In the 5G era, edge computing is used to accelerate the development of interconnected manufacturing

In recent years, 5G and the Internet of Things have continued to develop, smart terminal devices have become increasingly popular, and data on the edge of the network has exploded. These factors have greatly promoted the development of edge computing. For the industrial Internet, 5G edge computing technology can solve problems such as data latency, bandwidth, and security, and meet and accelerate overall construction needs. The manufacturing industry is also actively transforming towards intelligence and interconnection, and "interconnected manufacturing" has become a new turning point and historical opportunity in the development wave of manufacturing companies.

However, with the development of intelligent manufacturing, more and more data types have emerged. According to a report by IDC, the growth rate of real-time data is 50% faster than that of static data, and the compound annual growth rate of streaming data analysis is expected to reach 28%. This makes traditional data platforms dedicated to static historical data solutions and running locally or in discrete clouds unable to meet the current needs of manufacturing companies for real-time analysis. The reason why streaming data has grown so rapidly is that it can achieve real-time analysis and, more importantly, autonomous decision-making.

Factors that enable the transformation of traditional manufacturing to connected manufacturing include: economical process sensors tailored for specific purposes, powerful edge computing devices that can make repeatable autonomous decisions, cloud computing for analysis and storage, and upcoming 5G applications. 5G will open a data "highway" to free the manufacturing process from the constraints of connection lines; but these advantages of streaming data also make it more challenging to manage the huge amount of data in the business processes of various manufacturing companies, as well as the diverse data structures.

Traditional connected manufacturing data management solutions face challenges

With the rapid development of new data sources and the increase in data volume, many manufacturing companies are under pressure to address the complexity of digitalization. The main challenges facing enterprises in the management of connected manufacturing data include:

• Data management costs: Traditional data management mechanisms are costly and cannot capture and process PB-level IoT data streams from connected devices. Today, organizations need a more flexible and scalable data management and analysis platform that can easily collect, store, and manage streaming data at a lower cost.
• Handling the volume and variety of IoT data: To enable process monitoring and optimization, predictive maintenance, and emerging IoT applications, information architects need a platform that can handle a wide range of data structures and schemas, from second-by-second temperature, pressure, and vibration readings to pure unstructured data (e.g., images, video, text, spectral data), as well as other forms of data such as thermal imaging, acoustic signals, etc., from the edge through a variety of supported drivers and protocols.
• Managing the complexity of real-time data: To drive continuous process monitoring, yield optimization, or predictive maintenance, a data management platform needs to analyze streaming data in real time and efficiently collect, store, and process this data to provide timely insights and actions.
• Liberate data in silos: Due to the existence of special processes in the value chain (innovation platform, QMS, MES, etc.), different data sources and data management platforms need to be customized for each independent silo solution. Given that cross-enterprise data can only provide a small part of the insight, these narrow point solutions will limit the value of the enterprise; and duplicate silo solutions will divide the business, thereby limiting the opportunity for cooperation. Therefore, the platform must be able to combine streaming data from various points in the value chain with ERP, MES and QMS sources and form actionable insights.

Cloudera DataFlow Gain Insights from the Edge

Given the complexity and diversity of manufacturing and IoT data, manufacturing companies attach great importance to obtaining clear and visible insights from edge to artificial intelligence. Therefore, data should be placed in data lakes and enterprise data platforms from the beginning.

Cloudera Data Platform addresses these challenges through a combination of technologies in Cloudera DataFlow (CDF). CDF provides the following solutions:

• The ability to manage, control, and monitor all data flows and the edge of IoT initiatives. Cloudera Edge Management (CEM) consists of edge agents and edge management centers, which manage, control, and monitor edge agents to collect data from edge devices and send intelligence back.
• The ability to acquire and manage real-time streaming data. Cloudera Flow Management (CFM) is a data acquisition and management solution powered by Apache NiFi that does not require coding. Through NiFi's intuitive graphical interface and more than 300 processors, CFM can provide enterprises with highly scalable data movement, transformation, and management capabilities.
• Advanced messaging and stream processing capabilities through Apache Kafka. Cloudera Stream Processing (CSP) uses Apache Kafka to provide advanced messaging, real-time processing, and analytics of streaming data, as well as management and monitoring capabilities supported by Cloudera Streams Management.
• Real-time insights from Cloudera Streaming Analytics (CSA). Powered by Apache Flink, CSA provides low-latency processing capabilities to perform real-time actionable intelligence on streaming data from the edge.