Akamai and Limelight To Deploy P2P For Higher Quality, Not Cost Savings

While there's been a lot of hype about peer-assisted technology (P2P) over the past few years, in reality, the technology has gotten very little adoption when it comes to being used to distribute video, let alone streaming. Today, most of the content being delivered via P2P is for software and gaming downloads and so far we haven't seen any of the major CDNs offer a real peer-assisted solution on their network. Later this year, that is going to change.

Within the next two quarters, both Akamai and Limelight are going to bring to the market peer-assisted delivery services on their network. And unlike previous deals where some of the CDNs were simply reselling a third party P2P platform, these solutions are going to be their own, deeply deployed and integrated directly into their networks.

(Correction: In the case of Limelight, they will work with other existing P2P companies who want to use the Limelight Network for peer-assisted delivery.)

While some might think the reason Akamai and Limelight need peer-assisted solutions is to reduce their costs, or offer a cheaper level of delivery in the market, they would be wrong. This is not going to be about reducing a customer's bandwidth bill as many P2P providers pitch in the market today, but rather the ability for Akamai and Limelight to guarantee a certain level of quality. For all the talk in the market about HD video and the word "quality", the fact of the matter is that the CDNs don't control the last mile and have no way to ensure truly HD quality video.

In order to truly control quality to the end user you need a client and you need a control mechanism tied to the client. While many folks have client base solutions on the market, most of them don't have any central control mechanism and simply reply on the client itself to determine what's taking place on the network. The problem with that solution is that the P2P client cannot be self aware enough to know what really taking place on the network and can't adjust accordingly. You can't have every single P2P client on the network trying to act as the general. This is the primary reason why to date, none of the major CDNs have worked with any of the stand-alone P2P companies in the market and in the case of Akamai and Limelight, have decided to develop their own technology in-house.

While I can't give out too many details yet on the service from either company, I can say that Akamai's solution will be based on the Red Swoosh technology they acquired in 2007. While the technology didn't support streaming at the time of the acquisition, Akamai has been working for the past two years to further develop the Red Swoosh platform and continues to build out additional functionality. I'll give out more details on the solutions themselves when I'm allowed to.

Some might ask, why come out with a peer-assisted offering now, what changed in the market? It's a fair question and one that is important to answer. These new solutions won't be for every customer and for every type of content delivered over the CDNs. Large scale live events and other use case scenarios where large traffic spikes occur or the need for a guaranteed level of quality are the best fit for peer-assisted delivery. While we keep talking about the bitrate when it comes to HD video, delivering a quality user experience in HD has to be about more than just the bitrate the video is encoded for. That's really where peer-assisted technology comes in.

While some might speculate that the 2010 Winter Olympics would be the ideal event for Akamai to use a peer-assisted based solution for delivery of content, Akamai said no P2P based solution will be deployed for the event next year.

Even though P2P technology has been around for years now, the problem is that most P2P companies pitch it as a replacement for traditional CDN services, which is wrong. It's not a replacement for CDN, it's a compliment. There are instances where both sets of technologies have their own strengths and weaknesses and the value of a CDN is being able to offer different levels of delivery services based on the type of content being delivered, the type of device it's being played back on and the size and scale of the traffic.

For CDNs, having a peer-assisted solution in their bag of tools is a natural next step but it is really important for people to realize that the CDNs are not being driven to offer this service to simply drive down the cost of delivering bits. This is all about guaranteeing a different level of quality that today, the CDNs don't have the capability of doing since they don't control the last mile.

Related Posts:

- Aug 08': The Current State of the Content Delivery Market

- May 08': P2P Vendors Struggling, CDNs Not Interested In Adopting

- April 08': Do P2P Networks Really Support Streaming?

- May 07': Confusion Reigns Supreme When It Comes To P2P Delivery Networks

- April 07': Post Updated: Akamai Buys Red Swoosh - Making P2P Play?

- April 07': P2P Delivery Networks Can't Survive On Their Own

- Feb 07': Does p2p Really Have Any Traction In The States?

