Real time computing on the edge

Whether you like sports or not the Olympic motto: “Citius, Altius Fortius” (“Faster, Higher, Stronger”) should not be rare to you. We live in a society where boundaries are pushed higher every day and new goals may seem impossible to achieve. In this scenario we are progressing on fields that were science fiction just a few years ago. The possibility to measure and take immediate actions was only available to scarce industrial sectors, few economies and only embedded mostly in the arms industry for obvious reasons.

Nowadays we have at our disposal elements that are not only capable of sending information to the cloud for analysis and solution retrieval, but we are designing new concepts that require to send enormous amount of data which needs to be processed so fast that near immediate action can be taken. Sending and receiving information at the Access layer (the part of the network next to us) is not difficult.

4G WorldWide Coverage

4G – LTE (high speed mobile network) is widely available while initial 5G tests are on the way. So, if speed at access level is not the problem, where do we see the problem? Simple, where computing is done: At the core.

When we send data to be analyzed not only do we have to consider the time that the analysis requires, we also have to consider the time it takes for the data to leave our device and to get back. As data computing takes place in datacenters data needs to travel many kilometers, sometimes hundreds or even thousands. This may take from a few to hundreds of milliseconds in worst case scenario.

While the application of technology until now did not need to cut this time down we are faced with new challenges that is in need of this rapidness. Let´s take for instance the self-driving car technology.

A self-driving car has no less than a few hundred sensors, not only the “traditional” ones like speed, oil, tire pressure, etc but now there are many more like proximity to other cars, environmental hazards, etc. The values of those sensors can not travel away for the car to take a decision. The car itself has become a datacenter mobile where data computing and values received are transformed into actions. We just can not afford to send that key data away to receive an answer as a matter of milliseconds will be the difference between life and death.

 

Only certain data will travel away from the car, which will help the overall car network to stablish whether the traffic is fluid (based on position, speed and concurrent data received from other cars) informing of better alternatives to travel.

Likewise, bringing the datacenter to the edge will mean that less data has to be transported in between nodes of the network, making this transport better as latency, costs to increase bandwidth and QoS (Quality of Service) will improve.

This new paradigm means that all the access infrastructure needs to be redesign. It can´t just be a box that will link electronically places. It must have the ability to “think” by itself. Where we were aggregating links before we have to be able now to include servers for data analysis, firewalls so unauthorized users can not access the local or remote servers and move data quickly enough so real time decisions can take place. It may not be something that will happen immediately but customer demand for always-on, quick data retrieval and security is making this change happen faster than expected.

 

Be the first to comment

Leave a Reply