Thursday, April 19, 2012

OpenFlow Poses a Challenge to Net Neutrality, in a Good Way

In principle, some of us would argue, consumer users of Internet apps should have the ability to prioritize their own traffic, or have their ISPs optimize on their behalf, in ways that preserve user experience as deemed important by users themselves. Current network neutrality rules prevent that. OpenFlow might help. 


OpenFlow is software that could allow any end user to create priorities for their own traffic. In principle, ISPS could use OpenFlow to optimize on behalf of their customers. In practice they cannot, as it would violate network neutrality rules that only allow "best effort" for all packets. 


The advantages are relatively obvious. Business users for example, can buy services not subject to net neutrality rules, in ways that are helpful. 


Cadbury, for example, is managing its internal corporate network using application and other priorities after an audit found that 55 percent of its traffic was recreational, says Nolan Rosen, chief marketing officer at Exinda, a consultant to Cadbury, USA Today reports. 


Now, each office now gets bandwidth prioritization based on size of site or its contribution to overall business objectives. Such policies, based on type of application, time of day, day of week, type of user or site are permissible for private networks run by enterprises, but are not allowed for consumer broadband customers.


That's one reason some of us have argued that current network neutrality rules, intended to prevent unfair business practices, also have the effect of denying packet prioritization policies that actually can have high value for discrete end users, based on the priorities those users have. 


Video and voice are two applications most consumers use frequently, and both are real-time services that benefit from policies that preserve bandwidth for such applications at times of network congestion. Gaming is the other obvious application that can benefit from prioritization policies at times of network congestion. 

No comments:

Costs of Creating Machine Learning Models is Up Sharply

With the caveat that we must be careful about making linear extrapolations into the future, training costs of state-of-the-art AI models hav...