Bee Foraging Provides Insight on Optimization of Server Allocation
Perhaps you don’t instantly recognize the connection between the Internet and honeybees. But researchers from the Georgia Institute of Technology (where the school mascot is, coincidentally, the Yellow Jacket) do. A Georgia Tech group has developed an algorithm based on honeybee foraging strategies to optimize server allocation to keep websites running smoothly.
Honeybees and servers face similar optimization problems: efficiently allocating limited resources (e.g. worker bees or computing power) in the face of variable demand. Through swarm intelligence, honeybees operate entirely devoid of central command. The hive’s survival depends on worker bees finding and retrieving nectar. So how do the honeybees ensure that they work productively? They dance, of course.
Honeybees scout for nectar until they find it and then return to the hive. The scout bee crawls up on a vertical honeycomb and starts dancing. This jig is called the “waggle dance.” Every turn has meaning. The direction of the dance indicates the direction (relative to the sun) of the nectar, the number of waggles tells how far to fly, and the length of the dance connotes how sweet the source. The forager bees learn the directions by following the dance, turning the hive into a Broadway show-stopper.
When the source is very sweet, the waggle dance can go on for quite a while. This allows more forager bees to see the dance so greater numbers will be recruited to that nectar site. Yet during this time, as the sun continues to move, the position of the nectar source relative to the Sun constantly changes. Amazingly, the dancing bees adjust their angle to account for the Sun’s movement.
Returning bees keep the dance going as long as the source is still abundant. As the bees deplete the source, the dancing slowly tapers off. Meanwhile, scout bees from new sources enter the dance floor, allowing the hive to deftly shift from a dwindling supply to a fresh one. This competition ensures that available nectar sites are optimally staffed with workers at all times as the conditions change.
The researchers at Georgia Tech recognized that server optimization problems resulting from “abnormal” demand could be solved by developing an algorithm based on the bee foraging strategies. Servers allocate computing power to websites as the demand on those sites varies. Under previous strategies, sites were optimized for “normal” demand, but when strained these sites would crash. All the while, servers for other sites remained idle.
The honeybee algorithm guarantees less busy servers don’t sit by idly during times of high demand. When an overloaded server receives a request for access to a website, it places an ad to recruit other servers to the in-demand site. The higher the demand, the longer the ad remains posted, and the more computing power is contributed to the run the site (examples, like Ticketmaster or YouTube). “Computational results indicate that the new algorithm is highly adaptive to widely varying external environments and quite competitive against benchmark assessment algorithms.” 1 In laymen’s terms, this new algorithm is the bee’s knees!
The research from the field of biomimetics (biologically inspired engineering) continues to demonstrate that natural systems are endowed with specialized knowledge from which even the most intelligent human designers can benefit. Engineered biomimicry has brought society a wide array of technologies from robust adhesives to optimized server strategies.
Wisely, researchers heed the biblical imperative to “ask the animals and they will teach you.” In Proverbs 30:24-28, the Bible says, “Four things on earth are small, yet they are extremely wise: Ants are creatures of little strength, yet they store up their food in the summer; coneys are creatures of little power, yet they make their home in the crags; locusts have no king, yet they advance together in ranks; a lizard can be caught with the hand, yet it is found in kings’ palaces.”
What can we learn from the animals today?
Katie Galloway is an RTB volunteer apologist. She is currently completing her PhD at Caltech in chemical engineering with a minor in biology. Her research focuses on designing biological systems.