Software-defined networks (SDN) are a new paradigm shift in the world of network centralized command and control, providing network omniscience and separates control and data planes. Most of the research work till date focuses on increasing efficiency and manageability of computational and storage resources which results in emergence of current virtualization technologies. The feasibility and applications of SDN in current datacenters and network infrastructures is being studied by academia, industry, and the standardization bodies. This chapter explains SDN concepts and its difference from legacy networking, interrelated terminologies, protocols, programming languages, benefits, and shortcomings. Moreover, exploration of current research areas and techniques along with in-depth analysis and future research directions will be presented.
TopIntroduction
Software Defined Networks concept is not new and have started to evolve since 1998 from Ipsilon’s general switch management protocol (Network Working Group, 1998), IETF FORCES (Retana, Atlas & Brungard, 2015), IETF path computational element (Architecture Adrian Farrel), Clean slate 4D (Greenberg et al., 2015), Ethane (Casado et al., 2007)which was full scale research implementation.
Mostly SDN is confused with Network Function virtualization (NFV) terminology; Technically SDN is the architecture of network which separates control plane from data plane, whereas NFV is taking physical networking equipment such as routers, switches, firewalls, load balancers and running them as virtual machines (VM) (Network Function Virtualisation, 2012) as shown in Figure 1.
Figure 1. Relationship between NFV and SDN
SDN predominantly software based enjoys larger acceptability and ease of deployment; which is not true in case of hardware innovation cycle that may take years to reach market. SDN runs control software on modern multi-core processors resulting in easy updating of control protocols and faster processing. This software based modular approach enables devices in data plane to work without any processing overheads, resulting in simpler and lower cost switches. The approach provides better platform for protocol development as in case of Google which implemented their WAN connectivity in VAHDAT project (Software Define WAN, 2015). This methodology of software define WAN architecture enables Google to successfully improves Network utilization up to 100% as well as time critical traffic get prioritization and remaining bandwidth is filled with bulk data.
TopSdn Architecture
In conventional network, devices comprises of almost all the layers including management, control and data plane which increases touch points in network causing lack of visibility and control. Moreover legacy network works on the basis of Human in loop; this situation makes it more challenging and causes serious degradation of network performance, since humans cannot fix all the problems in real-time and their responses are more reactive rather than pro-active as shown in figure 2.
Figure 2. Layered Approach of Device
In case of SDN network, controller serves as a bridge between human in loop and machine. Controller gives real time situation awareness, effect, redirects the flows by sending configuration changes to network devices and modify flow table of network devices as shown in figure 3. Here the North bound and south bound interface are not related to flow of data but instead they are from perspective of SDN Controller in control plane.