In the past decade, we have witnessed the rise of hyperscale cloud computing, the benefits of which have been evident. But as Internet of Things (IoT) and artificial intelligence (AI) become more commonplace in digital strategies, businesses start to realise that a centralised implementation of cloud has its limitations.
“IoT solutions generate data that require near-real-time processing, and this can be done more efficiently when the computing power is closer to the source,” said Lisa Ong, Principal Lecturer and Consultant, Software Systems Practice, NUS-ISS. She was speaking at a keynote presentation titled ‘Intelligence on the (bleeding) Edge/Fog’ during NUS-ISS’ Learning Day 2019.
One of the architectures that has been identified as a solution to what is sometimes known as the ‘proximity problem’, is edge computing. While it is not a new concept, it has become one of the hottest buzzwords in technology.
“Edge (or fog) computing is a form of distributed computing dating back to the 1960s. Simply put, it enables data processing to be done close to the source of data generation, or the customer,” Lisa explained.
Research firm Gartner predicts that by 2022, more than 50 percent of enterprise-generated data are expected to be created and processed outside the data centre or cloud – closer to the consumer. Doing so would reduce the latency, bandwidth and overhead traditionally required for a centralised data centre. “These advantages become increasingly valuable as companies move towards extensive adoption of big data processing,” she added.
Addressing the limitations of centralised clouds
From storage solutions such as Dropbox to application hosting services enabling enterprises to modernise their applications at scale, the cloud has become a default for companies going through digital transformation.
“Besides reducing the cost of purchasing and managing IT systems, cloud computing also endorses a mobile workflow that has become essential in a digital workplace,” said Lisa.
However, as IoT increasingly becomes the fabric enabling the exchange of information in smart city processes, an estimated 41.6 billion devices will be connected by 2025. An exponential increase in the volume of data will bring about a need for IoT applications to have real-time responsiveness and reliability.
“This is essentially what drives the interest in edge computing,” Lisa said. She explained that the main issues with a centralised cloud infrastructure include latency, data privacy and security, as well as reliability. “We are relying more and more on technology for emergency situations, and in those cases, every millisecond matters. Some use cases would be the safety sensors in manufacturing plants, or in autonomous vehicle environments. When you transmit data to the cloud, you are constrained by how fast the internet can go – and sometimes, a latency of even one millisecond could cause accidents and cost lives.”
A decentralised complement at the edge
However, while edge computing has been posited as an alternative to the cloud, it will not replace it.
Rather, Lisa emphasised, it will complement and complete an enterprise cloud strategy. “With edge computing, some of the current cloud scenarios can be moved down to the edge.” The self-driving shuttle buses recently deployed in the National University of Singapore (NUS) is one example. Part of the data generated by the sensors can be sent to NUS’ central server for processing, to generate higher level insights and identify commuter patterns. But at the same time, the moving bus processes data locally for functions that require time-sensitive inputs, such as safety purposes.
It could be also used in the Lamppost-as-a-Platform project, which is part of Singapore’s Smart Nation agenda. Some of the processing could take place in the edge devices on individual lampposts. “The lampposts cannot be sending all the data collected from sensors to a central server – it’s going to take too long, and with that amount of data, it’s not going to scale,” Lisa said.
In addition, the imminent advent of high bandwidth 5G networks is another factor driving the push for edge devices. “The convergence of 5G and IoT devices will result in a formidable force, enabling smart devices to talk to and share data with each other at high speed.”
Intelligence on the edge
Businesses are expected to rely more and more on edge computing to optimise business processes and address the key limitations of centralised cloud computing.
In January 2019, the two largest international consortia in industrial IoT, fog and edge computing – The Industrial Internet Consortium (IIC) and the OpenFog Consortium (OpenFog) – merged. “On the macro level, there is hardly any doubt that industrial IoT effort is moving into edge and fog computing,” Lisa said.
“In short, small latency means big business; and as the digital world matures, edge computing is poised to help businesses unlock the value that real-time information can deliver,” she concluded.
Click here to find out more on Software Systems courses or Stackable Graduate Certificate in Architecting Smart Systems offered by NUS-ISS.