Tom Leighton
Abstract
Given the Internet's bottlenecks, how can we build fast, scalable, content-delivery systems?
- Afergan, M., Wein, J., LaMeyer, A. Experience with some principles for building an Internet-scale reliable system. In Proceedings of the 2nd Conference on Real, Large Distributed Systems 2. (These principles are laid out in more detail in this 2005 research paper.) Google Scholar
Digital Library
- Akamai Report: The State of the Internet, 2nd quarter, 2008; http://www.akamai.com/stateoftheinternet/. (These and other recent Internet reliability events are discussed in Akamai's quarterly report.)Google Scholar
- Anderson, N. Comcast at CES: 100 Mbps connections coming this year. Ars technica (Jan. 8, 2008); http://arstechnica.com/news.ars/post/20080108-comcast-100mbps-connections-coming-this-year.html.Google Scholar
- Horrigan, J.B. Home Broadband Adoption 2008. Pew Internet and American Life Project; http://www.pewinternet.org/pdfs/PIP_Broadband_2008.pdf.Google Scholar
- Internet World Statistics. Broadband Internet Statistics: Top World Countries with Highest Internet Broadband Subscribers in 2007; http://www.internetworldstats.com/dsl.htm.Google Scholar
- Mehta, S. Verizon's big bet on fiber optics. Fortune (Feb. 22, 2007); http://money.cnn.com/magazines/fortune/fortune_archive/2007/03/05/8401289/.Google Scholar
- Spangler T. AT&T: U-verse TV spending to increase. Multichannel News (May 8, 2007); http://www.multichannel.com/article/CA6440129.html.Google Scholar
- TeleGeography. Cable cuts disrupt Internet in Middle East and India. CommsUpdate (Jan. 31, 2008); http://www.telegeography.com/cu/article.php?article_id=21528.Google Scholar
Index Terms
- Improving performance on the internet
Recommendations
Internet of Things security
The Internet of things (IoT) has recently become an important research topic because it integrates various sensors and objects to communicate directly with one another without human intervention. The requirements for the large-scale deployment of the ...
The relationship between Internet identification, Internet anxiety and Internet use
This paper reports a study investigating the relationship between Internet identification, Internet anxiety and Internet use. The participants were 446 students (319 females and 127 males) from two universities in the UK and one university in Australia. ...
Reviews
Christian Esteve Rothenberg
As a cofounder of Akamai Technologies, Leighton is in a unique position to unveil current best practices on high-performance content delivery over the Internet. This article takes the reader on a walk through the state of the art in network technologies for content delivery. Starting from the key observation that the Internet's bottleneck is moving from the last mile to the middle mile (or the metropolitan area network (MAN) aggregation network), back-of-the-envelope calculations on bandwidth-intense applications, such as video streaming and peer-to-peer (P2P), suggest the growing importance of the content delivery network (CDN), an overlay solution for efficient data dissemination. In addition to congestion in the aggregation network due to the current content-centric use of the Web, the actual distance to the server from which the large media sources are fetched appears to be another bottleneck that is poorly handled by the transmission control protocol (TCP) and the bandwidth delay product (BDP). Despite advances in the quality of the single network link technologies, the probability of packet loss increases, due to longer distances between servers and end users; this, in turn, causes TCP to slow down. So, let's move popular content closer to the requesters. In order to guarantee the performance, reliability, and scalability of commercial-grade Internet applications, it seems that content delivery architectures follow four gradually distributed approaches. At one end, we have traditional centralized hosting that suffers from the above-mentioned bottlenecks plus flash crowds. At the other end, there are P2P approaches that involve final user hosts limited by their asymmetrical uplink capacity. Commercial CDNs offer their services either in a "big data center" fashion or highly distributed among multiple Internet service providers (ISPs). The latter appears to be the most flexible and best performing approach. In the end, what matters is the user's experience. To overcome the impasse of providing network-level quality of service (QoS) among multiple administrative domains and transport technologies, the common approach of CDNs is to employ application-level techniques to provide quality of experience (QoE); referred to as application accelerations, optimization techniques to accelerate both cacheable and noncacheable (dynamic, personalized, and application-driven) content include finding better routers, prefetching embedded content, and neat computation and encoding techniques at the network edges. The article ends with insights into Akamai's design methodology, which is based on taking into consideration multiple sources of failure when designing, rather than as an afterthought. The design principles ensure quick recovery times and failover, by using significant redundancy mechanisms and both reactive and proactive highly distributed control mechanisms. "Besides the inherent fault-tolerance benefits, a system designed around these principles offers numerous other benefits," such as faster software rollouts, reduction of operational expenses, and easier scalability. These issues will grow in importance as a means to fulfill the increasing demands of the future Internet. Aren't CDNs somehow breaking the end-to-end principle by leveraging the network with additional intelligence and functionality__?__ To what extent should the community rethink future internetworking research__?__ While ad hoc and overlay solutions such as CDNs are clearly an excellent short-term solution, one should consider whether the host-centric TCP/IP paradigm could be replaced by a content-centric or information-oriented paradigm, where efficient interconnection of information objects is the main architectural goal [1]. Online Computing Reviews Service
Access critical reviews of Computing literature here
Become a reviewer for Computing Reviews.
Comments