Authored By : Anandh Venkatraman, Chief Digital Officer, APJ, EMEA, Dell Technologies Consulting
There is no choice over where we will reach based on the road we choose. IOT as a service is one such path that exists for enterprises which are looking for one. After the last three decades of accelerated socio-economic growth with science and technology, there is the next wave of disruption. Internet of Things (IoT) is changing the way we communicate with everyday things. With real, practical products and services gaining importance in our day to day activities starting from fitness trackers to automating homes, offices the wave is already there. 5G connectivity will soon enable new technologies on scale, such as connected vehicles and even connected healthcare. New technologies, new products and new connective potential is also enabling new business models. The next generation of products will need to connect (globally), interact with new partners and ecosystems, and may be available in as-a-service formats (made possible through real-time remote connectivity).
The lack of in-house expertise in an enterprise to manage these disparate layers is among the top reasons that have constrained the potential growth of IoT. With numerous protocols and standards out in the market, understanding interoperability of these disparate layers is what adds the most complication.
The business forecast of IoT market is already impactful at a global level. For the next, five to ten years Gartner estimates a benefit of USD 2 trillion in 2020, IDC Estimates USD 1.7 trillion in 2020, and McKinsey estimates growth to go around USD 4 trillion to USD 11 trillion. The clear statistic is the there is a growth and curve is steep. For catering this forecasted growth, Gartner notes that IoT and the business models associated are not enterprise ready.
The platform of platforms approach towards designing, development, deployment, and operationalization of various IoT applications and business processes would be a solution for this.
Value Proposition
The Internet of Things has come a long way, from hacks to research projects to efficient, sustainable everyday products, services and applications capturing customer’s imagination and driving a technological revolution. While the industry is trying to realize the full benefits of this ecosystem, the enterprises should address the challenges faces by sensor hardware providers, big data platforms, service integrators, application developers and end users.
Platform of platforms
There is a need to have a holistic solution, reference architecture to cater most of the enterprise IoT needs. This solution while handling most of the IoT solutions should also be flexible enough for handling cross industry applications, hence a platform of platforms. Our approach on the solution to this problem for applications in capturing from several enterprises and telemetry data sources, processing of petabytes of byte-sized datasets originating from various sensors, advanced machine learning system for recommending actions based on real-time analysis and connecting the end user business systems
Reference Architecture
Over the last few years, we have worked on developing several IoT ecosystems in energy, traffic, healthcare, and automotive. While designing, developing and creating market-ready architecture, we have developed a cohesive IoT PaaS solution keeping in mind rapid design, seamless integration and scalable deployment as the principle. The application pretext for this platform to have capabilities for data streaming, smooth ingestion, event processing, and analytics, elastic hardware and software computing, machine learning compatibility and end-user application integrations.
Various Components of a generic IoT System would be:
Sensor Communications and Connectivity
The platform must support multi-protocol communications and connectivity support. Sensors are connected to the primary stream systems by either a direct internet connection if sensors are wifi enabled. Otherwise, a gateway like Dell Gateway 5100 acts like an edge server that communicates with sensors in various protocols. Edge nodes are responsible for the connections to any sensors, actuators, control systems and assets. For controlling the sensors too, the gateway acts as an interface. Hence a reliable and stable portal is the first touch point for this architecture.
Edge Data Manager
The primary role of gateway software after connecting with sensors is to tag them through a sensor management layer. Registering of the sensors, creating specific IDs, and connection management is the core component of the edge sensor manager. Once the sensors are designated, the data from sensors will be normalized to a standard format like MQTT. Either the data in the sensors is stored for edge analytics or will be exposed as JDBC connections or HTTPS based REST calls for further processing.
Edge Analytics
Analytics on edge systems becomes critical where the business-critical information needs information in low latency, for example, video analysis systems that need to do the analytics on real-time or healthcare systems that require to make calculations on the gateway. Hence, edge gateways also have an analytics component like EdgeXFoundry deployed to serve the edge analysis component.
Platform Data Manager
Platform Data Manager is the data center component that acts as a backend platform service to handle: – Data Ingestion: Ingestion manager caters to both real-time and batch-processing needs of the sensor data, in the Lambda architecture, – Data Pre-processing: Standardization, Normalization, Verification, of the the data are done as part of the platform ingestion – Data Aggregations: Raw data is converted into aggregated data marts for analysis and analytics at data manager
Platform Analytics Manager
Analytics Manager works on the information there in the data lake ingested and pre-processed through the data manager, and various analytics performed over the data in data lake like:
– Correlation Analysis: E.g. how to change the ambiance of the room according to the weather outside
– Forecasting: E.g. Energy forecasting – Predictive Analysis
– Prescriptive Analysis: E.g. Automatic Demand Response Systems
Application Manager and Business Layer
From the data after analysis and analytics, the platform needs to support business needs with various end-user applications using these could be plugging using PaaS systems like Pivotal Cloud Foundry.
Security Layer
At various system integrations security and encryption of data is very prominent, hence different touch points of data across the platform, a security protocol is enabled and logged for the understanding of further lineage and audit.
End to End IoT Solution
New IT revolution is underway, over the last few years, there have been advancements on scale-out architecture, distributed systems, cloud platforms and the end to end big data platforms and the next would be an IoT platform designed for enterprises.
This platform would need to comprise various paradigms for software advancements and cater to business functions and applications. Such platforms will have the capabilities from sourcing the sensor information, realtime ingestion in a distributed system-based hardware and software, data processing and normalization of the data, data science and insights, and lastly end user applications running on the platform as a service.
Conclusion
Internet of Things, distributed systems, advanced data engineering, machine learning and advancement in application development is changing the enterprise ecosystem. There is machine generated, hence well-structured new petabytes of data that is arriving every second, machine runs. Applications on this data require a historical understanding of this data and real-time analytics for determining patterns and recommendations. This ecosystem hence demands an enterprise-ready end to end platform, covering systems from edge to on-premise systems to cloud application systems.
