Over the years of delivering IoT infrastructure software for enterprises, we have found that enterprises mostly build single IoT applications in a scenario-first manner. However, with the gradual maturity of IoT technology, the data generated by massive devices in the IoT era provides a richer data base for enterprise digitization. As the data generation subjects shift from human behavior to device state, the IoT era will face more real-time stream data, and its volume will expand from TB level in the mobile Internet era to PB or even EB level.
There is no doubt that enterprises should rethink the logic of IoT application and business. With "data" as the center, on the basis of merging IoT data and traditional TP data, they can flexibly build diversified and innovative business to achieve digitalization, real-time and intelligent transformation.
According to the paradigm and the “IoT-Oriented” design principle, EMQ provides a portfolio based on a new paradigm of data infrastructure architecture - Data Infrastructure for IoT, including EMQX, a cloud-native distributed MQTT broker, HStreamDB, a cloud-native streaming database, eKuiper, an edge lightweight IoT data streaming/analytics engine and Neuron, an industrial protocol gateway software for IoT edge, which helps complete the data chain from sensing and collection to information conversion.
- Understand the key trends of digital transformation and the paradigm shift in data processing in the IoT era
- Understand the characteristics of IoT-oriented data infrastructure architecture
- Learn how to unify the "connecting, moving, processing and analyzing" of IoT data through a portfolio of products under the Data Infrastructure for IoT architecture paradigm
- Digital Transformation Trends in the IoT Era
- Edge Computing
- The Paradigm Shift in IoT Data Processing
- Data generation: from "human activities" to "connected things"
- Data form: from transactional data to streaming data
- Data volume: from TB era to PB/EB era
- Data processing: from batch to stream processing
- Data analytics: from Schema-on-Write to Schema-on-Read
- Towards a unified ontology namespace architecture
- Move computation prior to data
- Datastream reusability
- Edge-cloud orchestration
- Full-scenario adaptability
- Unified Namespace - New primitive for IoT architecture
- Seamless data convergence
- Data Model to Dataflow
- Data interoperability
- Exclusive Data source, Multiple user cases
- Modern Data Infrastructure for the IoT
- Architecture Paradigm
- EMQ’s Portfolio for Data Infrastructure for IoT
Now it's time to think about the business challenges and opportunities for enterprises in the IoT era.
Over the past decade, EMQ has been developing open-source infrastructure software for IoT applications and solutions. We have delivered EMQX, the world's leading open-source MQTT message broker that solves the challenges of massive device connectivity. We also created HStreamDB streaming database for the storage, processing, and real-time analysis of IoT data.
After years of delivering basic IoT software for enterprises, we have found that most enterprises build IoT solutions with an application-centric approach.
Today, we believe that the core logic of IoT scenarios should be data-centric and obtain business insight from data to create value. In particular, medium and large enterprises should focus on data and think about IoT business from the perspective of data, building diverse and innovative IoT solutions based on a unified data infrastructure.
In this white paper, EMQ formally proposes the architecture paradigm of "Data Infrastructure for IoT" to meet the key business challenges of enterprises in the IoT era together with industry to achieve business innovation and value creation.
Digital Transformation Trends in the IoT Era
Cloud-native architecture and technologies are an approach to designing, constructing, and operating workloads built in the cloud and taking full advantage of the cloud computing model.
Cloud-native features extreme elasticity, service autonomy, fault self-healing, and replication at scale.
Cloud-native systems embrace microservices to achieve speed and agility.
Cloud-native architecture is a design methodology that utilizes cloud services to allow dynamic and agile application development techniques that take a modular approach to build, run, and update the software through a suite of cloud-based microservices versus a monolithic application infrastructure. The microservice architecture best reflects the philosophy of cloud-native.
The classic cloud computing model uploads all the data that needs to be calculated and stored to the cloud data center and uses the cloud data center's supercomputing power to meet the application's computing needs in a centralized way. However, in the era of the Internet of Everything, the centralized processing model of classic cloud computing has three shortcomings:
- Real-time requirements and network coverage: As the number of edge devices increases, the amount of data generated by the devices continues to surge, causing network bandwidth to become the bottleneck of the classic cloud computing model gradually. The high cost of ensuring full-scenario network coverage also forces companies to decentralize part of their computing to the edge.
- Data security and privacy: With the popularization of data collection equipment, directly uploading the collected data to the cloud data center will increase the risk of leaking the core data assets of the enterprise. In addition, managing customer information on the cloud has also brought about personal privacy leakage. As a result, more and more companies are putting privacy-related data on edge for analysis.
- Energy consumption: As cloud servers run more and more applications, it won't be easy to meet the energy consumption requirements of large-scale data centers in the future. Improving the efficiency of energy consumption under the framework of the classic cloud computing model does not solve the fundamental problem by itself, and the issues in the era of the Internet of Everything will become more prominent.
In the edge computing model, the edge device has the processing capacity to perform calculations and data analysis. As a result, part or all of the computing tasks performed by the classic cloud computing model are migrated to the edge device, reducing the computing load of the cloud server, reducing the pressure on the network bandwidth, and improving data processing efficiency in the Internet of Everything era.
Whether cloud-native or edge computing is essentially thinking about software architecture from a technical perspective. Cloud-native brings many benefits, but it also brings new challenges. Edge computing has solved some of these challenges and has become the second force in designing IoT software architecture. Now, it is time to return to the essential needs of the business and think about what software architecture the IoT system needs itself.
In much frequent communication with customers, EMQ found that whether it is on the cloud/edge/end, the software architecture designed for the IoT system is quite different from other web applications and embedded systems due to the following characteristics:
Integrated system with hardware, software, sensors, connectors, and gateways.
Various interoperable communication protocols.
Support for data volume and velocity.
Support for multiple QoS of communication.
Live data stream analytics with complex event processing and analysis.
The traditional big data ecosystem cannot meet the above requirements. Therefore, IoT-oriented system architecture requires modern IoT data infrastructure software.