Edge Computing Needs an Active Archive
By Bhupinder Bhullar, CEO of Swiss Vault
Many data storage architectures are centralized due to the economies of scale they offer, as well as a greater ability to control costs. However, data decentralization has become inevitable following the rise of edge computing. Put simply, it isn’t feasible to offer high performance at the edge when large amounts of data are transmitted to a central data center. Thus, expect to see the emergence of edge nodes that can both store and process edge data.
Edge, in essence, brings computation closer to the data source. It reduces latency and bandwidth requirements while enabling more rapid decision making. Future edge systems will be required to collect, store, and process data from a wide array of sensors and systems. These systems include chemical, water, image, light, optical, motion, pressure, humidity, proximity, and temperature readings, among others, in use cases such as agriculture and healthcare.
Agriculture is going high tech. Farms now measure temperature, humidity, rainfall, and other parameters constantly both outdoors and in indoor facilities such as greenhouses. They keep track of soil conditions and plant conditions in order to optimize output and productivity. The array of equipment now at their disposal includes shade canopies, dosing and irrigation controls, lighting systems, passive/active cooling and heating, harvesting and spraying equipment, foggers, humidifiers, and a wide range of physical and virtual security systems. These systems must be able to serve up data to mobile devices and websites and send text alerts. They must host historical data for trending and analysis. They must monitor energy usage, supplier data, packaging data, crop pricing, and labor schedules. And they must be capable of analyzing large amounts of data rapidly to predict necessary changes. Systems are emerging that harness edge computing to provide real-time monitoring and analysis of conditions like solar conditions, weather patterns, or crop health. These systems allow farmers to access data-driven insights in remote areas.
In healthcare, data also must be collected in different ways. Wearable devices, imaging systems, diagnostics, and other applications are emerging that monitor and dictate actions based on an array of readings for blood pressure, body temperature, pulse, oxygen levels, heart rate, and more. Edge data will soon revolutionize healthcare by bringing data processing and decision-making closer to the point of care. Smart wound bandages, for example, detect a number of biomarkers and conditions at the wound site so that you don't have to remove the bandages to check underlying conditions. Nurses can leave them on, and the patient doesn't have to come in unless conditions dictate the need for a visit or an adjustment of the bandage. Instead, everything is remotely monitored, and actions are only triggered when attention is required.
Modern edge systems must be designed for efficient collection from these applications, systems, and sensors; they must be able to cope with high-frequency data collection and must be able to process and store data in a way that best serves the healthcare community.
The Edge Nodes of the Future
Edge nodes must be capable of conducting data filtering and pre-processing, such as noise reduction and data cleaning. That entails low-power GPUs that can support AI applications and run algorithms at the edge. After all, the edge is where a great deal of object recognition, anomaly detection. and predictive maintenance will be performed. There is no time to transmit all that data to a central repository to await instructions when the harvest may fail, a life is in danger, or a car traveling at 60 on the freeway is approaching an area of gridlock.
The edge nodes of the future must be able to run machine learning algorithms and real-time analytics to drive immediate insights. Beyond that, the edge must be optimized for access and long-term retention.
One way to achieve this is to distribute data across many edge data storage nodes. Using a geo-distribution algorithm for data fragmentation, data is stored on disk for easier accessibility and analysis while being transmitted to central servers for further analysis. Lastly, edge data must be encrypted and secure at rest and in transit.
You can find out more about emerging use cases for active archives by listening to the recordings of the 2023 Active Archive Virtual Conference.