Author | Li Jun, Sanheyi Technology
Company introduction
Anhui Sanhei Information Technology Co., LTD. (hereinafter referred to as Sanhei Technology), specializing in big data industry applications and industrial Internet solutions, is committed to work with customers from all walks of life to jointly discover new industrial value. At present, the 3H1 high-end equipment operation and maintenance service platform developed by Sanheyi Technology has been successfully applied in high-end equipment manufacturing, automobile manufacturing, environmental protection equipment, color sorting machinery, cement industry and other fields.
The business scenario
High-end forming equipment is a strategic pillar industry of the country, which is used in automobile, petrochemical, aviation, aerospace, military industry, construction machinery, household appliances and other important fields of national economic development, and is the foundation of many major projects. At present, with the rapid development of the new generation of information technology, the high-end forming equipment manufacturing industry is in an important stage of development from digitalization and networking to intelligence.
As a high-end equipment operational service platform, 3 h1 of the underlying Internet database to support hundreds of enterprises, hundreds of thousands of equipment access, after have always been traded using the open source InfluxDB, reason is on the basis of the stand-alone version can extend the multi-instance depots architecture, but they are also gradually exposed some disadvantages in use process, such as high cost, difficult maintenance, hardware It’s not easy to scale horizontally. Fortunately, I later encountered TDengine, a 10x high-performance database. After repeated tests, its various indicators meet business needs, and it has been used until now.
Why TDengine?
In the iot scenario of equipment industry, there is a huge amount of real-time data, including temperature, pressure, vibration, displacement and many other parameters. It is difficult to analyze and give early warning for these parameters. These requirements are summarized as follows:
-
High concurrency data write, each record need to be timestamp;
-
Different sensor devices need to record different data fields. It is hoped to build separate tables for different devices.
-
The original data must be stored for more than five years, and data compression must be supported to reduce data storage costs.
-
Support localization, support database vendor service fast response.
TDengine community version 2.2.1.1 was selected for distributed simulation test, and 3 servers with the following configurations were used:
Test 1: Verify the writing performance of sequential data on three database nodes of the sequential database product
Simulate the data of 10 workshops in 2 factories, 1000 monitoring points in each workshop, each monitoring point from 2017-07-14 10:40:00.000 start to write simulation data, record timestamp interval 0.001 seconds, each measuring point write 500000 records.
8 thread writes, stopping the writing program after writing more than 5 billion records. The test wrote 5 billion data records at an average write speed of 1.91 million records per second.
On the basis of test 1, check the actual file sizes of three database nodes as follows:
The size of all files is 36GB after the disk is dropped.
The original data size is 5000000000*20/1024/1024/1024=93.13GB,
The compression ratio is 36/93.13=38.65%.
Test 3: Time series Database product Performance of three database nodes to retrieve historical time series data by time
Select any measurement point randomly and query the historical data of this measurement point in a certain period, for example, 10001 data records (data output to file) from 2017-07-14 10:40:00.000 to 2017-07-14 10:40:10.000 10s.
The corresponding query statement in the database is:
Select * from d0 WHERE ts >= '2017-07-14 10:40:10.000' and ts <= '2017-07-14 10:40:10.000' >> /dev/null;Copy the code
TDengine has been proved to have high write performance, high concurrency and very short query delay. The overall cluster adopts distributed architecture, and the reliability, stability and data integrity meet project requirements.
After the selection result was confirmed, we immediately upgraded the original business system and formally introduced TDengine.
TDengine in 3H1 landing practice
3H1 High-end equipment operation and maintenance service platform focuses on solving the key problems of high-end forming equipment enterprises’ transformation from manufacturing to service, and provides enterprises with the overall solution of industrial Internet and intelligent operation and maintenance.
The overall architecture of the platform is shown in Figure 1. TDengine is connected with the intelligent data acquisition terminal module of high-end forming equipment to assist the acquisition terminal to complete the collection of equipment operation data and provide the equipment data foundation for the system. The industrial cloud computing service platform provides storage, conversion and analysis of system data, and provides business data support for the system. The intelligent operation and maintenance service system is composed of the equipment intelligent operation and maintenance service platform and intelligent operation and maintenance service APP, providing system and mobile service support for enterprise personnel respectively.
