What is SDN and NFV ?
The Internet has successfully been growing for more than 20 years the growth in demand has so far been met by introducing even larger and larger routers. This has been beneficial and to scale in public networks. However, in order to meet today’s steadily growing demand for Internet access and other packet-based services, there is a present need to deploy more efficient packet networks also within metro and aggregation network domain. The attempt to copy the approach from the coarsely populated, but large core network sites and migrate to metro and aggregation network sites may not be the most cost optimal approach. It may be time now to split the router architecture in similar ways as was done in the traditional mobile core network, in order to penetrate the highly dense metro/aggregation networks.
Splitting the router control and forwarding plane forms the initial idea of software defined networking (SDN).
The initial idea was born to decouple the routing intelligence software from simple forwarding hardware allowing, particularly for academic research networks and test beds, fast prototyping and evaluation of new control theories and algorithms. It was part of the clean Slate Internet Design initiative of Stanford University. The target is to develop a system that is amenable to high-performance and low-cost implementations and capable of supporting a broad range of research, can isolate experimental traffic from production, and is consistent with vendors’ need for closed platforms.
The key technical idea of SDN is to provide an open control interface to the operating system of the network device without compromising the details of the implementation, an important business aspect for equipment manufacturers. This is enabled by support of Open flow in the operating system and is based on the ternary Content addressable Memory (TCAM)-based flow tables, most routers and switches make use of. In a classical router or switch, the fast packet forwarding data path and the high-level routing decisions in the control path occur on the same device. An OpenFlow-enabled switch separates these two functions. The data path portion still resides in the switch, while high-level routing decisions are moved to a flow controller, typically a standard server. The OpenFlow Switch and Controller communicate via the OpenFlow protocol, which defines operation and management (OAM) messages.
Besides this technical view, this split design will enable a cost reduction and new market opportunities by the basic principle of modularization. This is of high importance for supporting flexible network innovations because the development cycles of hardware and software components are extremely different, and the modularization supports a decoupling of the innovations from a market perspective. The right layering approach will enable high market volumes for specific modules (software or hardware).
The introduction of the SDN concept into real networks would have a profound impact on the way in which networks are built and operated. In order to understand and evaluate the practical implications of the general concept, it would be beneficial to first test it in research networks. Feedback from the experimental implementation will be crucial in improving the overall concept and allow taking the concept to further applications in networking. First trials are currently under way in selected US universities, which focus on the easy management and reconfiguration of research networks, for example, for applications in the field of Clean Slate research.
While SDN brings innovative evolution to network routers and
switches, the network comprises other types of devices besides routers and switches. Network operators enforce network policies using a combination of switches and network functions (NF). Policies may be complex, such as ensuring that unauthorized users are prevented from accessing sensitive servers or malicious traffic is eliminated from the network. To do this, an operator could use a stateful firewall to ensure that only traffic initiated from within the network is permitted and in doing so protect users from malicious traffic. Indeed, today’s networks heavily rely on a wide spectrum of NFs. The diversity and complexity
of NFs have been further expanded as the proliferation of wireless devices and mobile applications. NFs offer a variety of valuable benefits, ranging from improving security (e.g., firewalls, intrusion detection systems, and deep packet inspection), improving performance (e.g., proxies, caches) and reducing bandwidth costs (e.g.,WAN optimizers, video transcoder). However, despite their benefits, NFs come with high infrastructure and management costs. One important reason is their complex and specialized processing. As a direct result of this complexity, configuration errors are common configuration errors comprise as much as 65% of the network outages. Other reasons of their complexity come from the lack of standardized management tools across different devices and vendors. Moreover, there is a need to consider policy interactions between these appliances and other network infrastructure, which cannot be easily troubleshot.
To facilitate programmability and flexibility of NFs, in 2012, operators initiated a new concept, called network functions virtualization (NFV) within the European Telecommunications Standards Institute (ETSI) consortium. Instead of building NFs in the form of proprietary hardware boxes, NFV calls for the virtualization of them. Using virtualization and cloud technologies, it allows legacy NFs to be flexibly deployed in the form of software on commodity servers. Sharing the same spirit of splitting the router’s control plane from forwarding plane, the decoupling of NF software from the hardware facilitates a faster pace for innovations and shorter development cycles, and result in shorter time to market of new services.
0 Comments