Modern navigation systems such as those available in cars, smartphones, and high-tech gadgets have become immensely popular. They efficiently guide you to your destination with real-time information. It is cumbersome and unsafe to look at paper maps while driving. Also, one does not need to stop and ask for directions anymore.

The reason navigation systems such as Google Maps are so accurate is because they use past, present, and future information effectively.

• Past: Navigation systems have access to a large bank of data that they can utilize to predict trends.
• Present: Navigation systems get current traffic information for a particular path and inform the user of the time it will take and the directions to get there. They get localized traffic information for the path they are interested in.
• Future: Navigation systems utilize past traffic patterns to inform the user of future traffic. One can search for directions to say the nearest Walmart from home at a time in the future. The system predicts traffic for that path from past data and informs the user of the time it will take and the directions to get there.

Navigation systems can enable the user to find closest locations of interest. They can redirect users on alternate paths when there are traffic jams. These same principles can be used by a network. Utilizing past, present, and future traffic information can enable network devices to route packets on network links efficiently.

Just as modern navigation systems have reimagined and replaced paper maps – we need a way to rethink networks. Legacy networking protocols such as EIGRP, OSPF, BGP, and their extensions are good at conveying connectivity. However, they are not good at conveying capacity, performance, and service needs. DNS-SD, DHCP, SLP, and other protocols enable service discovery but lack knowledge of path preference. OpFlex, NIC, OpenStack Congress, and others convey application intent but lack knowledge of network state. OVSDB, eAPI, and other protocols provide programmatic control and integration but lack real-time network knowledge. Each of these solutions perform a single task but are unable to integrate past, present, and future network information to provide a solution.

Many companies like Facebook, Google, and others have built their own systems to convey real-time information on capacity and performance of pathways to make these decisions. Some have even eliminated traditional routing protocols within their networks. 

Extending existing protocols to perform additional tasks has usually been the norm for network innovation. This leads to solutions mired in complexity that are prone to errors and impossible to debug.

Network innovation must imbibe skills from big data, machine learning, artificial intelligence, and high speed in-memory databases to integrate past, present, and future information.

• Past: Networks can build a profile of traffic trends over time. Big data analytics can utilize this to alert network administrators of traffic trends. This can be used to resource networks at appropriate locations in future.
• Present: High speed in-memory databases can be used to share real-time network traffic information with network nodes so that packets can take the best possible path at any given time. Instead of sharing the entire network state information – only path information of relevance would be shared similar to finding best path to get home from work during rush hour. These databases can also be used for elastic service discovery similar to finding the closest Walmart.
• Future: Machine learning and artificial intelligence can be used to predict future network traffic on given pathways similar to finding best path to Walmart tomorrow. Based on sequence of changes in the network in the past which may have resulted in a failure – the administrator can be alerted if a similar pattern is being repeated before the network fails.

Modern navigation systems have become a part of newer automobiles guiding them on existing roadways. Networks can work in a similar fashion with existing routing protocols doing what they do with the added intelligence from past, present, and future information to send packets efficiently.

Modern applications revolving around 5G, IoT, and cloud require networks to find efficient pathways to be responsive. There is a dire need for a navigation system for networks which combines past, present, and future information. Packets can reach their destinations today. However, they could get there by utilizing better pathways during times of congestion. Next time you are in a traffic jam and find yourself reaching out to your favorite navigation system to find an alternative route – remember the same can be done with your network!