Networking infrastructure has rapidly grown from traditional networking devices, single Bluetooth device to massive data centers and cloud computing facilities. Thus Internet of Things (IoT) becomes paramount important topic to the networking world. As long as internet is becoming more and more accessible and more usable the number of technology nodes connected to the inter-network is rapidly growing. Network heterogeneity and vendor specific protocols provide lesser flexibility to network growth and researches. Therefore modern network should be much more flexible and scalable. Along with the network growth technology should have a proper way of implementing network programmability to support dynamic network environment. Today’s networks are not programmable and because of that, in order to increase scalability and flexibility, massive amount of funding and man hours are required. There should be more proper way of implementing network programmability that count for more research and innovations in the networking, without affecting the heterogeneous network basis and various network protocols that run on the network itself nowadays. To achieve programmability and flexibility, network must be capable of understanding overall big picture of the network, as well as controlling upon the data forwarding. Thus centralization of control plane to support distributed data forwarding planes are required.

Growing complex network facilities including huge data centers, massive clouds as well as fast growing ISP/Telecommunication core networks and especially so called Big Data concepts are clamoring for faster forwarding, ACL applications and QoS (Quality of Services) adjustments and also security appliances. And also they required more traffic shaping, traffic engineering and more quality control over data forwarding to reduce traffic congestions and achieve faster data forwarding.

There are four major networking node categories can be identified in the existing scenario.

  • End devices,
  • Backbone Switches,
  • Edge routers
  • Backbone routers.

In each of this devices there are four major levels of processing.

  • Interface Level
  • Protocol Level
  • Flow Level
  • Packet Level

And also there are some severe drawbacks of prevailing networking technologies. Mostly data path processing done by the hard wired approach controlled by the distributed control planes. Therefore most of the time protocol understanding and conversions, QoS, payload transcoding and other custom networking functions are carried out in the higher abstract layers which is affecting the high speed packet processing functions such as de-capsulation (de-multiplexing), processing and forwarding.

Main causes is that, although layered network architecture supports good point to point as well as end to end connectivity, it creates a significant bottleneck with the layered model of unable to jump over layers. And also due to this narrow overview and flexibility it account lesser support for flow level processing such as QoS assignments than other levels of processing such as interface level, protocol level and packet level. But when you see the core of any network it is a bottleneck of having single packet based processing rather than flow based processing where millions of flows can be identified exploiting layered approach.

As long as we focus on more higher and abstract layers, it is more lucrative for cost benefits, but not good to cater massively parallel and high speed processing. In other hand our focus is towards the more hardware and physical acceleration, it is more costly, but providing high speed data plane. Therefore advanced and distributed data plane that can process either the flows or packets that can be controlled via centralized controller may be a good option to satisfy the needs of the networking world.

Software Define Network (SDN) concept which compromises the separation of data forwarding plane from control planes, has revolutionary changed the traditional network architecture consist of data forwarding with distributed controlling architecture. SDN has emerged its association with the ideas of network virtualization involves with Network Function Virtualization (NFV), virtualized network infrastructure (VNI) and virtualized computing resources (VCR). Therefore SDN is basically designed to make higher level abstraction on top of which can be built the hardware/software infrastructure. This concept is paramount important when we comes establishment and maintenance of Data centers and carrier networks and cloud networks in new cloud computing era which prefer much more virtualized resources and connection than traditional network architectures. Although Prevailing fixed network infrastructure which is the caliber of growth in size and scope of the computer network, It is less flexible, less performance and inefficient when we consider carrier and enterprise networks. In the concept of SDN there are 3 main planes which can be communicate via secured and specific protocols.

Centralized SDN control plane is the vital and most significant concept in the SDN technology which lies in between higher level abstraction of application layer and high speed real time data processing and forwarding planes. SDN controller itself have the overall understanding and the big picture of the whole network. If it is a massive network centralized SDN controlling plane can be distributed among several SDN controllers. So with SDN concept we can achieve the necessary,

  • Flexibility
  • Scalability
  • Programmability
  • Extendibility
  • Controllability
  • Modularity

That we are lagging behind.

What is Artificial Intelligence?
Artificial Intelligence is becoming more and more realistic and applicable in the technologies that creates more processing and decision making powers without any manual intervention. This approaches is very good for the complex and complicated scenarios where human mind cannot understand and process based on the overall view. Based on the developments of the neural networks, Pattern recognition and regressions and logical inference, Intelligent Agents have evolved from simple applications to large intelligent units for better applications such data mining and searching. Companies such as IBM, Google have emerges the applicability and suitability of the intelligent agents in high tech applications. More appropriately artificial intelligence may be the best solution ever for the secured and efficient centralized controller. Any intelligent agent need facts, environment perception, logical functioning as well as responses.
SDN Controller?

If we combined these two scenarios we can see the actual potential of perceiving SDN controller as an intelligent agent who has the abilities to perceive its network and control distributed data planes with less intelligent to optimize overall networking functionalities and throughput. Rather than looking forward into more locally optimized network control, we can have globally understandable and optimized network where any network functions and virtual-functions can be run upon that basis. Understanding the whole network will be an additional advantages to avoid vulnerable security issues in distributed network nodes, but creating a single point of failure if we not identified and set better security policies on SDN controller. If we have an intelligent agent that can perceive the entire network facility and take necessary decision based on global scenario, it will be a massive leap forward in networking technologies. Data Planes are not completely dumb devices, but have little amount of intelligent to abbey to its controller and communicate securely throughout the network. Independent distributed network nodes only need to inform its superior controller about the conditions and situation around it. More intelligent agent capabilities and functions that embedded into controlling agent, more security and programmability and scalability will be provided. In this special scenario, protocols are the way of perceiving network and enabling data plane response and processing.

Distributed Data Planes only need to understand and process protocol information and execute it upon the packets that it need to be forwarded. Data Planed switches do not need to process packet by packet and forward them based on higher layer data, but it need to identified flows and forward packets based on the relevant flow. Therefore higher layer functionalities such as routing, ACL processing, QoS (Quality of Service) assignments, Security applications can be performed in the data plane apart from their conventional point to point connections and L2 forwarding.

Most important part is that controller agent need to understand the whole network and function over the information sent by data planes and make appropriate decisions suitable for global scenario and program the data plane to work like it has intelligent. Therefore controller agent functions are paramount important here if we will deploy intelligent agent instead of traditional SDN controller. For this purpose to be adopted to the network environment some ontology based detection, identifying and interpretation capabilities are essentially required. But the emerging innovations will provide better solution for artificial intelligence and intelligent network agent to handle large scale network such as clouds, core networks and isp providers.

We have seen some emerging deep learning and artificial intelligence applications that are recently gives more benefits. As the ultimate goal of centralization of the network control means that we can easily make the network intelligent to understand and response to various situations. And also we see the potential of SDN to deploy massive learning and optimization algorithms on networks which are more complex today that cannot be catered with existing solutions. SDN itself identify the networks as flows and even in the control part we have to consider multi-dimensional system. Therefore intelligent networking platform has a lots of caliber to repidimize the networks.