The year 2007 produced ample proof of consumer demand to consume and produce all manner of media and social communications on the best-effort Internet. In 2008, that proof will add up to a competitive imperative among backbone, content delivery and local access network operators to invest in not only more capacity, but in enablers for what might be called a ‘better-effort’ Internet.
One area of industry response to market demand has already proved obvious and widely reported: both fixed and mobile operators are ramping up the capacity of last-mile access networks in 2008 with deep fiber, VDSL, DOCSIS 3.0, 3.5G, Long Term Evolution (LTE), WiMAX and other technologies. By late 2008 broadband operators could begin to add more upstream as well as downstream capacity. That would radically multiply the sources of multimedia traffic on the Internet, including consumer generated video, multiplayer game-playing and multimedia telephony and messaging traffic generated by millions of users.
However, in that scenario, international, national and regional backbone transport networks could become an equal gating factor, as they must support the aggregation of millions of multimedia uploads, downloads and real-time streams, or what the Internet Innovation Alliance (IIA) calls the “exaflood.” The term ‘exaflood,’ based on the measure ‘exabyte’ (more than 1 billion gigabytes) and coined by Bret Swanson of Seattle’s Discovery Institute, is meant to describe the torrent of data that the Internet will have to handle in the very near future.
Video, of course, is the major driver of the coming flood. The IIA – a non-profit coalition dedicated to universal broadband availability, with members including AT&T and Level 3 – notes that downloading a single half-hour television show consumes more Internet bandwidth than receiving 200 emails a day for a full year, and downloading a single high-definition movie consumes as much bandwidth as 35,000 Web pages.
Further, even more bandwidth-hungry high definition video streaming will become a reality in 2008. Walt Disney Company already delivers a form of HD from its Web destinations. South Korean ISP UBalance Co. Ltd. is set to deploy a Matrixstream 1080p HD-capable platform for over-the-top IPTV service in Korea early in 2008 (see ScreenPlays, October 2007). And other streaming video providers like the well-funded Joost and Hulu are all but certain to follow suit to full-screen and HD in the coming months.
According to the IIA, the world created approximately 161 exabytes of digital information in 2006 and by 2010 might produce as much as 988 exabytes of data. Such a demand leap is cause for concern, the Alliance says, based on a study released by Nemertes Research in November. That study concluded that Internet access infrastructure, specifically in North America, will cease to be adequate for supporting demand within the next three to five years.
“The financial investment required to ‘bridge the gap’ between demand and capacity ranges from $42 billion to $55 billion in the U.S., primarily to be spent on broadband access capacity; this is roughly 60-70 percent above and beyond the $72 billion service providers are already planning to invest. Required investment globally is estimated at $137 billion, again primarily in broadband access,” Nemertes says. “It’s important to note that even if we make the investment necessary between now and 2010, we still might not be prepared for the next killer application or new Internet-dependent business like Google or YouTube,” says IIA co-chairman Larry Irving.
However big the flood proves – and that depends keenly on the still uncertain ability of industry players to monetize multimedia content – operators of global, national, regional and local IP networks are in for another race to boost bandwidth. And with expanding bandwidth comes the need to tame the complexities that attend the rapid rise of video, social networking, photo-realistic role-playing environments and other ‘killer applications’ yet to emerge.
Backbone Capacity
Near the end of 2007, U.S. backbone operators began to commit to substantial capacity upgrades in the coming year. IP Multi-Protocol Label Switching (IP/MPLS), optical switching and Carrier Ethernet technologies figure prominently in those commitments.
XO Communications announced in December that it will double its backbone capacity in 2008. The upgrade employs the Cisco CRS-1 Carrier Routing System in the core of a nationwide IP/MPLS network to support services including high-speed Dedicated Internet access, IP Transit, IP-VPN and VoIP services. It follows a recent XO deployment of 800 Gbps of capacity in core long-haul network nodes on major coast-to-coast network routes, which itself augmented a 2006 capacity expansion to deliver 400 gbps of capacity across the entire XO long-haul network.
AT&T, which also made a major purchase of Cisco CSR-1 backbone routers in early December, and Verizon communications have indicated their intentions to upgrade their transport networks to 100 gigabits-per-second from 40 gbps.
