Posts for 'Akamai'

As mentioned in my earlier post, today I was at Akamai (disclaimer: Akamai is a VideoNuze sponsor) for their annual Analyst Day. I only stayed for the morning sessions, but, reflecting how important the Media and Entertainment (M&E) vertical has become for Akamai, they put a huge emphasis on M&E updates, so I think I was exposed to all their news. Following is an objective review of my takeaways:

As the largest CDN and established incumbent in broadband video delivery, Akamai's capabilities and plans are crucial to broadband video's continued growth. To be sure, the company's market dominance has been challenged by newer rivals (which continue to emerge seemingly daily) and talk of alternative approaches such as P2P continues to swirl.

As with its underlying web content delivery model, Akamai stakes its differentiation - and premium pricing - on the company's highly distributed content delivery architecture. The company now has 27,000 servers in 1,600 locations within 900 networks. As Tom Leighton,Chief Scientist and co-founder pointed out repeatedly in his presentation, this model contrasts with its competitors' model of co-locating their servers exclusively in big data centers (where Akamai has servers as well).

Akamai's distributed footprint addresses a key chokepoint in the Internet: the lagging performance the "middle mile", the vast expanse of routers that forms the guts of the Internet between the first mile data centers/Tier 1 interconnects and the last mile cable/DSL/Ethernet connections. By Akamai's calculations, the performance of the "middle mile" has grown by only 6x, while the first mile has improved by 20x and the last mile 50x (think how your cable model blows away those pokey old dialup modems).

In Tom's view, competitors' performance delivering video and other applications is handicapped because their servers don't fully overcome the middle mile's deficiencies. Akamai's goal is to be serving content from servers within 10 miles of the user. Tom's estimates are that Akamai can serve content 2-6x faster than competitors and with 2-8x more reliability. These figures are magnified when it comes to delivering high quality video files. In a test Akamai did for a major media company trying to serve 4GB video files to Japan, it took Akamai 12.2 minutes to deliver the file, while servers located 500-1,000 miles away took 2.2 hours, 3,000 miles away took 8.16 hours and 6,000 miles away 20.hours. This was of course an Akamai test, but, assuming the data holds up under closer analysis, this would be pretty compelling for media companies looking to optimize user experiences.

To debunk pundits who have suggested the Internet cannot scale to deliver the explosion of video, particuarly HD, Tom drew a hypothetical case of 50M users each streaming at 2 megabits per second resuting in total bandwidth needs of 100TB of capacity. Akamai's goal is to have just north of that deployed by 2010. Estimating usage is notoriously difficult so who knows whether having 100TB will be sufficient or not, but again by Tom's calculations, none of its competitors will be able to come close.

A recent challenge to Akamai's architecture approach has been P2P's potential. Akamai acquired Red Swoosh and outlined how it is integrating this technology. For Akamai, standalone P2P delivery isn't realistic, rather it requires both CDN-assist and also cooperation from willing customers and ISPs who could squelch its adoption. It sounds like the Red Swoosh technology is being slowly introduced to select customers and to test its acceptance. Of course Akamai would like to reduce costs for its customers, but the company made clear P2P is no panacea for content owners expecting to dramatically slash their delivery costs.

Yet another challenge has been the prospect of service providers and telcos building out their own private CDNs using gear from companies like Cisco. In an offline chat with Tom, he mentioned that this approach has been around for years, but hasn't seen significant adoption. So he expects some providers will continue to pursue it, but ultimately, Akamai's overall solution approach will always give it an edge over these homegrown alternatives.

It's clear that Akamai also sees delivering HD as a key driver of its future success. Customers are expecting ever-higher quality, and this plays into the company's positioning as a non-commoditized provider. So while Akamai acknowledges that price is a key factor in customers' decision-making, it's quite clear that the company is relying on data from its own testing and actual customers' performance to demonstrate why quality and reliability matter, and how these translate into greater revenues and user loyalty. Given the wealth of data they shared today, it is evident that Akamai's sales approach is pretty sophisticated.

Having developed and seen myriad business cases for small and large video providers, I know first-hand how significant the delivery cost line item is in these P&Ls and how tempting it is to zero in on low price as the key decision criterion. Expect Akamai to keep making the case that while important, it is far from the only variable.

Tomorrow I'll be at Akamai's annual analyst day (disclaimer: Akamai is a VideoNuze sponsor). The morning speaker line-up includes Paul Sagan, President and CEO, Tom Leighton, Chief Scientist and Co-Founder and Mike Afergan, CTO. I attended last year and found it to be an extremely informative day, especially since Akamai is the leading CDN and has been very focused on the media and entertainment space.

I'll be listening for information on 3 specific areas:

• Update on pricing pressure and what this means for customers?
• How Red Swoosh P2P integration is coming along and are any customers using it yet?
• Any insight on service providers' (cable operators and telcos) motivation to build out their own private CDNs with gear like Cisco's CDS?

I'll try to provide an update before hopping a plane to Dallas to speak about broadband video trends at a large broadcasters' executive offsite.