using the Dimes Project data sets that describes the structure of the Internet, Chris harrison of Carnegie mellon university created this visualization illustrating how cities across the globe are interconnected (by router configuration and not physical backbone). In total, there are 89,344 connections.

 

mance in video distribution, it is natural to consider a P2P (peer-to-peer) architecture. P2P can be thought of as taking the distributed architecture to its logical extreme, theoretically providing nearly infinite scalability. Moreover, P2P offers attractive economics under current network pricing structures.

In reality, however, P2P faces some serious limitations, most notably because the total download capacity of a P2P network is throttled by its total uplink capacity. Unfortunately, for consumer broadband connections, uplink speeds tend to be much lower than downlink speeds: Comcast’s standard high-speed Internet package, for example, offers 6Mbps for download but only 384Kbps for upload ( one-six-teenth of download throughput).

This means that in situations such as live streaming where the number of uploaders (peers sharing content) is limited by the number of downloaders (peers requesting content), average download throughput is equivalent to the average uplink throughput and thus cannot support even mediocre

Web-quality streams. Similarly, P2P fails in “flash crowd” scenarios where there is a sudden, sharp increase in demand, and the number of downloaders greatly outstrips the capacity of uploaders in the network.

Somewhat better results can be achieved with a hybrid approach, leveraging P2P as an extension of a distributed delivery network. In particular, P2P can help reduce overall distribution costs in certain situations. Because the capacity of the P2P network is limited, however, the architecture of the non-P2P portion of the network still governs overall performance and scalability.

Each of these four network architectures has its trade-offs, but ultimately, for delivering rich media to a global Web audience, a highly distributed architecture provides the only robust solution for delivering commercial-grade performance, reliability, and scale.

 

application acceleration Historically, content-delivery solutions have focused on the offloading and delivery of static content, and thus far we

have focused our conversation on the same. As Web sites become increasingly dynamic, personalized, and applica-tion-driven, however, the ability to accelerate uncacheable content becomes equally critical to delivering a strong end-user experience.

Ajax, Flash, and other RIA (rich Internet application) technologies work to enhance Web application responsiveness on the browser side, but ultimately, these types of applications all still require significant numbers of round-trips back to the origin server. This makes them highly susceptible to all the bottlenecks I’ve mentioned before: peering-point congestion, network latency, poor routing, and Internet outages.

Speeding up these round-trips is a complex problem, but many optimizations are made possible by using a highly distributed infrastructure.

Optimization 1: Reduce transport-layer overhead. Architected for reliability over efficiency, protocols such as TCP have substantial overhead. They require multiple round-trips (between the two communicating parties) to set

 

48 CommunICatIons of the aCm | feBRuaRY 2009 | vol. 52 | No. 2