Comments

[Matt]

Red Swoosh did have streaming and showed it off with their abandoned Bittorrent client Foxtorrent .

https://addons.mozilla.org/en-US/firefox/addon/4844

[Mayeul]

I don't if it's intentional, but you don't mention Velocix (ex Cachelogic) who is a major UK-based CDN, and they've been running a hybrid P2P delivery platform for years now.

[Dan Rayburn]

Hi Mayeul, the story is about the major CDNs who traditionally have had no P2P offering adopting P2P. Velocix is not a major CDN and their network was built from P2P technology from day one.

[Dan Rayburn]

Hi Matt, that client you reference only allowed you to watch the video and "stream" it back while it was downloading. Not really streaming in the true sense of the word.

[Anonymous]

Dan,

Where do you think this leaves the major CDN's that don't have this functionality?

It seems like LL and Akamai are pulling away.

[Anonymous]

CDNetworks announced a deal with Octoshape last year. I don't know if it's still in effect, but their website does mention that P2P is available for both on-demand and live streaming.

If Akaimai and/or Limelight can find a way to measure and guarantee the quality of delivery through P2P, that would be very interesting.

[Jonathan]

While no where near the size of the above stated companies, we have been sucessfully using P2P (yes Dan, for live streaming) for some of our clients through a realtionship with a live P2P company based here in the states- NiFTy TV. There is demand, and there are success stories out there. Keep an eye on P2P -live and ondemand - in STB.

http://www.streamguys.com/streamingservices/distributedbroadcasting.html

[Steve]

Here's what I think is really going on:

Akamai found out that LL was going to announce P2P and decided to "leak" their own solution - but their solution isn't ready. There's all the reason in the world for Akamai to deploy P2P for the Olympics, one of the largest live events in the world, and the only reason they wouldn't use it for that event is that it isn't yet ready for Prime Time.

Makes you wonder what they really have if anything.

[Dan Rayburn]

@ Anonymous - Many CDNs have announced "partnerships" with P2P companies, including CDNetworks, Internap and others. But so far, those partnerships have amounted to no substantial client wins or revenue and to date, are more "reseller" agreements. The CDNs love those deals as the P2P vendor agrees to use their CDN for traffic, so the CDN is more than happy to take their money each month. But I would not consider that to be a real P2P offering by a major CDN.

@ Anonymous - Until CDNs start offering peer-assisted solutions in the market and we see they are successful with these offerings, we can't say other competitors are behind them in the offering. That said, Akamai and Limelight, and in particular Akamai, have such a deep pool of development resources in-house that they will always be ahead of the pack. They can afford to work on many products all at the same time, even if there is no revenue for them yet in the market.

[Kevin]

Dan, Limelight on their twitter said...

"Before rumors get started, this post is inaccurate: http://tinyurl.com/kpntzr We have no plans to bring a P2P product to market. @llnw"

the post they are linking to is yours.

Any comments?

[Dan Rayburn]

There was some confusion on how the service was described to me, hence why I added the clarification to the post:

(Correction: In the case of Limelight, they will work with other existing P2P companies who want to use the Limelight Network for peer-assisted delivery.)

In other words, Limelight won't be rolling our their own P2P service but rather working with others.

[Steve Lerner]

Maybe its just me- but I view video from a gazillion sites every day and never, ever, encounter a requirement to download a p2p app - Move Networks- yes, but P2P? Still haven't seen a single request...

[Russell]

Dan - although the larger CDNs say that they are not providing this service to lower bandwidth costs, the fact is that their costs will be lowered by using the p2p solution.

The very same reliability can be achieved by using a major CDN, such as LL and Akamai or Level 3, and simply signing up back-to-back with an independent p2p hybrid provider (for example piCast) directly and leveraging the p2p efficiencies from there. In fact, you will save on your delivery costs with LL, Akamai or Level 3 if you do it this way; so even if they do not want to share their saving with the user directly, there is a way to force their costs down and keep the saving yourself.

[someguy]

Doesnt the move to fragmented codecs over http - with the ability to leverage the cacheing that exists at the ISP level - basically make P2P redundant?