HUAWEI CLOUD cloud server evaluation｜Performance test

foreword

In today's digital age, cloud computing, as an efficient and flexible computing method, is attracting more and more attention from enterprises and individual users. As one of the leading cloud service providers in China, HUAWEI CLOUD is constantly committed to providing users with stable and reliable cloud server services.
In this article, we will conduct a comprehensive performance test on HUAWEI CLOUD cloud server to explore its performance and performance advantages in various indicators. Performance testing is an important means to evaluate the service quality of a cloud server. It can help users understand key indicators such as server processing power, network speed, disk read and write speed, and stability.
Through this performance test, we will conduct a comprehensive evaluation of the HUAWEI CLOUD cloud server in terms of multi-core CPU performance, memory performance, network transmission speed, and disk IO performance. At the same time, we will also compare with other cloud service providers to better understand the competitive position of HUAWEI CLOUD cloud servers in the market.
We believe that through in-depth testing and analysis of the performance of HUAWEI CLOUD cloud servers, we can provide references for enterprises and individual users and help them make informed decisions when choosing cloud server services.

Software Installation

We need to use software to test the Huawei Cloud cloud server this timesysstat

We enter the following command in the terminal to installsysstat

sudo apt-get update
sudo apt-get install sysstat

expand knowledge

sysstatis a collection of utilities that provides some system performance monitoring and reporting capabilities. By using sysstat, you can get detailed information about system resource usage, including CPU usage, memory utilization, disk I/O, network traffic, and more.

Collect server load information

Use the command sar -u 1 10to collect server load information

command parsing

-u means to collect CPU usage, 1 means to sample once per second, and 10 means to collect data for 10 seconds.

Explanation of the fields:

• %user: Percentage of CPU occupied by user-mode processes.
• %nice: Percentage of CPU used by user-mode processes running at a lower priority.
• %system: The percentage of CPU occupied by the kernel mode process.
• %iowait: Percentage of CPU waiting for disk I/O to complete.
• %steal: In the virtualization environment, the percentage of CPU occupied by other virtual machines.
• %idle: The percentage of CPU idle time.

Note:
The server is not currently running any programs, as can be seen from the test results

At this point in the afternoon at 01:27:15 PM, the total CPU usage is: 0.50% user mode, no lower priority processes, 0.00% kernel mode, no waiting for disk I/O, no virtualization CPU usage in the environment, 99.50% idle time.

start pressure test

1. Run stress --cpu 2command to create 2 threads occupying CPU
2. Run sysbench cpu --cpu-max-prime=20000 runto test CPU performance, memory operations, file IO, etc.

Background processstress

As shown in the figure, we can see that during the running of the stress --cpu 2 command, the CPU usage is as follows:

Time: 02:03:31 PM - 02:03:41 PM
CPU usage:
%user: 99.85%
%nice: 0.00%
%system: 0.15%
%iowait: 0.00%
%steal: 0.00%
%idle: 0.00%

These data show that during this period, the CPU is almost fully utilized, user processes occupy the vast majority of CPU time, and the utilization rate of system processes and other processes is very low. It can be considered that the performance of the CPU is good

runsysbench

As shown in the figure, you can see sysbench the statistics during the CPU stress test:

• Tested with 1 thread
• The speed of the CPU is 446.87 events/s (number of events processed per second)
• A total of 4470 events were executed
• 10.0013 seconds in total
• 95% of events delayed within 2.26ms
• In terms of thread fairness, the number of events executed by each thread is basically the same

From these data, it can be concluded that the CPU performance of the system is relatively good, and the load level of the system will not have too much impact on the CPU execution capability of a single thread.

Test network bandwidth

Enter the following command in the terminal to test the network bandwidth of the server
sudo apt-get install iftop
sudo iftop -i eth0

The effect is as follows:

According to the network bandwidth test results, the analysis is as follows:

• hcss-ecs-0975 server sends data:

  • The amount of data sent to the host with IP address 183.242.65.238 is 1.02Kb, 1.16Kb, and 1.95Kb respectively.
  • The amount of data sent to the host with IP address 100.125.1.250 is 0b, 0b, 147b respectively.
  • The amount of data sent to the host with IP address 100.125.11.131 is 0b, 0b, 143b respectively.

• hcss-ecs-0975 server receives data:

  • The amount of data received from the host with IP address 183.242.65.238 is 208b, 208b, 299b respectively.
  • The amount of data received from the host with IP address 100.125.1.250 is 0b, 0b, 224b respectively.
  • The amount of data received from the host with IP address 100.125.11.131 is 0b, 0b, 9b respectively.

• Overall data transfer status:

  • The total amount of sent data is 9.48KB, and the amount of received data is 2.21KB.
  • The peak send rate is 5.70Kb and the peak receive rate is 3.09Kb.
  • The average sending rates are 1.02Kb, 1.16Kb, and 2.23Kb respectively.
  • The average reception rate is 208b, 208b, 533b respectively.

Summarize

The server is 2核2G 3Mrunning, so it can be seen from the results of this network bandwidth test that its network transmission performance is basically normal. The average sending rate is about 2 Kbps, and the average receiving rate is about 200 bps, which is consistent with the network transmission performance when used with 2G memory.

Test disk I/O performance

Enter the following command in the terminal to test disk I/O performance
sudo apt-get install ioping

Through the command, lsblkyou can see that there is a /dev/vdadisk device named, which is mounted on /the path
, so we run the command sudo ioping -R /dev/vdato perform a random read performance test on the /dev/vda device

I/O Performance Evaluation

sudo ioping -R /dev/vdaThe result of running the command is shown in the figure

From the disk performance test results provided, the following conclusions can be drawn:

1. Disk random read performance: 4.52 K IOPS. IOPS (input/output operations per second) is one of the common metrics to measure disk performance. Higher IOPS values ​​indicate better read capabilities of the disk.

2. Disk throughput: 17.6 MiB/s. Throughput refers to the amount of data transfer per unit time. Higher throughput means that the disk can read data faster.

3. Response time: min/avg/max/mdev = 130.6 us / 221.3 us / 2.68 ms / 94.0 us. These values ​​represent the time it takes for the disk to respond to read requests. The average response time is 221.3 us (microseconds), and the maximum response time is 2.68 ms (milliseconds). The shorter the response time, the faster the disk read speed.

Summarize

According to the test results, the disk shows good performance and can meet the general read operation requirements.