Judging from the buzz at this year’s CDN World Summit, CDN federations are a hot topic—and not just because they were the focus of my keynote. In short, the industry has moved beyond “if” and is now talking about “when” and “how.” This is good news because I believe CDN federations will play an important role in creating new opportunities for service providers to monetize their services.
As consumers demand greater amounts of high-quality content for their in-home and mobile devices, service providers (SPs) are finding it difficult to increase revenues while containing costs. In response, many SPs have implemented their own CDNs to reduce content-transport costs and improve the quality of content delivery to customers. While this approach has helped, results have been limited. Read More »
By Francois Le Faucheur, System Architect, Cisco Systems
With the CDN World Summit coming up this week in London (Oct 26-28), it seems a good time to describe something that’s been a working passion of mine: CDNI, where the “I” stands for “Interconnect.”
What’s CDNI? It’s a new working group within the Internet Engineering Task Force (IETF), co-chaired by myself and my colleague Rich Woundy, Senior VP/Software and Applications for Comcast. The working group mission is to develop standards for the interconnection of CDNs run by different operators, so they can inter-operate and collectively provide a multi-footprint CDN — thereby pooling resources that are owned and operated by multiple operators.
It’s all about how to link up the world’s CDNs, so that they can exchange information and work collectively. Just like the Internet was about linking up the world’s IP networks.
What if your mobile device allowed you the freedom to seamlessly roam across any network in the world, regardless of location or operator and with all the attributes you would expect, security or privacy… With LISPmob, we may have gotten a giant step closer as we open sourced a network stack for network mobility on Linux platforms, an implementation of basic LISP mobile node functionalities.
Imagine how “happy” your eyeballs would become when you realize that your Internet connection failover time was drastically reduced from a full minute to less than half a second, Dan Wing and Andrew Yourtchenko of Cisco developed a methodology to do just that.
The Internet is changing. Network operators and content providers are beginning the widespread global deployment of IPv6, while keeping IPv4 up and running until IPv6 is ready to take over. Dan and Andrew have contributed to the cause of easing the adoption of IPv6 by documenting a methodology that will enable client applications to react more responsively in dual-stack failure scenarios by aggressively rectifying intermittent access issues and therefore preserve the end user experience for dual-stack IPv4 and IPv6 devices. This solution is documented in their IETF draft, cleverly named Happy Eyeballs. It is designed to keep the eyeballs of a computer end user “happy” in the face of problems that may exist when a host is attempting to establish IPv4 or IPv6 connectivity. The IETF draft document describes how client applications should behave when establishing IPv6 and IPv4 connectivity simultaneously, preferring IPv6 if the connectivity is successful, and disconnecting any remaining redundant (IPv4 / TCP) connections. By failing over quickly from IPv6 to IPv4, or from IPv4 to IPv6, the user is not affected by problems that occur in only one of the two IP versions in a dual-stack deployment. This can greatly reduce the connection times in problematic situations -- from minutes to milliseconds, compared to the typical behavior in many implementations today.
In anticipation of World Ipv6 Day, Google Chrome has adopted a similar approach to what Dan and Andrew have documented, under the somewhat less light-hearted name “IPv4-Fallback”. This modification promises to ease potential trouble spots on World IPv6 Day, as well as future browser interactions with dual-stack network configurations. Google’s Internet browser, Chrome 11, uses a “hybrid” variation of Happy Eyeballs that is responsible for establishing, monitoring, and management of simultaneous parallel IP connections. This software enhancement produces significant results by reducing the fallback latency of a problematic IPv6 connection from between 20 and 75 seconds as is often seen today, to as little as 300 milliseconds.
Widely deployed from the core to the edge, IP/MPLS has achieved huge success as a mature, standards-based technology now deployed by virtually every service provider worldwide. As a result, the industry has chosen to extend IP/MPLS and develop a transport profile called MPLS-TP (MPLS Transport Profile). MPLS-TP is the packet transport technology of choice being developed by the IETF to allow service providers to cost effectively migrate existing transport networks to packet based solutions.
Recently EANTC conducted an MPLS-TP Interoperability Event which focused on testing and demonstrating interoperability of key IP/MPLS and Carrier Ethernet features between multiple vendor platforms. This represented a critical technology demonstration for service providers as they begin their migration to packet transport networks.