Akamai

Last June, Netflix announced plans to build out their own content delivery network by giving ISPs free caches to place inside their networks. Called Open Connect, Netflix’s new platform allows network operators to provide higher quality streaming and more importantly, gives them control over the video that flows through their pipes. This morning, Netflix announced which ISPs have joined the program which include Cablevision (U.S.), Virgin Media (UK), British Telecom (UK), Telmex (Latin America), Telus (Canada), TDC (Denmark) and GVT (Latin America).

Netflix says that Open Connect is now serving the “vast majority” of Netflix video in Europe, Canada and Latin America. Netflix wouldn’t clarify exactly how much of their traffic that is, but did say back in June that more than 50% of their traffic in the UK alone was coming from their new CDN platform, so clearly it’s grown quite a bit since then. Netflix said their goal is to have “all of our members served by Open Connect as soon as possible” and while they haven’t given out a time frame just yet, realistically it would probably take them 24-36 months to have nearly all of their International and U.S. video streams being delivered from inside ISP networks.

In addition, Netflix says that their Open Connect partners now have the ability to offer a limited number of videos in Super HD and 3D streaming, with 3D streaming being limited to North America only. I tried to see if I could test one of the videos, but my ISP Verizon isn’t in Netflix’s Open Connect program. Netflix has launched a new page on their website where consumers can check to see if their ISP is in the Open Connect program by simply going to www.netflix.com/superhd.

Netflix told me their Super HD videos are encoded for 7 Mbps and that the 3D streaming videos require 12 Mbps at the high end. Currently, only the PS3, WiiU, Windows 8 devices, Roku, Apple TVs (1080p model) and select smart TVs and Blu-Ray players are supported.

It’s also interesting to think about how Netflix’s Open Connect program could help protect their business. For some time now, service providers have been feeling the pressure by having to backhaul a lot of Netflix’s traffic, at their own cost. They only thing they can do to combat it is spend money to built-out, which means they put that pressure back on the consumer by raising rates or implementing caps. Over time, it will be very interesting to see if any of the ISPs that have Netflix caches inside their network allow content from Netflix not to count towards consumers caps. I don’t know of any ISPs currently thinking of doing that, but it’s something to keep an eye on.

Very_few content owners can build out their own CDN, but for Netflix it makes perfect sense. They have enough traffic to make it more cost-effective than using third-party CDNs and even more importantly, it allows them to provide an even better user experience. Based on recent data Netflix has given out, their average broadband stream is delivered at just over 2 Mbps and for mobile devices, those streams average 600 Kbps. So over time, their Open Connect initiative will allow customers to be able to get much better quality video, something every content owner is always striving to improve on.

If you’re interested in hearing more about Netflix’s CDN plans then save the date to join us on Monday May 20th at my Content Delivery Summit in NYC, where Ken Florance, VP of Content Delivery for Netflix will be our opening keynote speaker.

Read more:

– Netflix’s Streaming Cost Per Movie Drops 50% From 2009, Expected To Spend $50M In 2012

– Netflix’s CDN News Being Overblown By Many Wall Street Analysts, Focus On The Facts

– Average Family Of Four Probably Has 10 Netflix Enabled Devices In Their Home Today

– Netflix Announces New Content Delivery Network, Offering Free Caches To ISPs

The media is spending a lot of time regurgitating the Akamai and AT&T press release but from what I’ve seen, none are digging under the surface to see what’s really going on. Some like the WSJ even said that as a result of the new deal, AT&T and Akamai will now “end competition” amongst each other, which is laughable. Anyone who follows the CDN space, which clearly the WSJ doesn’t, knows that AT&T was never competing with Akamai, or anyone else for that matter when it came to their failed CDN business. The simple fact that the WSJ and other sites don’t even mention what AT&T or Akamai’s CDN revenues are shows they really don’t get the market. Others have made comments like, “the deal eliminates AT&T as a rival” which is also not true. Based on numbers Akamai has given out in the past, and the AT&T numbers I know of, AT&T is doing around 1% of Akamai’s CDN revenue. How is that a “rival”?

Reuters said that the AT&T deal, “follows Akamai’s similar agreement with France Telecom’s Orange last month”, which is not accurate. Akamai’s deal with AT&T is not a licensed CDN deal. It’s a straight reseller deal. Akamai’s announcement with Orange is for a licensed CDN deal where Orange licenses Akamai’s software to run on the Orange network. AT&T is not deploying any software from Akamai on their network, so the two deals are not at all alike, something that Akamai confirmed with me on a call with them this morning.