According to a December 4, 2007, report by Light Reading, AT&T will have spent most of $18.5 billion in 2007 on capacity upgrades, and CEO Randall Stephenson insists demand for these upgrades is no longer speculative. “The bandwidth glut of the 1990s, all those companies that took writeoffs – it’s gone,” Light Reading quoted him as saying.
In late November, Steve Craddock, senior vice president, new technology, office of the CTO, for Comcast, told attendees of the ScreenPlays CableNEXT conference and expo that one of Comcast’s top goals now is to add itself to the select few national long-haul IP providers able to carry that and other IP traffic entirely on its own backbone. “A Gigabit Ethernet backbone quickly turned into four gigabits,” Craddock said, “and now maybe 100.”
Service-Intelligent Transport
As always, operators see pure capacity as short of a complete response to market demand, particularly as they face the complexities of managing divergent service types over increasingly integrated networks and protocols. Ongoing projects to more fully migrate from legacy frame relay and ATM networking technologies to more efficient IP, MPLS and Carrier Ethernet technologies provide evidence that operators of all stripes are prepared to invest in service-intelligence technologies to support easier launch and management of new categories of services.
In late 2007, for example, tier-1 backbone and metro network operator SAVVIS deployed a Cisco-based multi-service, application-intelligent transport platform that integrates datacenter, metro and backbone networks – an initiative, the carrier says, that ensures customers end-to-end control of individual applications, and that sets a competitive benchmark for the economies of carrier-managed application transport services.
Similarly seeking a combination of greater capacity and granular, multi-service intelligence, Cox Communications is looking beyond 100 gigabits with its deployment of Sycamore Network’s SN 9000 Intelligent Multiservice Switch, announced in September 2007. The switch incorporates networking and management software to support automated provisioning of high-speed pipes at aggregated capacities of 80-, 160- or 320-gigabits-per-second.
In addition to massive capacity, the Sycamore switch supports circuit, cell, frame and packet-based services including Metro Ethernet Forum (MEF) private line and virtual private line service profiles. This any-port, universal service capability is in line with operator demand for increasingly unified Layer-1, Layer-2 and Layer-3 platforms as they seek to more cost effectively funnel all manner of enterprise and consumer service traffic into ‘exaflood-ready’ metro and backbone networks.
Along these same lines, British Telecom in December expanded its network procurements from Ciena, adding metro-to-backbone optical Ethernet transport and switching solutions for BT’s 21st Century Network. Leveraging the remote provisioning and monitoring virtues of Ethernet services, the operator says, will produce “exceptional” service management, quality of service (QoS) guarantees and end-to-end service level agreement verification.
These and other operator moves suggest a consensus that renewed, long-term investment is warranted in advanced technologies designed to impose granular, per-service delivery controls as the industry struggles to meet transport demand in 2008 and beyond.
The Better-Effort, Managed Internet
A key theme defining much of 2008 network technology investments: the boom in multimedia traffic will continue to be taken up by a next-generation managed Internet, or intranet, connecting multimedia content owners with billions of consumers. The industry, in fact, has begun to overbuild the ubiquitous, but less than fully managed, best-effort Internet with a complex of fully managed content delivery networks (CDNs).
Defining themselves by the ability to ensure end-to-end QoS controls for real-time applications like video, voice, online games and other photo-realistic online environments, CDN operators are both expanding capacity and fundamentally refining their storage, transport and asset management architectures to intelligently distribute content to local sites. Their common goals are to achieve shortest-hop transport, minimized latency and quality of experience (QoE) levels that aim to match those of self-contained consumer HDTV, DVD, DVR and game console devices.
In the December issue of ScreenPlays, CDN operator Limelight Co-founder and Chief Strategy Officer Mike Gordon explained how Limelight is aligning its business processes and technology development with leading software, infrastructure services, managed hosting, e-commerce and other partners to assemble an ecosystem of value-added services that content owners, broadband operators and consumers can leverage to support the emerging rich media Internet.
Started specifically to support video, music and games, Limelight expected its revenues to top $100 million in 2007, up from $63.5 million in 2006 and $24 million in 2005. The company’s media centric benchmarks for end-to-end application performance exemplify the new bar that is being set for a more finely managed, better-effort Internet around the concept of end-user QoE.
“It should start as soon as I push ‘Play,’ and it should play continuously,” Gordon says. “It should take full advantage of the screen or form factor that I’m connected to. That set of things should happen every time, and it shouldn’t be limited to only the most popular 50,000 or 100,000 songs in the Rhapsody Music library, but for all 3 or 4 million. That’s the vision that…defines and drives our infrastructure. Our design point is to deliver a media-quality experience to every user for every title every time.”
