Almost every day, tons of new data is produced, processed, shared and moves across the globe. The current scenario of mass data generation led Intel to predict that by 2020, an average internet user would consume around 2GB traffic within 24-hours with daily video streaming to stagger approximately 1PB (1 Petabyte is equal to 1,000 Terabytes).
The exponential growth is further fuelled by sheer expansion of IoT devices like smart home technologies and those autonomous cars. In short, with every single purchase of smart devices available currently, we’re only adding to the mass data production posing new set of challenges for data centres around the world.
In combination with the revolution of 5G technology and super-fast worldwide connections, data housing facilities are under massive strain just at the thought of all the processing capacity expectations. Even as these facilities begin upgrading to prepare for the ambush of data and analytics, a new genre of processing seems to have risen which is now a buzz among many and that is Edge Computing.
The concept is comparably simple being a distributed information technology architecture. It allow client side data processing right on the “edge” of the network which is as close to the source as possible. The technology utilises processing power of intermediary and side-line servers that speeds up the mobile services response without straining core network servers.
Time-sensitive data is processed through intermediary servers that are locally housed while less time-sensitive data is migrated to the cloud for long-term storage and big data analytics. Sending sheer volume of raw data over a specific network strains the network resources which is the reason processing data closer to its source makes sense, instead of being transmitted to the data facility only to be retransferred again later!
Many tech and industry experts even suggested that smart devices can be used only to transmit particular or specified data to reduce strain on service networks with the rise of information over the next decade. For instance, monitoring of oil level or brake fluids by autonomous cars and transmitting the data only when the level exceeds the pre-defined limit.
Likewise, a Wi-Fi powered security camera can use edge analytics to only transfer information when percentage of pixels in between the frames varies. As IoT devices further accelerate the processing power, each can be effective mini data centres.
Peter Levine, a venture capitalist at the Wall Street Journal commented that a self-driving car with ability to easily amass above 200 CPUs can be considered a data centre on wheels. In fact, it can be concluded that a self-driving car simply produce too much data which is simply impractical to send it to the cloud server only to retrieve it immediately! A vast field of other smart/intelligent devices would also try using the cloud and must be considered.
Levine further declared that there won’t be enough bandwidth and network speed in between just to achieve all of it. Each device in this scenario would process, store and send relevant data to the cloud on its own which is much like being a brain or system’s core that’s able to analyse all the data and send back the results of what’s learned to the devices. This way, all the devices are able to self-learn from each other which is totally radical.
Other tech experts envisions a future with cloud and edge computing works in coalition which is something that would transform the very nature data centres operates today. Edge computing is more likely to ease the workload from the current cloud platform while processing data for individual devices and let it focus of big data analytics from many different devices.
The very concept of edge computing has become increasingly feasible due to the imminent outgrowth of IoT devices and constantly declining costs of essential computing equipment.
From the details outlined above, it can be concluded that edge computing is ought to change the very picture and notion of data centres by reducing response time to milliseconds, counter bottlenecks and network latency.