Whilst the digital environment has been expanding rapidly for many years, the pandemic ushered in, by necessity, a degree of digital transformation that is unprecedented in both its scale and scope. With organisations throughout private and public sectors alike forced to roll out digital systems, there has been a sharp uptake in the adoption of connected technologies.

As the Internet of Things (IoT) landscape experiences large-scale growth - from automated supply chains to help maintain social distancing, to more efficient and convenient smart cities and vehicles - the amount of data produced grows rapidly, as well. It is estimated that by 2025, connected IoT devices will generate 73.1 zettabytes of data.

Not only does this data need to be captured, it also needs to be stored, accessed and transformed into valuable insights. This process requires a comprehensive data architecture that can accommodate the demands of a large range of use applications throughout the data journey.

WHAT IS THE IOT DATA JOURNEY?
The vast majority of IoT data is stored in the cloud where high-capacity drives - now reaching 20TB - store massive amounts of data for big data and fast data workloads. These could include genomic research, batch analytics, predictive modelling, and supply chain optimisation.

For some use cases, data then migrates to the edge, where it is often cached in distributed, edge servers for real-time applications such as autonomous vehicles, cloud gaming, manufacturing robotics, and 4K/8K video streaming.

Finally, we reach the endpoints, where data is generated by connected machines, smart devices, and wearables. The key aim here is to reduce network latencies and increase throughput between these layers (cloud-to-endpoints and endpoints-to-cloud) for data-intensive use cases. A potential solution could be 5G, by using millimetre wave (mmWave) bands between 20-100 GHz to create "data superhighways" for latency and bandwidth-sensitive innovations.

WHAT IS THE VALUE OF YOUR IOT DATA?
Data infrastructure is critical in our digital world as data must be stored and analysed quickly, efficiently, and securely. Thus, data architectures need to go beyond simple data capture and storage to data transformation and creating business value, in a 'value creation' approach. Examples include:

• Autonomous vehicles - These vehicles are loaded with sensors, cameras, LIDAR, radar, and other devices generating so much data that it is estimated it will reach 2 terabytes per day. That data is used to inform real-time driving decisions using technologies such as 3D-mapping, advanced driver assistance systems (ADAS), over-the-air (OTA) updates, and vehicle-to-everything (V2X) communication. In addition, IoT data creates value in personalised infotainment and in-vehicle services that improve the passenger experience. In order to enable real-time decision making, which is crucial for passenger safety, the priority for this data architecture is reducing network latencies, along with enabling heavy throughput to facilitate predictive maintenance.
• Medical wearables - It has been predicted that in 2021, worldwide end-user spending on wearable devices will total US$81.5 billion. These devices generate important data to track sleep patterns, measure daily movements, and identify nutrition and blood oxygen levels. This IoT data can be transformed into daily, monthly, and yearly trends that can identify opportunities to improve health habits using data-informed decisions. Such data could also create more personalised and proactive treatments, especially as telehealth and remote healthcare continue to progress, even after the pandemic subsides. Here, the storage priority for data architecture is offering long-term retention for critical health records.

In addition, the following IoT applications provide key examples as to why storage considerations vary according to each specific case, and how the requirements can be met.

• Search-and-rescue drones - This is a key example of an IoT use case which requires a very specific data storage solution to get maximum value from the application. Such drones are often required to operate in harsh natural environments with extreme temperatures and weather patterns. Therefore, the storage solutions used in these technologies must be especially durable and resilient, such as the high-endurance and highly reliable industrial-grade e.MMC and UFS embedded flash drives. Search-and-rescue drones are also commonly used in combination as part of a wider network, utilising optimised routes and shared automated missions. This means that the data architecture must be scalable, enabling the operation of multiple technologies in conjunction with extreme efficiency, performance, and durability.
• Smart cities - For smart cities to function, they require the storage of huge amounts of both archived and real-time data. In order to analyse and act on real-time data, IoT technologies are relying on storage at the edge and endpoints. For example, smart public transport systems require real-time data on traffic, in order to quickly and accurately adjust to spikes in demand, such as rush-hour traffic. This means that, similar to smart cars, this application requires data storage that facilitates low network latencies.

The storage for archival data, in comparison, requires less of a focus on real-time rapid transfer, instead prioritising long-term retention. Here, cloud solutions come into play. Intelligent carbon mapping tools enable another IoT use case which relies on historical data of carbon emissions in order to identify trends and deploy carbon reduction measures.

GENERAL-PURPOSE TO PURPOSE-BUILT ARCHITECTURE
Various connected technologies have different requirements when it comes to how data must be stored in the most appropriate way and how to get the best value from it. For example, NVMe storage solutions are ideal for use cases that require very high performance and low latency in the data journey. Specialised storage is therefore necessary in order to create optimum value from IoT data, which must be considered when building out the wider data infrastructure.

Many businesses, however, still use general-purpose architecture to manage their IoT data. This architecture does not fully meet the varying needs of IoT applications and workloads for consumers and enterprises.

For example, whilst search and rescue drones prioritise endurance and resilience, storage solutions in digital healthcare applications must focus on offering long-term retention and security for critical health records. Therefore, there must be a move from general-purpose storage, to purpose-built data storage and different solutions for different needs.

For any data architecture, the goal is to maximise the value of data. For real-time IoT use cases, your storage strategy has to be designed specifically for IoT, and address the following considerations:

1. Accessibility: what is its serviceability, connectivity and maintenance?
2. Wear endurance: It is WRITE-intensive or READ-intensive?
3. Storage requirements: what data and how much needs to be processed, analysed and saved at the endpoints, at the edge, and in the cloud?
4. Environment: what is the altitude, temperature, humidity and vibration levels of the environment in which data will be captured and kept?

SPECIALISATION FOR OPTIMISATION
Taking optimal advantage of the evolving IoT data landscape means using specialised storage solutions to bring unique business value. It is no longer sufficient to rely on standard, 'one size fits all' storage solutions, when the requirements for different IoT applications vary so drastically. The deployment of innovative and specific data storage solutions will help businesses and enterprises to navigate the accelerating journey of the IoT landscape, and will ensure that the value of data isn't lost unnecessarily in the process.

More info: www.westerndigital.com