To meet these benchmarks, Limelight owns and lights up its own dark fiber transport in metro areas, and it rents or leases by wavelength in the wide area network. In early 2006, Limelight reached the one terabit-per-second delivery capacity milestone. As of October 2007, Level 3 had deployed 700 gbps of CDN capacity and, like Limelight, expected to exceed one terabit per second by the end of the first quarter of 2008.
“We won’t use the Internet to move it [multimedia content],” Gordon says. “We’ll use an optical network that Limelight operates, that is used only for this just-in-time delivery of content objects, and that runs at 20 to 40 gigabits per second depending on which city you’re in. You end up with a combination of lots of content that’s really close to users married to an infrastructure that, for that content that is sort of too long-tailish to be kept everywhere, we’ll move it on a just-in-time basis right to the end user.”
While some operators like SAVVIS have exited the consumer CDN business to focus on what amounts to managed enterprise CDN services, others are entering the consumer CDN business and investing in long-haul and metro transport to interconnect with local broadband operators, and investing in highly distributed multimedia servers and other datacenter infrastructure. Both backbone and metro network operators must build more and more fiber and intelligent routing capacity to meet this demand for managed transport, as increasing numbers and kinds of CDN operators accommodate this Internet overbuild for multimedia content distribution.
Companies taking a high profile in the CDN and distributed media hosting markets now include not only pioneers Akamai Technologies and Equinix IBC, and relative newcomer Limelight, but also InterNAP, which acquired CDN operator Vitalstream Holdings in 2006 as an overlay upon InterNAP’s global private Internet services infrastructure; tier-1 global Internet backbone operator Level 3, which acquired SAVVIS CDN assets in 2007; and VeriSign, which acquired peer-to-peer (P2P) media infrastructure operator Kontiki in 2006 as a complement to its extensive, global data center, domain management and secure transaction management infrastructure. In December 2007, Walt Disney Co. venture capital arm Steamboat Ventures invested in CDN operator EdgeCast Networks.
To accommodate all this CDN activity, Equinix IBC has been building new major data centers in Chicago, Tokyo and Washington, D.C., in 2007. The cost of Google’s new data center in Lenoir, N.C. has been pegged at $600 million, only hinting at the cost of more secretive and massive Google construction in The Dalles, Ore. According to it.slashdot.com, Google has been constructing two more large data centers in South Carolina with an estimated $950 million price tag, while Microsoft pursues a $550-million build in Texas.
Billions more dollars will be spent by CDNs in filling these datacenters with multimedia servers and in acquiring fiber metro and backbone transport in and out of the datacenters.
Content Delivery Management
Heightened CDN competition is leading to further investment in advanced distribution architectures and management intelligence, as well as in advances including policy-based automation for syndication support, advertising data collection and other value-added services.
“The real thing in growth is not just the underlying network but services and functionality on top of that,” says Akamai CTO Mike Afergan. “That means services that help enable customer business models – things like syndication, upload, pay-per-view or ad-based models.”
Major media companies are seeking “a comprehensive solution, not just to deliver a file, but policy around that file is critical,” he says. “U.S. companies are interested in delivering internationally, as are Nintendo, Sony and companies in Korea looking for a global platform that eliminates the need to build data centers around the world or the need to worry about an earthquake hitting Taiwan.”
The drive toward such comprehensive solutions will define CDN operator investments that just began to gain steam in late 2007 when, for example, Akamai launched a new TheHDWeb.com portal providing a package of services to content purveyors including support for VC-1 and MPEG-4 encoding; delivery and storage of files greater than 2 Gigabytes in size; support for files with resolutions of 720p, 1080i and 1080p; and provision of “client-side technology as appropriate.”
In October, Limelight introduced LimelightHD to meet demand for high-definition content delivered via the Internet to PCs, TV set-top boxes, media players and game consoles. In that same month, Level 3 announced a new IP video streaming service that will be packaged with an HD video player that content owner customers can offer to consumers as a downloadable client application.
On the advertising and monetization front, Limelight logs every delivery of every object at its edge servers, then centrally aggregates the data married to geographic information and exposes the results to content owners and advertisers via its Lux reporting interface. “There is a wealth of information there about what their customers, in aggregate, are doing, which is so important to our customers in terms of their monetization strategy,” Gordon says. “We’re making some significant investments in improving the depth and breadth of the reporting information.”