I’ve also seen a couple of telco bloggers say this deal isn’t good for EdgeCast, which isn’t the case at all. While none would go on the record for this piece regarding numbers, I talk often with executives at all three companies and one critical fact that’s missing from every story so far is that Akamai committed $100M to AT&T to secure a resale deal. That might seem smart on the surface but if you know the space well, you know that it’s just not a scalable strategy. It doesn’t surprise me that this deal would be applauded by shareholders but most don’t have any insight into AT&T’s CDN business and to date, I have never seen a single report that even says what AT&T’s CDN revenue is, aside from the number they used from my blog, which was $10M last year and less than $20M this year.

While EdgeCast is nowhere near the size of Akamai, I expect they will do about $75M in 2012, EdgeCast is years ahead of Akamai in the carrier space and have vastly more white-label resellers overall. By my last count I think it is more than 60, including telcos like DT and large web hosts such as Softlayer. EdgeCast earned millions from AT&T in licensing fees (my sources estimate the amount at $25-50 million) and they will continue to do so for some time. Meanwhile, Akamai is spending $100M partly to try to stop their momentum and also get access to AT&T’s network. Sure, that’s a much cheaper way to buy into the telco space than acquiring EdgeCast would be, but it’s a staggering price to pay to pick up a reseller. Committing eight and nine figure sums to resellers is simply not a repeatable model, especially since EdgeCast already has deep relationships with many of the largest resellers.

Yet many are suggesting that what Akamai is doing with AT&T is a blueprint for what they will do with other carriers when it’s not. Akamai only has a few hundred million in cash, so they definitely can’t afford to buy their way into all these resellers and operators, especially for a service like CDN, which has low margins. And most telcos and carriers have been actively building out their own CDNs, or using licensed and managed CDN products, which is a business Akamai has only just entered, years behind others. Most telcos, carriers and MSO don’t want to simply resell services, but rather have more control over them since they own the network.

Some have suggested to me that once Akamai’s LCDN product is out of beta that AT&T could buy out their EdgeCast contract and replace EdgeCast’s LCDN solution with Akamai’s, but that’s not likely. I’ve also heard some say that Akamai bought Verivue to try to get an LCDN product to market faster so they could try and get it into AT&T and push out EdgeCast, but that’s not happening. Remember that Juniper spent $100M buying Ankeena just to please AT&T and then AT&T didn’t end up going with them. Akamai’s deal with AT&T is not around LCDN and won’t be any time soon.

So, what will be left when the dust settles? There will be AT&T’s wholesale CDN platform, with hundreds of servers running EdgeCast software and delivering paid customer traffic for massive brands that AT&T services today on their CDN. None of these big brand customers that I speak with are interested in having their production traffic moved off a stable platform and onto a new one. So AT&T is at risk of churning the CDN business they do have if they try to force their customers onto a third-party platform they don’t operate.

As for the growing federated traffic being sent to AT&T by EdgeCast and Pacnet, clearly that cannot migrate to Akamai’s platform under the newly announced deal since this is not a licensed CDN (LCDN) agreement with AT&T and right now, Akamai has no LCDN product outside of what’s currently in beta. So AT&T can either keep their wholesale platform up, or go back on their commitments to the global carrier community (via OCX, etc.) to promote CDN Federation.

There will also be the interesting drama of AT&T trying to figure out how to resell Akamai services to AT&T enterprise customers, a large number of which are already Akamai customers. One final fact is that the CDN that AT&T launched (at the cost of well over $100M) before they signed with EdgeCast is still running. Now they are going to add an Akamai CDN to this mix, which means running three separate CDNs until they shut down their own and migrate over to Akamai. That’s a lot to ask for a company that has never shown any expertise with their CDN business or strategy.

As I wrote last year, the way AT&T could get serious about CDN would be to acquire a CDN operator, but they don’t seem interested in doing this. One could suggest they don’t want to acquire low-margin CDN products, but they also had the right of first refusal to acquire Cotendo, who had high-margin value add services, and they passed on that before Akamai acquired them, even though AT&T told me that Cotendo’s services were very important to the company with, “40-60% of all new customers in their sales pipeline” needing such a solution. So AT&T is not really serious about this business if all they want to do is resell it. Some would suggest the opposite by saying that AT&T can now rely on Akamai to help them do it right, but if AT&T could not sell a simple CDN service that they owned, how are they going to resell CDN services they no longer control? It’s asking a lot from AT&T.

