Liquid Platforms to Develop IoT Analytics

With a forecast of 30 billion Internet-connected IoT devices by 2020, the Internet of Things represents an immense amount of data never seen before, and today’s systems make it difficult to analyse such a quantity of information. Combining with the public cloud allows this IoT data analysis to be addressed in terms of both volume and speed so that once data has been analysed, it can re-enter the business value chain and enhance the information.

IoT poses real technological challenges, especially in industry, where thousands of sensors monitor real-time status and are stored every day, which is why IoT is also a huge and complex data source providing much value. In this environment, the key is how to take advantage of that data, transform it into value and leverage it to create new business models.

All IoT analysis should be oriented to generate business value. For this purpose, we propose the simplified IoT Analytics process in four phases: information intake, transition to back office platforms, the events themselves and their analysis.

In this second phase of transition to platforms, difficult connectivity situations in industrial environments can prevent this transition. To solve this limitation, Edge Computing comes into play, which is the capacity to compute, analyse and process information as close as possible to where it is produced, which greatly facilitates the development of IoT environments.

Platforms that adapt to change

30 billion devices by 2020, 40 billion by 2021, 50 billion by 2022, according to Jupiter Research reports, and so on and so forth. IoT is a sector that is both growing and yet immature, something that has been caused by the changing speed of technology in terms of devices, connectivity and analysis, which is so great that to propose global strategies is very complex since it is not permanent in time.

This is why IoT’s liquid or flexible analysis platforms, based on the public cloud and aligned with each company’s cloud strategy, where enabling capabilities are deployed, are the best way of tackling this type of project. One example is the case of big data intake and processing capacity. In this respect, the public cloud is the best platform for large volume because it allows flexible management of volume, variety of data types and complete governance throughout the process.

Data enrichment is also essential in IoT analysis. On many occasions, IoT data by itself is not useful for analysis, it has to be enriched with business data. Mixing business and sensor information is essential to get valuable analytics.

Another key aspect of developing an IoT analytics platform is training analysts to work with the information. Traditionally, creation of these platforms has been very close to control systems with little analytical capability. In cloud platforms, the platforms must allow analysts the environments in which to explore information, both historically and with high granularity in real time. In order to propose liquid solutions that will enable evolution, data science tools that the public cloud can provide are necessary.

As a final feature, in many cases the liquid platform seeks to return information to its own control systems. There will be algorithms and models that should be deployed quickly; a classic example is failure prediction. This can be achieved by running models in the cloud but, sometimes it is not possible for various reasons, but it may make sense to run the models on Edge, ie situate a device that is capable of computing the model near to where the data is produced, connect locally and run the models.

With this approach, the option with more prospect of success for companies is to build a platform for IoT analysis to adapt to the changing needs that will inevitably occur. It is possible that certain IoT use cases planned on the basis of a global cloud strategy will fall by the wayside, or new case requirements arise that need to be accelerated and priorities rearranged. A liquid, flexible platform, leveraged on public cloud technologies, is the best option so that the changing and dizzying pace of technology does not lead us to create obsolete functionalities at the time of going into production and thus we can devote our time to what is really important: finding solutions of value for business (the ‘what’) rather than spending time on the technology itself (the ‘how’) when it is impossible to compete with the innovative capabilities of public platforms.

Image: pixbay