Figure 1 Overall architecture of the platform
The system application service is divided into six functional modules for enterprise application scenarios.
(1) Enterprise cockpit: it mainly serves the managers of equipment manufacturing enterprises, which is convenient to understand the platform data and key business process indicators. From the overall interface, it is very convenient to know the sales of equipment, enterprise access information and platform data collection. At the same time, some key business processes can be tracked and managed, including enterprise equipment monitoring, alarm information display, maintenance efficiency management, equipment failure and three-packet task information, as shown in Figure 2.
Figure 2 Enterprise cockpit
(2) equipment resource management: the main purpose is to give each a high-end forming equipment set up electronic archives, in order to understand the equipment history, current situation, optimize the equipment operation, to predict the future equipment, check the specific device information mainly display equipment four dimensions of the current condition, health analysis, maintenance, and historical conditions.
The current working condition shown in Figure 3 is convenient for users to understand the basic information, key indicators and alarm situation of the equipment, and can also provide an overview of the current situation of the equipment. Health analysis is shown in Figure 4. The purpose of health analysis is to enable device users to have a deeper understanding of the current status of the device and the changes in the health status of the device over time. If the device is facing a fault risk, the fault cause and fault module can be found quickly. The maintenance situation gives the user an overview of the equipment maintenance information and the flow tracking of the current maintenance task, as shown in Figure 5. Historical working conditions are used for pre-troubleshooting of faulty modules, as shown in FIG. 6.
Figure 3 Device Resource Management – Current operating condition
Figure 4 Device Resource Management – Health analysis
Figure 5 Equipment Resource Management – Maintenance status
Figure 6 Device Resource Management – Historical operating conditions
(3) Maintenance service management: mainly provide the current and historical efficiency analysis of maintenance tasks for the personnel of the maintenance service department. Maintenance tasks display the current number of tasks to be handled, such as orders to be received, orders to be sent and return visit. At the same time, you can also view and operate each maintenance task, which is specific to each link of the maintenance process, as shown in Figure 7.
Maintenance efficiency analysis is a statistical analysis of key efficiency indicators in maintenance, changes in order quantity in the past year, changes in maintenance response time, distribution of fault types, and task statistics of maintenance personnel, so as to facilitate maintenance management personnel to manage maintenance services and efficiency, as shown in FIG. 8.
Figure 7 Maintenance Service Management – Current maintenance tasks
FIG. 8 Maintenance service Management – Maintenance efficiency analysis
(4) Equipment health analysis: By analyzing the history and current equipment information of the equipment, the possible future failures of the equipment can be predicted, and the possibility and type of the failure can be given, so that the maintenance department can specify maintenance strategies for users and actively contact users, as shown in Figure 9.
Figure 9 Device health analysis
(5) Sanbao Service management: Serve the Sanbao department, provide current maintenance activity reminder, equipment maintenance activity record, equipment maintenance expiration warning and other functions.
(6) Spare parts management: Spare parts management shall file all spare parts related to maintenance. Users and personnel from relevant departments can query, apply and approve spare parts on the mobile terminal and the system terminal, reducing unnecessary workflow and improving maintenance efficiency. At the same time, the demand for spare parts can be predicted through data analysis to ensure the demand and reduce the inventory cost of enterprises.
After the application of TDengine, these six functional modules in the use of the effect has also been significantly improved, not only reflected in the data writing, query performance, but also reflected in the efficient compression efficiency, really achieve the optimal balance between performance and cost.
The future planning
At present, with TDengine, 3H1’s original business system has been greatly improved after upgrading, which not only reduces the cost of r&d and maintenance, but also achieves horizontal expansion. TDengine’s excellent query performance gives us a great surprise, high compression efficiency, but also to save a lot of storage resources. In the future, we will try to apply TDengine in more scenarios and strengthen in-depth cooperation with TDengine.