I’ve also seem some that suggest the CDN business with AT&T isn’t really important as it’s all the “value added services” that AT&T will resell that will bring Akamai revenue. While that’s a nice idea, it’s not the majority of what AT&T will be reselling. Akamai confirmed with me today that AT&T will sell CDN including their media services, software downloads, small object delivery and DSA. But that’s a far cry from the complex, customized, high-margin services Akamai offers that require a lot of professional services. That’s not something AT&T will be able to sell at scale.

The real benefit to Akamai is not what AT&T might be able to resell, it’s the fact that Akamai now gets to place their servers inside AT&T at the regional level. Akamai confirmed for me that before this deal, they didn’t have any of their servers directly within AT&T’s network and had to peer with them instead. So from a QoS and capacity standpoint, this deal is good news for Akamai and AT&T customers but Akamai isn’t saying how much additional capacity it gives them or the difference in QoS. But at the cost of $100M, Akamai paid a lot of money to get inside AT&T’s network, with the hope that AT&T can also bring them a lot of reseller revenue.

Another thing to watch from this deal is Akamai’s costs. While Akamai talks a lot about how many servers they have and how they are at the “edge”, which is a vague term, the deal with AT&T shows that Akamai never actually had servers within AT&T’s network. Now they are spending a lot of money to get that access and this might open them up to other telcos who now see that Akamai is willing to write big checks to get inside telco networks. It will be interesting to see if other telcos now ask Akamai for a large sum to have access inside their network as well. If that happens, and Akamai needs to have more than just peering access, then this could get expensive for Akamai in the long-term. While I don’t know if this will happen, it’s reasonable to think it could, considering Akamai’s deal with AT&T is public for all the other telcos to see.

AT&T has simply never been able to get their CDN act together and now they are going to try another approach, by reselling Akamai, and change their strategy once again. Time will tell, but in two years we could be looking back on this recent agreement between Akamai and AT&T and realize that Akamai picked up the tab for the most expensive deployment inside a carrier in history.

[See my follow up post here: Inside The Akamai and AT&T Deal and Why Akamai May Have Paid Too Much] In August I reported that AT&T was planning to shut down their in-house CDN and re-sell CDN services from Akamai or Limelight Networks and that Akamai would probably win the deal since they were willing to guarantee AT&T more business than Limelight would. This morning, AT&T and Akamai made the deal official with an announcement saying the two companies will work to jointly sell CDN services in North America to start, expanding to outside the U.S. in twelve months. For AT&T, this signals what is almost a thirteen year effort to try to get their CDN business off the ground, dating back to 2000 when they launched their ICDS platform (Intelligent Content Distribution Service).

While today’s announcement is good for Akamai, there’s not a lot of revenue attached to it. AT&T will do less than $20M in total CDN revenue this year and it will take them and Akamai a long time to sell a joint solution in the market, let alone one that can handle content delivery outside the U.S. I don’t expect today’s announcement to affect AT&T’s wholesale CDN services and federation model, so I would expect AT&T would still manage that portion of their CDN business with EdgeCast’s platform. Customers who are currently buying this solution from AT&T purchase it from a wholesale division of the company, not from an enterprise sales team, so the new re-seller deal with Akamai should not impact AT&T’s wholesale CDN business, which continues to grow.

While enterprise customers could also go direct to Akamai, most of AT&T’s large enterprise contracts are for multiple products, including things like co-location, transit and managed services, which are services Akamai does not offer. So AT&T isn’t trying to get CDN only business with a re-seller deal like this, but rather want to use CDN to keep or get them more of the non-CDN business they already have. It will be interesting to watch how both companies manage channel conflicts, since a very large percentage of enterprise customers are already taking services from Akamai, but one would assume that’s something they have already worked out with this deal.

At the time of this post, Akamai’s shares are up $3.20, which makes no sense since the revenue associated with this deal is so small to start. I’ve seen more than a dozen articles taking about the deal and about Akamai’s shares being up, but none of them discuss what the value of this contract could be worth to Akamai, or what AT&T’s current CDN revenue is.

This morning Akamai announced they have reached an agreement to acquire privately held Verivue Networks, which providers a transparent caching and licensed CDN platform to service providers. Akamai expects the deal to close by the end of this year and said they will pay all cash for the company, but has not disclosed any other terms of the deal. Verivue doesn’t have a lot of paying customers and has a small number of customer deployments with 2012 revenue in the $20M-$25M range, so I don’t expect Akamai to be paying much for the company.

