SD-ANTs

Ok, I admit I’m a geek. That said, I am not just a network geek. I have a real geekiness when it comes to biology. Since childhood, I’ve had a keen interest in it, and could spend whole afternoons observing insects. I still marvel at how things so small can be so intelligent about their collective ability to survive in a dangerous world. Yes, as a boy I did have an ant farm in my room, and I spent quite a bit of time studying it. My mother was unhappy with me when I spilled it on the carpet but accidents happen.

Swarm intelligence is one of the areas of biology that has always interested me. How it is that ants, bees, and termites are as intelligent as a colony yet as simplistic as individuals?

Sci-Fi frequently pits humans against a swarm intelligence. H.G. Wells has Empire of the Ants, Star Trek has the Borg, Enders Game has Formics, Star Ship Troopers has the Arachnids (book please!… let’s pretend the movie never happened), and even Terminator’s Skynet is a form of swarm intelligence. Unfortunately, in Sci-Fi stories, swarm intelligence is usually the enemy of humanity… but hey it is fiction and fun.

Let’s go back to my harmless but innocent nonfiction ant colony of our discussion and my childhood. Each worker ant is really a quite simple and rather dull creature. They cannot survive long away from their colony. The magic is how each individual worker’s experience is shared via olfactory cues, sounds, touch, and movements when they contact other ants from the same colony.

That said, I’ve always found it clever when worker ants, distant from the colony, are still able to respond, rapidly and appropriately, to many situations. They don’t need to check in with the nest for guidance to make quick, tactical decisions, which would only slow things down. Workers have just enough sensing and intelligence to react and adapt to the most probable situations appropriately. The colony has a distributed, time-efficient intelligence.

On the other hand, the worker ants are not wandering around blindly without guidance. The nest is centralized and intelligently synchronized. The central location of the nest and its hierarchy provides a foundation for the colony’s goals of continued existence and the eventual creation of new colonies. The nest provides both guidance and the central knowledge base of the collective knowledge and experience of the colony and the workers that come in from the field.

Interestingly, the queen ant is not that intelligent either. It’s the nest surrounding the queen that is an intelligent super organism. This intelligence is shared with the queen and among workers in the nest. It is the connection and sharing of information that makes the nest smart not individual ants. The queen ant just provides the center and synchronization point for the colony.

It is fascinating how resilient ant colonies are. They adapt so rapidly and seamlessly to issues at the nest or in the field for opportunity and hardship. If a good food source is identified in the field, the workers relay the message to the colony, and rapidly expand resources to exploit it to the colony’s advantage. If the colony does not react rapidly, another colony may jump in and take the food source resulting in a lost opportunity. Rapid decision making for an ant colony aligned with its core, strategic purpose is amazing to me.

So ants have tiered levels of intelligence, the distributed, timely tactical intelligence of remote workers, and a centralized, strategic intelligence of the nest driving the colony-wide survival and reproduction of the colony.

Why do I ramble on such things when I am a network guy dealing with hardware and software? Because swarm intelligence is an inspiration for how Talari developed its products, and I believe it should be contemplated by others as we evolve into a Software-Defined (SDx) universe. SDN controllers are very good with centralized intelligence at the data center, but they run into problems when time becomes a factor in the outer reaches of the network because of the distance to the SDN controller. If all the brains are centralized, the remote nodes of the network will be slower to respond to situations. In the DC, or even in a campus environment, this issue may not be as severe when latencies are low and bandwidth is plentiful. But in the real world of WANs, as well as ants, there is time and space between the centralized nest of intelligence and the remote outreaches of the colony. The worker ants and the SD-WAN appliances must be able to think on their feet to react to real time situations that occur to both exploit opportunities and avoid adversity. This does not mean that the remote nodes require burdensome hardware and software. The outer reaches of the network can use simple, standard hardware that is more than capable of performing these functions. The worker ants don’t need to be brainiacs, but they do need to have brains.

At the same time, information learned in the outer parts of the network should be gathered and brought back to the nest so the colony and the network may strategically adjust its overall behavior. So distribute intelligence for timely adaptation, central policy, and intelligence for strategic purpose and adjustment. Talari does this with our Talari Aware (nest), Talari Network Control Node (queen), Talari Client (worker), and Adaptive Private Network technology (colony) to provide a product that has an almost biological intelligence to provide industry leading service levels to the network users.

I promise that we are not the creators of anti-humanity technology. Our technology is only intended for the greater good of humans and ants alike.

Trust that our ants were always good ants as long as you avoid spilling them on the carpet.

Categories: Software Defined WAN (SD-WAN)

Tags:


Subscribe