Photo by Jonas Morgner on Unsplash
Introduction
Recently, there’s been a wave of manufacturing innovation around the world, with smart factories at the heart of it. IoT and IIoT are gradually making their way into our lives. Combined with other technologies of the Fourth Industrial Revolution, such as big data, artificial intelligence, and the cloud, they are increasingly influencing various sectors, such as smart homes, smart buildings, smart factories, and smart cities. In response, more and more technologies and solutions are emerging, and new markets are being created, especially by combining time series databases.
About Smart Factory
A smart factory is a production system in which information and communication technology (ICT) and traditional manufacturing technology, production and manufacturing technology are converged, and equipment and device parts in the factory are connected and interact with each other through technologies such as IoT, big data, cloud computing, intelligent robots, and cyber-physical systems (CPS).
An intelligent factory is different from simple factory automation. Factory automation is automated and optimized on a per-process basis, with no connection between processes. The key difference is the connection of data across all processes and assets in the factory. In addition, by fusing sensor data with production and management data such as ERP and MES, management decisions can be reflected directly to the production equipment in the factory.
Smart factories conserve resources by using energy and materials optimized for factory operations, improve asset utilisation through predictive maintenance of equipment failures, and enable factory operations that are automatically linked to production and management plans. However, handling the large amounts of data generated by the numerous devices and sensors installed in modern factories is a problem. This article presents recent trends and related technologies from a data storage perspective for storing tag data generated in smart factories.
Data flow in smart factories
Implementation techniques
Connectivity
Various sensors, such as temperature, pressure, and vibration, are attached to manufacturing equipment, and data is collected in real-time through the application of IoT technologies. In existing manufacturing facilities, there are various sensors, actuators, and PLC devices, and the communication protocols of the data generated by these devices are also diverse, such as Wifi, Zigbee, and BLE. Standardization of these communication methods, such as OPC UA, has recently begun.
OT(Operation Technology)
It is the primary collection and monitoring of data from a manufacturing facility by personnel working on the floor of the facility. It leverages historians or real-time databases (RTDBs) to collect and store data in real-time, monitor conditions in real-time, and alert personnel when something is out of normal range so they can take action.
IT(Information Technology)
IT departments use business systems such as MES, PLM, and ERP to store various data and perform big data analysis. Since the data is collected and analyzed not from individual pieces of equipment, but from the entire plant, factory, or even the entire company, the amount of data collected is huge and requires large-scale processing technologies such as Oracle Exadata or Hadoop. We also use professional analytical tools such as R and SAS to perform correlation and multi-dimensional analysis. Advanced analytics are also performed through machine learning.
Need for new solutions
High-speed data capture and storage
As the speed and volume of data to be collected increases, a data storage solution is required that can collect and store data at ultra-high speeds of more than 1 million tags per second.
In addition, a solution that can compress and store data is required for system resource efficiency and storage savings.
Scalability
Most systems select hardware based on preliminary calculations of CPU, memory, and storage to meet business requirements prior to deployment. However, due to changes in the working environment, there are situations where more hardware resources are required due to deviations from the initial prediction. In particular, in a smart factory, the increase in the number of sensors due to facility expansion or the increase in data due to a change in scan rate can be very large, so it is necessary to be able to expand on demand rather than prepare in advance. In other words, it is possible to configure a multi-node cluster to respond flexibly to changes.
High Availability (HA) Service
Production facilities operating in most manufacturing sites are always running 24 hours a day, 365 days a year, and tag data generated by them must be collected and stored 24 hours a day. This requires coping with server failures and requires HA configuration.
Machbase, the Time Series DBMS for the Smart Factory
ㆍ Machbase is an ultra-fast time series DBMS that ingests, compresses, and stores 5.7 million tags per second in real-time. It also creates indexes in real-time, allowing searches to be performed simultaneously with data entry.
ㆍ The Enterprise version is configured as a multi-node cluster, enabling distributed storage and retrieval, and a scale-out configuration that can respond to data growth by adding nodes.
ㆍ It also has a replica in an active standby configuration, enabling not only HA but also failover in the event of a failure. It also has the advantage of providing the interface of a traditional RDBMS, so existing DB engineers can easily use it without additional learning.
A smart factory is a factory that collects and analyses data in real-time by installing sensors on equipment and machines in the factory, monitors all situations in the factory at a glance, analyses them, and controls itself according to purpose.
As the number of sensors and the amount of data has increased in recent years, the need for a solution that could quickly process and scale data collection and storage has become necessary. The Machbase database is a suitable repository for storing and retrieving tag data generated by equipment in smart factories.
Conclusion
As the number of sensors installed in smart factories has grown rapidly, so has the amount of data generated by them, which is difficult to store effectively with existing RDBMS relational databases, so many companies have recently moved to time series DBs. As a platform for storing IoT sensor data, the presence of time-series DBs will continue to grow.
We recommend taking a look at the USE CASE using the Machbase time series database.