This is a blog post by Frank Yue, Technical Marketing Manager at F5.
Recently, the Seattle Seahawks hosted the Green Bay Packers for a Monday night football game in Seattle, Washington. The stadium, CenturyLink Field, holds 65,000 screaming fans that are all watching the game while trying to post pictures to Facebook, Twitter and send SMS messages to their friends with each play. With the sudden increase of network users and their demand for high bandwidth communications, it is not surprising that Internet access for these fans is spotty at best. Imagine the surge in demand when the controversial final play unfolded and decided the game. Network connectivity at major events such as this is always an issue. There is a high density of people in a single area and they are all trying to use the same local radio network to connect to their provider’s network.
Oversubscription is a known phenomenon. It often occurs around disasters and major sporting events like American football and NASCAR races. If there is enough advance knowledge, resources can be added to alleviate the problem. At NASCAR races, wireless providers bring portable cell towers to increase their capacity to handle the extra 100,000 fans that show up for the weekend. When the race is over, all the equipment is removed and stored for use at a future event as necessary.
Wouldn’t it be nice if capacity and availability issues could be resolved dynamically and in near-real time? Electronic signs are placed on frequently used highways to warn motorists of congestion problems before they reach the traffic jam. This encourages the drivers to change their route, using less used roads. Networks today use dynamic routing protocols like BGP, OSPF and IS-IS to automatically change traffic patterns when there are problems in the network infrastructure.
Dynamic Carrier Architecture?
It only makes sense that this functionality migrates to the application layer both from a local and global perspective. Imagine a framework where all the components of the network are providing metrics to a central intelligence. This intelligent system is able to correlate the data to make informed decisions to modify the network to optimise the application delivery.
Imagine a scenario where the DNS infrastructure detects an increased number of requests for a specific website. Information is also gathered from the local load balancer that the number of requests to the website tied to the DNS is increasing as well. Because of the surge of requests, the system decides to increase the number of available servers for the application. The virtual server manager enables more servers to handle the expected load. The local load balancer changes its configuration to add the new servers and make them available once they are up. The surge in traffic can now be handled without any major issues. Once the load drops back to normal levels, the central intelligence can inform the various systems to revert to their previous configuration, freeing up the resources for the next event.
The Internet has an advantage over the NASCAR races where most of the resources are virtual and changes can be made quickly with minimal advance planning as long as the framework is in place. This dynamic framework is developing. Keep an eye on this technology and determine how you can leverage this flexible self-healing architecture.