territory ‘ in the UK BT HAS to carry the BBC’s iPlayer traffic, and in the US Comcast (and its rivals) have to carry the Netflix traffic ‘ or give up their hard won broadband customers.
Dolce is the first person we have come across who has an understanding of the Comcast versus Level III and Netflix dispute. ‘Comcast says it can change the price of each megabyte of data it carries for its customers because this is no longer a peering relationship ‘ because the traffic is all one way ‘ and that’s why it feels that it can change the peering charges that have been the anchor of the internet for the past 20 years. They have never had to be changed before because one ISP would have an imbalance of traffic routed in and out of a neighboring ISP. That imbalance was payable in terms of a price for each price of excess data shifted. As the networks get better at shifting vast quantities of data, the peering charges go up, but the price per megabyte doesn’t.
Comcast felt that the one sided nature it has with Netflix, with it making up 30% of its Friday and Saturday night overall DOCSIS load, made it worthwhile upping the ante and claiming that peering is no longer the issue. It has become a delivery channel for a rival. We’re not sure there are any rules in peering relationships which allow this sudden change from peering to another type of relationship, but there may well be.
The interesting thing is that Dolce sees this idea of being a telco or cableco and having to shift your own VOD traffic, as one which naturally turns into another, where you move that same traffic over a broadband line to multi-screens, like Tablets. Even if you don’t have a relationship with someone like Netflix, it is likely that your CDN will save your network a ton of work. Better still, at some future point, assuming that Netflix isn’t killed off by companies like Comcast, why would Comcast not bid to carry the Netflix traffic anyway? If the network is doing the work anyway, why not get paid for it.
Dolce sees a future of independent CDN layers within all telco networks, supporting multiple unmanaged Pay TV vendors and video that is stored could eventually be filtered through trading hubs so consumers can look around for the cheapest deal on a film, right now, from multiple suppliers ‘ but in the end, you’d get the same CDN cached copy, it would just have a different label on the header. The studios could offer the same content to each of the ISPs I in the same way. This is a model that Faultline put forward five years ago, but it has been slow to arrive.
In a way this, reshaping into a cloud based stream capable CDN, could also change the nature of DRM and conditional access as well. You will have noticed that most of the DRM’s announced for operation in a multi-device OTT world are not the same smart chip based systems from the TV world. Instead this is a great opportunity for the pure play, software only, IPTV initiated AES based encryption systems. This is why Widevine was a good purchase for Google, and why Secure Media is perhaps right for Motorola. We know that Irdeto has its own software only system, ActiveCloak for Media, underpinned by its Cloakware system, and we know that IPTV market leader Verimatrix has its eyes firmly on supporting multi-screen devices.
Verimatrix has upgraded its VCAS DRM server side module so that it can support existing bundled DRM systems on other devices but also its own Viewright web DRM on the PC, Apple’s iOS, Android and some set top SoCs and this supports the HTTP Live Streaming protocol. If these software only, return path solutions win out for protection in what is suddenly a world crazed with the need for tablet support, then half of the world’s pay TV connections could suddenly no longer be owned by the traditional hard wired conditional access players. It’s like the landscape is up for grabs all over again. And each one needs some form of trick ‘ it must offer encryption in adaptive streams, it must enable swapping between different stream types, perhaps be distributed in its own right inside a network, or it must embrace the other existing embedded DRM systems ‘ or perhaps all of the above.
Steve Oetegenn, Chief Sales and Marketing Officer at Verimatrix said to us, ‘It’s true that there are a lot of OTT RFPs out, particularly in Europe right now and by the end of 2011 we shall see these all launched and they must address multiple devices ‘ set tops, PCs, mobiles, tablets, but you shouldn’t see these as replacements for TV services, instead they are being designed as augmentations of existing services.’
He makes a good point ‘ it is the threat of brand new internet based services which has led initially to free to air broadcasters all across Europe building their own internet delivered catch up services. The issue with free to air content is that some of it is very good, but not all of it. So that when you watch only the best programs on a catch up basis, this provides a threat to pay TV operators ‘ and it could end up looking every bit as good as a pay TV service ‘ leading to churn. Virtually all of the pay TV operators in Europe have understood that it is time to make a move on OTT ‘ but this is about containing the threat from broadcaster services, not shifting their model from their own networks to the internet.
In a funny way it risks being something of a phony revolution, with some programming available to tablets, but nothing like the full bouquet which consumers would want, and in many cases this being tied to the broadband line at the same household where the TV service is being delivered.
There is still the potential for a pure internet play to emerge, and when we talked to Sezmi at the show, a company we have come to think of as Los Angeles based, and operating a long way from Europe, we can see the reason to fear a pure internet play. Initially Sezmi was a three way hybrid, broadcasting some cable channels, picking up ATSC broadcasts with an innovative indoor high gain antenna, as well as taking VoD programming over the internet. Outside of its native LA it now does everything which is not already being broadcast over the internet, whatever the internet carrier technology.
This is now available in two US bouquets, one for $20 and one for $5 in 30 US TV markets, with retail help from Best Buy and Amazon. But the company, driven by two experienced entrepreneurs, Phil Wiser who made a killing with Liquid Audio, and Bruno Pati, who ran Numerical Technologies, are not the type to rest on their laurels.
The surprise was the way in October the company announced its second OTT TV service in partnership with Malaysian power utility YTL which is building a hybrid pay TV service as part of a WiMAX based quadruple play, initially in Malaysia then spreading across local territories.
At the show Sezmi launched yet another deal, this one in Mexico and Wiser’s off-hand comment that ‘there will be many more in Europe and Asia,’ is enough to send a shudder down the backs of the pay TV operators who may think that a half-hearted, interactive friendly portal, which only uncovers a few choice pieces of video, will be enough. It won’t be enough against this type of threat, not with its own $149 Sigma chip based DVR and a dramatic cost saving for any pay TV home.
Wiser says that he favors approaching telcos who missed the IPTV boat, and the US system is sold their by smaller incumbent telcos. So expect the same approach in Europe and Asia, a major business that wants to shake up broadcasting and Pay TV, a major retailer and a distribution chain of ISPs and other telcos. Effectively Sezmi built its own system at home in the US, and is now exporting the concept through partners, selling the software and partnering on the upside.
The new Sezmi deal is with Grupo Salinas is for an OTT hybrid multi-screen TV service and the company was not originally a pay TV operator and is instead growing out of its banking and financial routes in to broadcasting and is now building out fiber to become a superfast telco service. The company has taken all of its infrastructure from Sezmi and is planning a market launch within 6 months. If you can undercut pay TV in Mexico you can probably undercut it in most countries, although the ARPU of Pay TV rivals there is around $22, so still way above the Sezmi target pricing, even for the US and it’s price is bound to be lower in Mexico.
So suddenly the CDN has become fascinating, perhaps not the precise technology of how it works, though expect further innovation at the network level, but because of the co-operative business models that it can bring to enrich both the free and paid video eco-system.