Yeasterday I made an invited talk on Software Defined Networks (SDN) at the Workshop “Network Operators’ Views on Future Networks Landscape” (organized by the University of Athens, Dept. of Informatics and Telecommunications). Other Operators, technology Providers and Academic Partners joined to debate which is ways for future networks, with the related biz.

We’ve already elaborated about Software Defined Networks in a previous post. In simple words, in a SDN control and data planes are decoupled, so that control intelligence and state can be logically centralized, whilst the underlying network infrastructure is virtualized, abstracted from upper network’s applications. As a result, they claim, network is gaining programmability and flexibility in order to adapt to changing business needs. Is this a way for commoditizing network equipment or a deeper transformation ? Is this a potential milestone on the way to become Smart Operators (vs Lean) ?

Recently Google has announced that they have built their own OpenFlow-based SDN (G-Scale Network) to manage all of their production WAN traffic. Indeed this is new way of looking at networks, not only within data centers but also for interconnecting them with WAN, more flexible and less expensive. I’ve already mentioned this paper (Above the Clouds: A Berkeley View of Cloud Computing): in the same direction, they claim that the potential exploitation of SDN in WAN may provide costs dropping more quickly than they have had historically. On the other hand, as an Engineer, I would argue that the more controlling software we put in a networks the more throughput performance are reduced, and we know – as a Network Operator – that carriers’ core routers have to be faster and faster … so is this the way ?

Well, my take today is that if SDN is going to happen, it seems to be reasonable to expected first impacts “at the edge”: it is “simpler” and it guarantees immediate advantages. For example, Users’ (with smart and customizable edge nodes, devices) may access more services and Operators can build (at the edge) “smart” network of networks, even with resources belonging to different Providers.

This is the presentation that I’ve made, proposing such reasoning on SDN: SDN-Athens-6th-June1.wmv

During the workshop several other presentation have been made by Operators, Technology Providers and Academic Partners. I’m just giving short snapshots of the presentations made by Huawey and Telefonica, which I’ve found particularly interesting.

Huawey (Advanced Technology Dept.) presented the Soft-BNG (Broadband Next Generation), a network architecture (for metro backbone) based on SDN paradigms. They are claiming the availability of prototypes of the pieces of equipment of a Soft-BNG network (starting by the edge).

Telefónica Digital talk started by pointing out the expected growing complexity of networks and the more and more dynamic (biz) game of diverse Players in the Telecom and ICT arena. Interestingly, presentation showed the example how most P2P applications can be modeled using  stable matching theory, then focusing on the intertwining between Users’ behaviors and complex network dynamics. In view of these relations, talk argued about the need of developing more RTD activities on these aspects of Network Economics: this paper has been mentioned as one example of the directions in pursuing said investigations towards 0 Capex networks (which is the dream of a Smart Operator).

I’ll keep you posted about follow-ups.

Tags: Industrial Mathematics, Network Economics, Software Defined Networks

This entry was posted on Thursday, June 7th, 2012 at 6:00 am and is filed under Pervasive Computing & Networking. You can follow any responses to this entry through the RSS 2.0 feed. Responses are currently closed, but you can trackback from your own site.