Businesses face an unusual dilemma as they prepare for a data-driven future. First is big data analytics, centralising data to ensure as much information is available as possible. Second is automation, applying data to increase efficiency and reduce manual intervention.
This creates a problem however. In order to function correctly, automation systems needs to process and action data at the point of collection, at the edge of the network. This is in complete opposition to the centralised model favoured by data-driven industry.
What does this actually look like?
Take self-driving cars for instance. Each vehicle is equipped with thousands of sensors to navigate routes and avoid collisions. In order to succeed, information must be processed in real time – the vehicle cannot tolerate any latency, ruling out cloud-based systems.
At the same time, vehicle manufacturers need to collect data from onboard sensors to drive product development and safety improvements. And this is where centralised cloud systems do make sense.
Autonomous vehicles are just one example of this dilemma. Factories, retailers, operators and producers all face the same challenge as they try to embrace the data-driven future. Any business deploying smart sensors, IoT devices and predictive analytics will encounter similar issues.
Ever-increasing data volumes
The introduction of IoT devices has exponentially increased the volumes of data being generated. Each sensor can output multiple messages every second. Although small in size, each signal needs to be analysed and actioned immediately.
In most cases, sensor output is nothing more than ‘status ok’ type messages and can be safely ignored, and simply sent to archive storage. In fact, it may be perfectly reasonable to discard them entirely as they offer little long-term value.
Without rules that filter and direct this constant stream of information, businesses will see their data capacity requirements – and costs – escalate even faster than anticipated. The right information must be retained however, otherwise the results of your predictive analytics efforts will be unbalanced or incomplete.
The fundamental challenges you face
In order to succeed in a data-driven operating environment, your business needs to adapt to computing at the edge. You will need to address:
- How to provide adequate processing power to deal with incoming data in real time.
- How to specify storage for machine-generated information.
- How to provide sufficient network bandwidth between the edge, data centre and cloud.
With the right infrastructure, these challenges can be overcome. And the benefits of edge computing are significant – you can read more in part 2 of this blog series next week.