SDN is used in data centers for internet protocol (IP) routing and switching, and is increasingly being used in carrier networks.
To illustrate, imagine you drive a car from A to B, where your car represents an IP packet moving from server A to B.
In conventional networks, road signs will tell you where you need to go. In an IP network, routers and switches will direct data packets through the network. The reality is however that the routing tables (road signs) were probably installed many years ago, and have not adapted to the reality that is constantly changing. They don't know about road works or traffic jams ahead, and will not be able to determine alternative routes that will lead to point B faster. In those networks, it is virtually impossible to update the road signs, as technicians are needed to reconfigure each router and switch manually. This is expensive and slow.
This results in inefficient use of the available roads and will make the driving experience poor.
When SDN is implemented in a network, it corresponds to using a central traffic command system directing self-driving cars. Real-time traffic information is used by SDN controllers and cars are steered remotely at each intersection without any human intervention.
Applications can also control traffic directly. Applications that are sensitive to jitter and latency like gaming and voice over IP can move through the network in fast lanes, while data backups and non-critical downloads are sent to back roads.
With networks being able to be controlled dynamically, bandwidth and performance is improved dramatically. Previously, congestion had to be relieved by building faster links (bigger pipes) and by over-dimensioning. Although this was reliable, it was also expensive. These methods are also no longer sustainable due to the continued increase in demand for bandwidth. SDN reduces choke points and improved network throughput without having to widen roads and highways.
Any comments? Please post them below. Thank you.
Share this post