If it makes sense, it works - Information Centric Networking

Darjeeling Tea is famous throughout the globe for its unique taste and aroma. This is produced from a small leaved variety of Chinese tea called Camellia Sinensis. But these tea leaves acquire their world famous unique flavor because of the typical weather condition and the geographical landscape of Darjeeling district in eastern part of India – and therefore not produced anywhere else in the world.

Imagine, on a sunny California morning you want to sip through a hot cup of this delicacy. Do you need to fly down eight thousand miles to procure it and have it a couple of days later than your cravings came? The answer obviously is a big NO because you can get packaged Darjeeling tea in a nearby store in California. In fact this is how it works for every product in the world where international trade tries to bring the product as nearer to the customer as possible. In this process two things are achieved – one, reduced time to get access to the product i.e. minimal latency and second, reduced effort to get access to the product i.e. reduced bandwidth requirement.

Now that I could not resist using a couple of networking terminologies in the above statement, let me come straight to the point. If this is the globally accepted efficient way of delivering product to the customers, is this the same way our Internet also works today? In a more specific terms, when I want to watch an YouTube video sitting in a sea beach in Mauritius, does the video get streamed locally from a router node of my Internet Service Provider (ISP) or is it actually coming from a distant YouTube data center located somewhere in US? Again, the obvious answer is – it’s coming from a server which is on the other side of my ISP’s network. This can be compared to the case when a local shop does not cache the product and brings it from the distant source every time a customer requests for it.

Unfortunately, our Internet works in this same way – with some exception of a few Content Delivery Networks like Akamai, BitTorrent etc.  Internet was designed for point to point communication in its early days. Therefore the building block protocol of Internet i.e. TCP/IP is designed for source to destination based communication and binds the communication channel to the location of the server using IP addresses. Even IP addresses are used to uniquely identify the source and the destination node of the communication channel instead of identifying the content being communicated.

But as our needs evolved over time, the biggest usage of Internet today is for delivering content – video, audio and other multimedia. The amount of data getting transferred over Internet every minute is a huge 1354440 GB approximately and consists of huge variety of contents. The following diagram shows different types of contents transferred over the Internet in a minute. Social Networking platforms such as Facebook, LinkedIn etc. and Over-the-Top (OTT) subscription services such as Netflix, Hulu etc. are the biggest contributors to this ever growing data explosion.

figure 1 : Contents transferred over Internet in a minute

In order to support this, the Internet had to evolve. Overtime, ISPs have built overlays (MPLS, IPSec etc.) on top of the basic TCP/IP infrastructure. But more and more overlays in last two decades had made our network very complex – difficult to manage and limiting us to scale.

Let us take a look at one more example. Days are gone when Internet could only be accessed from a personal computer connected through an Ethernet cable. With wireless access technologies becoming more and more efficient, mobile devices are becoming biggest users of Internet. With predicted 38 billion IOT devices being connected to Internet in 2020, mobility is going to become a crying need of the hour. I may start watching a live streaming of a football match in my mobile device at my home. At that point of time, I am connected to Internet through my home WiFi. Now, I want to continue watching the telecast as I get into my car heading towards my office. But my Internet connectivity switches to my mobile data service provider from home WiFi as soon as I get out of my home. This makes me a different destination of communication channel and identifies me through a different IP address altogether. My need is to receive content continuously even if the communication channel changes. But Internet was never designed to support this as TCP/IP based sessions are tied up to the end points. We cannot support this unless our communication paradigm is redesigned to work based on the requested content instead of the communication end points.

The new design of future Internet should ensure the contents residing closer to the access points and should not have any binding with the source or destination of the access – rather a content should be uniquely identifiable globally irrespective of the server from where it is being sourced.

figure 2 : Comparison between Existing vs Proposed Internet design

In the above diagram, let’s assume the big circle as the Internet. The model at the left hand side shows contents residing outside the Internet and being accessed through the Internet. In the proposed model at the right hand side the content is cached and replicated in different access nodes inside the Internet. As a result it can be accessed with lesser latency, less bandwidth usage of Internet core and from different access points with mobility.

Information Centric Networking (ICN) or Content Centric Networking (CCN) is an approach based on the above philosophy which is becoming popular as a proposed long term solution in this space. The story started with Mr. Van Jacobson (along with others) publishing a paper – “Networking Named Concept” at the SIGCOMM, 2009. IRTF (Internet Research Task Force) is doing extensive work in standardizing the proposal into protocol form. The ICN protocol will transfer uniquely identified data over the network in a request–response format while caching the data in the network nodes traversed. The unique identifier of the content will be used for packet switching in the forwarding plane. The protocol does not require any existing layer 3 connectivity like IP and is designed to work agnostic of the underlying layer.

figure 3: Packet exchange in ICN Protocol

While Information Centric Networking comes with promises of solving lot of problems, it has its challenges. The biggest challenge of a paradigm shift in networking philosophy has always been requirement of forklift upgrade of the existing infrastructures. In addition to that - ICN is yet to be validated for high performing applications. On the other hand, the disruptive requirements of 5G networks may just provide enough justification for the Service Provider’s investment and ICN may be able to ride the wave of 5G. “5G will herald an era of faster access, more security and much greater coverage. 5G will also see a new protocol called information centric networking, which in some ways encapsulates what 5G is really all about – a flat ‘any-access’ architecture where content distribution, mobility and security are inherent.” – says Chris Heckscher, Vice President of Cisco who publicly announced ICN as part of their 5G road-map as a network equipment manufacturer.

Only time can tell whether ICN is going to be the future of networking but in the history of the world, any disruptive idea that had made sense, have come!