While many are calling this a licensed CDN acquisition by Akamai, it’s also a transparent caching play. Verivue has two technologies that will benefit Akamai, a licensed CDN platform, and transparent caching technology, which is where Verivue gets a lot of their revenue from. Over the past year, many vendors have been working to combine the two technologies into one platform for service providers. PeerApp, the leader in the market has teamed up with EdgeCast, Juniper teamed up with Tata and Cisco is also working with PeerApp. For a better understanding of how transparent caching fits into the CDN market, read my previous post entitled “An Overview Of Transparent Caching and Its Role In The CDN Market“.

For those not familiar with the term “transparent caching” the idea is simple. Transparent caching platforms make intelligent decisions about which content can and should be cached inside a carrier’s network. By deploying intelligent caches strategically throughout their networks, operators can cache and deliver popular content close to subscribers and reduce the amount of transit traffic across their networks. CDNs like Akamai are good at what they do, serving content to subscribers quickly, and in the process alleviating peering costs for service providers. But there’s an enormous amount of traffic that CDNs do not serve, and that’s where the value of intelligent caching comes in.

Deployed throughout a carrier network, transparent caching improves subscribers’ experience and reduces carriers’ peering costs, but only if it delivers the features and intelligence required to adapt to constantly changing user behavior and content patterns, and most importantly, scales economically to tens or hundreds of gigabits per second. Akamai’s plan is to take the best of both technologies, licensed CDN and transparent caching and combine them into one platform, sold to service providers.

When Akamai announced their licensed CDN product in February, the company didn’t mention anything in their release about it having any transparent caching functionality. But in a briefing I had with the company when they announced it, Akamai divulged that their licensed CDN product would have transparent caching functionality build in. That was an interesting piece of news because Akamai’s platform would then offer all three services to an operator, those being on-net services/software, off-net delivery and transparent caching. At the time, Akamai said that 100% of the technology for their managed and licensed CDN product had been developed in-house and that they did not plan to licensed any transparent caching technology from any third-party. Clearly the company has changed their mind and acquiring Verivue will help them get to the market faster.

Since Akamai’s announcement of their licensed CDN product nine months ago, the company has yet to mention a single customer by name that is using their platform. This morning they mentioned to me that they have “three customers in trial” for their licensed CDN platform, which is still in beta but that they have “some paying customers today” for their managed CDN product. While we don’t yet know how much money Akamai is paying for Verivue, clearly they aren’t making too big of a bet on the licensed CDN business, otherwise they would have acquired EdgeCast, the leader in the licensed CDN market. But with expected revenues of around $70M this year for EdgeCast, coming from both regular CDN services and their licensed CDN platform, realistically, Akamai would have had to spend a few hundred million for EdgeCast, something I doubt they are paying anything near for Verivue.

As a stand alone business, the transparent caching market is expected to be $142M this year and the licensed CDN market is even smaller. So there is not a lot of revenue here – yet. Also, the market is crowded with transparent caching vendors including PeerApp, Qwilt, Juniper, Huawei, Fortinet, Conversant, BTI Systems, Brocade, Blue Coat, VidScale and Allot Communications who just acquired Oversi. Add to that all of the licensed and manged CDN vendors in the market like EdgeCast, Limelight Networks, Jet-Stream, Cisco, Alcatel-Lucent, XDN and others and it’s a very crowded market. But many of the players in the space are small, some are having trouble growing revenue and it’s only a matter of time before the market consolidates even more. No one truly knows what the size of the licensed CDN market is today, or what it can grow to because it’s only just getting started and we need to see a lot more customer deployments before we have enough data in the market to see how service providers use these platforms. But there is a real opportunity with the right kind of licensed CDN platform and it’s a market that is going to grow nicely over time. Acquiring Verivue should allow Akama to help get deployments in the market much faster.

Apple’s live webcast of the launch of their new iPad mini was a failure today after multiple users, including myself, had problems getting the stream to start or staying connected to the stream once it began. I tried the stream in the Safari browser at 1pm ET and got the spinning wheel with the player trying to load, but it took till 1:14pm for the stream to work. Once it did load, it worked for a few minutes before I lost all audio. When the audio came back, the video looked bad with lots of pixelation and twice the video re-wound and went back to a point in the stream it had already played. At 1:26pm, the stream died for me completely and I could not get it back.

Akamai was delivering the live stream for Apple and clearly had problems. While I hear from customers all the time that Akamai’s HLS delivery is often not reliable, I’ve now experienced it for myself. While I only tested it on Safari, other viewers I was live chatting with during the event also experienced problems on the iPhones and Apple TV. Looking at Akamai’s chart at 1:43pm ET that shows the number of real-time connections to their network for live and on-demand videos showed 943,000 concurrent live video streams, for all of their customers combined. And their 24-hour peak was 1.3M. So either Akamai was not counting Apple’s live stream numbers in their chart, or it shows just how few people were able to get the live stream as Apple’s webcast alone should be in the multiple millions of concurrent connections.

