OPCUA Driver Performance Testing

Test Purpose

In the scenario where OPCUA driver connects to devices for large-scale data collection and device control, verify the resource usage of NeuronEX, and continuously monitor: CPU, memory, network IO, and device control delay.

Test Architecture

Test Environment and Testing Tools

• Prosys OPC UA Simulation Server ：This software is an advanced simulation tool developed by Prosys OPC, specifically designed for the development, testing, and educational training of OPC UA (OPC Unified Architecture) technology. OPC UA is an open, standardized communication protocol widely used in the fields of industrial automation and smart manufacturing. It supports cross-platform, cross-language data exchange and provides high levels of security and reliability.

• Hardware resources of the Linux machine deployed with NeuronEX:

NeuronEX Version	Operating System	CPU	Memory	CPU Model
NeuronEX 3.2.1	Debian GNU/Linux 12	4 cores	30Gi	Intel(R) Xeon(R) Platinum 8269CY CPU T 3.10GHz

• Monitoring the usage of CPU, memory, network IO, and other resources of NeuronEX software on the Linux machine through Prometheus.

Test Scenarios

Data Collection Scenarios

• Scenario 1

NeuronEX is configured with 1 OPCUA driver, which includes 10 collection groups, each group collecting 1,000 Float type data points per second, totaling 10,000 data points.

• Scenario 2

NeuronEX is configured with 5 OPCUA drivers, each driver including 10 collection groups, each group collecting 1,000 Float type data points per second, totaling 50,000 data points.

• Scenario 3

NeuronEX is configured with 10 OPCUA drivers, each driver including 10 collection groups, each group collecting 1,000 Float type data points per second, totaling 100,000 data points.

• Scenario 4

NeuronEX is configured with 1 OPCUA driver, each driver including 1 collection group, each group collecting 1,000 Float type data points every 100 milliseconds, totaling 1,000 data points.

• Scenario 5

NeuronEX is configured with 5 OPCUA drivers, each driver including 1 collection group, each group collecting 1,000 Float type data points every 100 milliseconds, totaling 5,000 data points.

• Scenario 6

NeuronEX is configured with 10 OPCUA drivers, each driver including 1 collection group, each group collecting 1,000 Float type data points every 100 milliseconds, totaling 10,000 data points.

Device Control Scenarios

• Scenario 7

In NeuronEX configured with 10 OPCUA drivers, each driver including 10 collection groups, each group collecting 1,000 Float type data points per second, totaling 100,000 data points, dispatch 100 data points.

Overview of Results

Data Collection Performance Test

Scenario	Number of Drivers	Number of Groups per Driver	Number of Points per Group	Collection Interval	Total Points	Point Type	Memory Usage	CPU Usage	Network IO
Scenario 1	1	10	1000	1 second	10,000	Float	172MB	7%	receive: 147kb/s transmit: 186kb/s
Scenario 2	5	10	1000	1 second	50,000	Float	323MB	31%	receive: 738kb/s transmit: 934kb/s
Scenario 3	10	10	1000	1 second	100,000	Float	594MB	60%	receive: 1.49mb/s transmit: 1.88mb/s
Scenario 4	1	1	1000	100 ms	1,000	Float	140MB	4%	receive: 148kb/s transmit: 188kb/s
Scenario 5	5	1	1000	100 ms	5,000	Float	162MB	20%	receive: 741kb/s transmit: 939kb/s
Scenario 6	10	1	1000	100 ms	10,000	Float	199MB	36%	receive: 1.49mb/s transmit: 1.88mb/s

Device Control Latency Test

Scenario	Dispatch Method	Number of Points Dispatched	Test Count	Minimum Response Time	Maximum Response Time	Average Response Time
Configured 10 OPCUA drivers in NeuronEX, each driver containing 10 collection groups, each group collecting 1,000 Float type data points per second, totaling 100,000 data points under normal collection conditions.	API Dispatch	100	100 times	6ms	19ms	12ms

TIP

This test used simulator devices, and the data point addresses were all continuous segments, so the system resource usage when NeuronEX collects data from real devices will be higher than the results of this test.

If using NeuronEX data processing functions for data cleaning and filtering, edge computing, and algorithm integration, additional CPU and memory will be consumed.

Detailed Test Results

Scenario 1

NeuronEX is configured with 1 OPCUA driver, which includes 10 collection groups, each group collecting 1,000 Float type data points per second, totaling 10,000 data points.

• Memory Usage ：172MB

• CPU Usage ：7%

• Network IO ：receive:147KB/s; transmit:186KB/s

Scenario 2

NeuronEX is configured with 5 OPCUA drivers, each driver including 10 collection groups, each group collecting 1,000 Float type data points per second, totaling 50,000 data points.

• Memory Usage ：323MB

• CPU Usage ：31%

• Network IO ：receive: 738KB/s; transmit: 934KB/s

Scenario 3

NeuronEX is configured with 10 OPCUA drivers, each driver including 10 collection groups, each group collecting 1,000 Float type data points per second, totaling 100,000 data points.

• Memory Usage ：594MB

• CPU Usage ：60%

• Network IO ：receive: 1.49MB/s; transmit: 1.88MB/s

Scenario 4

NeuronEX is configured with 1 OPCUA driver, each driver including 1 collection group, each group collecting 1,000 Float type data points every 100 milliseconds, totaling 1,000 data points.

• Memory Usage ：140MB

• CPU Usage ：4%

• Network IO ：receive: 148KB/s; transmit: 188KB/s

Scenario 5

NeuronEX is configured with 5 OPCUA drivers, each driver including 1 collection group, each group collecting 1,000 Float type data points every 100 milliseconds, totaling 5,000 data points.

• Memory Usage ：162MB

• CPU Usage ：20%

• Network IO ：receive: 741KB/s; transmit: 939KB/s

Scenario 6

NeuronEX is configured with 10 OPCUA drivers, each driver including 1 collection group, each group collecting 1,000 Float type data points every 100 milliseconds, totaling 10,000 data points.

• Memory Usage ：199MB

• CPU Usage ：36%

• Network IO ： receive: 1.49MB/s; transmit: 1.88MB/s