One of the hottest buzzwords sweeping the networking world is software-defined network (SDN). The basic premise of SDN involves decoupling the control plane from the data plane in networks of routers and switches to create a logically centralized control plane that also allows for slicing and virtualization of the underlying network. Result: developers can program new services quickly and easily on the network without having to worry about the underlying infrastructure.
It's been well understood for some time now that telecom operators need more flexible networks to keep up in the age of customer-centricity and Big Data - hence the move to all-IP architectures, for example. The problem, according to SDN proponents, is that even in an all-IP environment, you only get so much flexibility with routers and switches running various protocols and embedded firmware with proprietary interfaces.
Between that and the added complexity of throwing data centers into the mix, networks are becoming increasingly complex beasts that are tough to manage and costly to run, and make new service provisioning a bigger chore than it needs to be at a time when IT technology is evolving faster than the ability of telecom technology to cope with it, making nimble service creation more crucial than ever.
The SDN approach is now being widely touted as the way forward, giving next-gen networks the operational scale, opex efficiencies and service innovation that operators will need to serve their customers.
To that end, a group of service providers - including Deutsche Telekom, Facebook, Google, Microsoft, NTT Communications, Verizon and Yahoo! -banded together last year to form the Open Networking Foundation (ONF), a consortium tasked with developing a de facto standard for SDN.
The ONF roadmap for SDN envisions the logically centralized control plane using a network operating system that "constructs and presents a logical map of the entire network to services or control applications implemented on top of it." The network OS "provides a central vantage point and well defined API" that would allow operators or third parties (researchers, for example) to write new network management and control apps.
The key to all this may hinge on the protocol used for the interface between the control plane and the data plane. The ONF is backing OpenFlow, which is being touted as an open, vendor-agnostic protocol to enable data/control plane separation, leading some to describe it as "the Android of networking". OpenFlow supporters see its openness as the key to SDN's value proposition.