As an industry, we’ve been streaming live events since 1996. This technology has been around for 16 years now and there is no excuse whatsoever for a live webcast not to work. Yet this is the same technology that Akamai and others keep talking about that is supposed to rival broadcast TV in terms of quality and reach? I don’t think so.

The term “value add services” is used a lot by content delivery networks to describe services that don’t fall under the typical commodity CDN business. It’s hard to define exactly what those services are as most CDN vendors don’t define it down to a product level. I get many questions asking what services are classified as value add, but the answer all depends on who you ask. The way I define it for people is that CDN services typically include four kinds of content delivery. Those are – audio/video streaming, audio/video downloads, software downloads and small object delivery.

Services outside of those would be considered value add and would include things like, application acceleration, dynamic site acceleration, front-end optimization, mobile content acceleration, media management (transcoding, ad insertion, content protection) and a host of other services for the purpose of security and commerce. That said, many of those value add services also utilize the CDN’s network to deliver this content, so not everything that falls under CDN can be that easily quantified as value add versus non-value add.

Each CDN vendor looks at and defines their services differently and many confuse the industry even more when they break out their revenue under the generic term of “value add services”, but then don’t define what those services are. Akamai in particular confuses the market because they break out part of their revenue based on services (products), while the other half is broken out by media and entertainment, which is not a product or service, but rather a vertical. All of this aside, here are what I consider to be the most common services that would fall under the term “value add services”, the way I define it, and definitions on what those services are.

• Dynamic Site Acceleration (DSA): Dynamic site acceleration is a suite of technologies and products that deals with optimizing dynamically served content across the network. Traditional DSA services often include TCP optimization, route optimization, connection management, on-the-fly compression, SSL offload and pre-fetching technologies.

• Front-End Optimization (FEO): Front-end optimization technologies help to reduce the number of page resources required to download a given page and makes the browser process the page faster. FEO technology isn’t used to bring content closer, but rather makes the content itself faster by optimizing the client side delivery of website resources.

• Transparent Caching: Transparent caching platforms make intelligent decisions about which content can and should be cached inside a carrier’s network. By deploying intelligent caches strategically throughout their networks, operators can cache and deliver popular content close to subscribers and reduce the amount of transit traffic across their networks.

• Licensed/Managed CDN: Licensed and managed CDN refers to software and services aimed at helping telcos, carriers and service providers build and deploy their own CDN services inside their network. Licensed CDN refers to the licensing of CDN software to the carrier who then builds a CDN solution on their own. Managed CDN is when a service based content delivery vendor helps build and manage the CDN component of the carrier’s network for them.

• Application Acceleration: Application acceleration is a suite of technologies that combines fast packet processing with SSL acceleration, connection multiplexing, dynamic caching and adaptive compression to improve application response times. These technologies enable enterprise customers to accelerate the delivery of internal, external and latency sensitive applications to distributed users across the Internet or via their enterprise network.

• Mobile Content Acceleration: Mobile content acceleration technologies are designed to specifically eliminate latencies found on mobile broadband networks to reduce page load times on mobile devices.

In addition to these listed, you also have a whole host of different services to handle things like DoS attacks (security), custom and managed DNS, custom reporting, tracking and analytics (especially for ad delivery) and lots of pieces in the video ecosystem for managing and adding business rules around the monetization of video. This is how I would classify value add services, but others may define it differently. How you would define it?

At the end of last week, and over the weekend, multiple people from the industry were sharing with me what they know about the negotiations between Akamai and Limelight, who are both competing for a CDN reseller contract with AT&T. Everyone I spoke with said they expect Akamai to win the contract and all of them also said that as part of their proposal, Akamai is guaranteeing AT&T at least $100M in revenue, over a multi-year deal. Some people told me the deal size was $100M and others said it was “more than” $100M, but either way, it sounds like Akamai has put forth the best revenue numbers.

While $100M may sound like a lot of sales, when it is spaced out over a couple of years, it’s really not that much revenue for AT&T. But I’m also hearing that Limelight was only willing to guarantee half as much revenue to AT&T, between $40-$50M, which if true, it’s probably one of the reasons why everyone I speak to keeps saying Akamai will win the deal. Of course, until a contract is signed, none of this is official and we still need to see AT&T execute on this new strategy. Companies put a lot of ideas and plans on paper, without following through with them, but from everything I am hearing, it sounds as if AT&T is looking to wrap this up pretty soon and sign a contract with Akamai or Limelight.