Competition over CDN efficiency also will continue to drive investments, particularly in refining distribution infrastructure that promises to reduce latency between a content server and the end user.
Level 3 claims its network offers the fewest average hops or connections to any given IP address in the world. The company says it facilitates the shortest route to 75 percent of ‘eyeballs’ in the United States through direct Internet connectivity to the largest broadband providers in North America. Similarly, VeriSign’s “Intelligent CDN” service emphasizes intelligent routing – rather than the more random routing of traditional P2P – to cut media delivery costs by up to 40 percent through what VeriSign calls a ‘peer-assisted’ CDN architecture.
Going forward, Limelight’s Gordon says, network operators must invest in accommodating a fundamental shift “to a Web that’s two-directional and conversational – it’s dialogue rather than monologue. It’s sharable, within whatever limitations content rights holders impose. It is on demand. It is deep, and it is searchable and/or discoverable.”
Web Publishing Infrastructure
While media company spending on Web distribution promises to bankroll continued investment in both backbone and CDN infrastructure, both media and network owners additionally are investing in Web publishing management technologies and infrastructure that also must be advanced to support the multimedia exaflood.
In the past two years, firms including The New York Times Company, Hearst Corporation and IAC/InterActiveCorp. have invested more than $60 million in video publisher Brightcove, and Comcast has invested more the $80 million to acquire and further develop Web video publisher thePlatform. Both claim client stables that include a who’s who among studio and media companies, as well as a growing stable of Web-age content creators.
These and other Web video publishing companies also are investing in integration projects – both internally and with partners – to create turnkey publishing platforms that can offer a full range of services in an ASP (application service provider) model that increasingly requires flexible bundling of services from multiple providers. Video ASP Origin Digital, for example, hosts and manages a range of services for content owners, including video transcoding, formatting, asset management and distribution to Web portals and devices.
Beyond such foundation Web video publishing services, thePlatform CEO Ian Blaine says his company is now focused on implementing functionalities at the management core of the IP content ecosystem, all within a context of open integration with partner applications that can be offered to content owners in bundled or optional add-on forms. thePlatform’s Sandbox program provides start-ups a real-world operations environment where they can test, demonstrate and refine products as they bring them to market.
“Because we manage many-to-many relationships between content owners and distributors we can drop new technologies into our system to see how they work,” Blaine says. “We’re not making cash bets like a VC, but we’re helping these companies manage their opportunity costs.”
For example, the Sandbox process has led to thePlatform integrating adapt.TV, a service that enhances the relevance of an advertisement to a specific situation using speech-to-text correlation between ads and media content audio. adapt.TV was able to use thePlatform’s player development kit to create client software that content owners and distributors on thePlatform’s service can readily access for engagement with adapt.TV’s service. Any content owner using thePlatform services to publish can turn on adap.tv by selecting the option.
The same model applies to another Sandbox partner, Move Networks, which provides end-device-specific content quality enhancement through efficient bit delivery – a capability that may become an element of a new category of CDN-optimization tools that thePlatform intends to offer, enabling content owners to wrest QoE controls back from the hands of CDN operators.
The key to husbanding and monetizing the explosive marriage between entertainment and social networking-based self expression, Blaine says, lies in giving customers the power to dynamically manage all the underlying components for delivering personalized content and advertising to any device anywhere from a single “holistic dashboard.” He notes there is tight coordination between thePlatform and Comcast teams on many fronts, including development of CableLabs-compliant XML applications that would facilitate publishing into existing VOD systems.
“Content from various Web sources would come into us, and we’d spit it out in the right format to the cable operator’s VOD servers,” he says. Indeed, the marriage of thePlatform’s publishing and CDN management suite with Comcast’s growing national backbone and metro transport infrastructure could effectively turn Comcast into a next-gen CDN operator competitive with the likes of Limelight, Akamai and Level 3.
If the industry is to respond effectively to what now appears an inevitable exaflood of multimedia traffic over the Internet – as well as to increasingly complex integration of layered services required to monetize that traffic – then continued investment in capacity, application-intelligent networking and advanced CDN and publishing management technologies to meet not only the scope, but also the complexity, of the exaflood now appears all but inevitable.
Market Data 
Markets 
AI
