With data exponentially outgrowing the capabilities of centralised storage and management, here’s how edge analytics can help organisations overcome this challenge
As the world continues to grow more interconnected, the range of intelligent devices capable of collecting and analysing data is greater than ever. Today’s cars generate large amounts of data from sensors and computers built into their designs. Retail stores collate data on everything from inventory and shipments to customer purchases. The wind turbines that produce our renewable electricity are built with hundreds of sensors and generate millions of data points per minute.
The volume of information required to power these everyday occurrences is growing faster than the available bandwidth of the networks designed to hold and organise it. It’s no longer possible to transfer all the generated data into a central place where it can be organised and analysed via conventional means.
Enter, edge analytics: a type of decentralised data analytics wherein data is analysed at its source – at the ‘edge’ of the information network.
In traditional systems, data is transferred from where it’s being collected to a central repository where it can be analysed. But even today’s most powerful networks don’t have sufficient capacity to transfer all the data generated in most use cases, meaning decisions must be made about what gets left out.
By allowing raw data to be analysed at its source, edge analytics avoid the need to transfer data back to a central system, while still bringing all the insights together for centralised decision-making. This drastically accelerates the speed at which analytics can take place, without compromising on the quality of the results.
>See also: How edge computing will benefit from 5G technology
Why you should adopt edge analytics
Given for the potential value of analytics for improving decision making processes and business outcomes, businesses cannot afford not to explore their options. Edge analytics has a wide range of benefits in terms of analysing more of their data, faster, and potentially for cheaper.
Analytics at the edge when using an in-memory database accelerates these benefits further, as organisations can analyse raw data as it comes in. This gives organisations real-time results, enabling changes and adjustments to be made rapidly.
Edge analytics can also help address one of the most common headaches of today’s digitally-savvy companies: cloud costs. Storing data in the cloud costs money, as does transferring data between the cloud and on-premises storage, or between cloud services providers. Those costs tend to escalate very quickly as cloud use scales up – so by doing more analytics at the edge, companies can reduce spend on cloud storage and transfer costs.
Security is another benefit of edge analytics, especially if organisations are dealing with sensitive data such as personally identifiable information (PII). Having all of a company’s raw data in one central location can be inherently risky. By employing edge analytics, organisations can keep sensitive data where it is and only transfer pre-aggregated data to the central data warehouse, meaning that it doesn’t have to host and protect sensitive information.
Industries on the edge
So: where do these benefits apply? Where can we see companies benefitting from edge analytics?
One of the most important is the renewable energy sector, which has been quick to adopt edge analytics. For example, wind turbines have hundreds of sensors built in to ensure each part of the turbine is functioning properly and can adapt to external conditions. If data analytics can be done at each individual turbine, predictive maintenance is possible, pre-emptively identifying and isolating any problems for each one, minimising the impact on the entire group in the event of an accident.
Another area where edge analytics will gain traction is supply chain management due to its complexity. Supply chains often have hundreds of individual moving parts, including the sourcing and tracking of raw materials which need to be transferred to multiple locations for manufacture, the management of storage facilities, and the monitoring of thousands of IoT devices such as RFID chips that keep track of items being shipped.
In this situation, supply chain managers might have a centralised place where the movement of products around the world is organised, and an ‘edge’ could be the storage facilities with RFID tagged goods. Analysing data directly at those storage facilities in close to real-time with edge analytics is useful for coordinating the rest of the chain. Supply chain managers also need to take into consideration the weather, for example, which is where more advanced analytics like AI and ML modelling might come in that can be crunched at the edge too.
>See also: Digital transformation of supply chains needs online and offline integration
Ready, set, go
Implementing edge analytics doesn’t need to come at the cost of traditional central databases — in fact, they’re best used together. Raw data can be analysed at the edge before being aggregated and sent to a central database or data warehouse for storage and more advanced analytics as needed.
Setting up edge analytics in your business begins with the right in-memory database. To analyse large amounts of data in the edge in real-time, you need a high-performance analytics solution, and one that can integrate infrastructure on the edge and central data warehouse infrastructure. The right database allows you to build a grid consisting of a centralised system and the edge, with the platform acting as a conduit helping direct data where it needs to go.
As the world of IoT continues to expand and organisations begin to add more AI and ML integrations into the edge to automatically optimise real-time data analysis for better and faster outcomes, edge analytics will only become more important to avoid a deluge of data needing to be processed and analysed centrally. Suffice to say, the key to better data analytics means living on the edge.
Jens Graupmann is senior vice-president of product and innovation at Exasol.
Four practices in IoT software development — Vincent Delaroche, CEO and founder of CAST, identifies four key IoT software development practices.
What the PSTN switch off means for IoT — Exploring how the pending Public Switch Telephone Network (PSTN) switch off could impact IoT operations going forward.