Terminal management and control performance test method and device based on dual-garden dual-active

Abstract

The invention provides a terminal management and control performance test method and device based on dual-park active-active, and the method comprises the steps: carrying out the benchmark test for different business types in a single park, and monitoring and recording the performance data of each business type; the terminal management and control performance of the double parks is estimated in combination with the performance data of the benchmark test of the single park and a comprehensive attenuation factor model, and the comprehensive attenuation factor model comprises a network delay attenuation factor model, a data synchronization attenuation factor model and a resource competition attenuation factor model. According to the method, the terminal management and control performance condition of the dual-garden dual-active architecture can be accurately and scientifically estimated.

Description

Technical Field

[0001] The present invention belongs to the technical field of performance testing, and in particular relates to a terminal management and control performance testing method and device based on dual-park dual-active. Background Art

[0002] In the field of computer networks, terminal control is a system used to manage and control desktop operating system terminals. It implements centralized management, authority control, and security monitoring of terminals by establishing domains. Terminal control can provide unique access control, unified user identity authentication, precise access control, and rigorous resource management, simplifying the management work between internal levels of the enterprise.

[0003] Terminal management and control clients refer to computer terminal devices that are managed under the terminal management and control platform, such as employee workstations, terminal servers, etc. These terminal devices access the terminal management and control platform through a specific network and are managed and controlled by it. The terminal management and control platform can perform policy management, restrict terminal user behavior, monitor terminal user behavior, and manage patches on terminals to ensure the security and maintainability of terminal devices.

[0004] However, with the deepening of localization, the expansion of enterprise scale, the growth of internal organizational levels and the increase in business needs, in order to improve the disaster recovery and reliability of the system, more and more data centers are using dual-park dual-active mode to deploy system services. Dual-park dual-active data center refers to two geographically independent data centers (for example, set up in two parks). Through efficient and reliable data replication and synchronization technology, real-time mirroring of key data and application status is achieved. Both data centers are active, processing business requests at the same time, and backing up each other to ensure that when any server or network has problems, the other data center can quickly take over to ensure service continuity and stability.

[0005] In order to ensure normal operation and stability, performance testing is required to evaluate its performance under different loads and concurrent access conditions. However, the existing traditional performance testing is usually conducted on a single campus, and it is difficult to directly deduce the performance in a dual-campus environment. If factors such as network latency, data synchronization overhead, and resource contention involved in a dual-campus active-active environment are not taken into account, the accuracy of performance prediction may also be reduced. Summary of the invention

[0006] The object of the present invention is to provide a terminal management and control performance testing method and device based on dual-park dual-active, which can accurately and scientifically estimate the terminal management and control performance of the dual-park dual-active architecture.

[0007] In order to achieve the above object, the technical solution of the present invention is as follows:

[0008] A terminal management and control performance testing method based on dual-park dual-active, comprising:

[0009] S1. In a single park, benchmark different service types, monitor and record the performance data of each service type;

[0010] S2. Combine the performance data of the benchmark test of a single park and the comprehensive attenuation factor model to estimate the terminal control performance of the dual park. The estimation formula is: dual park performance = single park performance × N × Ftotal; where N is the multiple relationship of the single park performance, and Ftotal is the comprehensive attenuation factor model;

[0011] Ftotal=Fdelay×Fsync×Fresource; where Fdelay is the network delay attenuation factor model; Fsync is the data synchronization attenuation factor model; and Fresource is the resource competition attenuation factor model.

[0012] Furthermore, the network delay attenuation factor model Fdelay in step S2 includes:

[0013] Fdelay = 1-(x × D);

[0014] Where D represents the average network delay between the two campuses; x represents the loss ratio of network throughput caused by network delay.

[0015] Furthermore, the data synchronization attenuation factor model Fsync in step S2 includes:

[0016] Fsync = 1 - (S × P);

[0017] Among them, S represents the frequency of data synchronization between the two parks, and P represents the performance overhead caused by each synchronization.

[0018] Furthermore, the resource competition attenuation factor model Fresource in step S2 includes:

[0019] Fresource=1-R;

[0020] Among them, R represents the performance loss ratio caused by resource competition;

[0021] R=1-(service throughput during resource contention / service throughput during non-resource contention).

[0022] On the other hand, the present invention also proposes a terminal management and control performance testing device based on dual-park dual-active, comprising:

[0023] Single-campus test module: Benchmark tests are performed on different service types in a single park, and the performance data of each service type is monitored and recorded;

[0024] Dual-park estimation module: Combines the performance data of the single-park benchmark test and the comprehensive attenuation factor model to estimate the terminal management and control performance of the dual-park. The estimation formula is: Dual-park performance = single-park performance × N × Ftotal; N is the multiple of the single-park performance, and Ftotal is the comprehensive attenuation factor model; Ftotal = Fdelay × Fsync × Fresource; Fdelay is the network delay attenuation factor model; Fsync is the data synchronization attenuation factor model; and Fresource is the resource competition attenuation factor model.

[0025] Furthermore, the dual-park estimation module includes a network delay attenuation factor model calculation unit, and the network delay attenuation factor model calculation unit includes:

[0026] Fdelay = 1-(x × D);

[0027] Where D represents the average network delay between the two campuses; x represents the loss ratio of network throughput caused by network delay.

[0028] Furthermore, the dual-park estimation module includes a data synchronization attenuation factor model calculation unit, and the data synchronization attenuation factor model calculation unit includes:

[0029] Fsync = 1 - (S × P);

[0030] Among them, S represents the frequency of data synchronization between the two parks, and P represents the performance overhead caused by each synchronization.

[0031] Furthermore, the dual-park estimation module includes a resource competition attenuation factor model calculation unit, and the resource competition attenuation factor model calculation unit includes:

[0032] Fresource=1-R;

[0033] Among them, R represents the performance loss ratio caused by resource competition;

[0034] R=1-(service throughput during resource contention / service throughput during non-resource contention).

[0035] The present invention also proposes a computer-readable storage medium, which stores a computer program, and the computer program is used to execute the above-mentioned terminal management and control performance testing method based on dual-park dual-active.

[0036] The present invention also proposes a computer program product, including a computer program, which implements the above-mentioned terminal management and control performance testing method based on dual-park dual-active when executed by a processor.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] The present invention estimates the terminal management and control performance of dual-park dual-active by combining the performance data of the benchmark test of a single park with a comprehensive attenuation factor model. It can obtain accurate estimation results while considering multiple factors such as network delay, synchronization overhead and resource competition, understand the performance of terminal management and control under different pressures, scientifically evaluate the system's carrying capacity for different business scenarios, and provide a basis for resource planning and deployment. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 This is a schematic diagram of the overall process of Example 1 of the present invention;

[0040] Figure 2 This is a schematic diagram of the process of a single park performance test according to Embodiment 1 of the present invention;

[0041] Figure 3 This is a system architecture diagram of dual-park dual-active terminal control according to Example 1 of the present invention. DETAILED DESCRIPTION

[0042] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0043] The design idea of ​​the present invention is to conduct a benchmark test of a single park through traditional performance testing methods, so as to obtain the performance of a single park in different business scenarios, and then further combine the attenuation factors of various factors such as delay, synchronization overhead and resource contention to accurately estimate the performance of the dual parks.

[0044] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0045] Embodiment 1:

[0046] Example 1 proposes a terminal management and control performance testing method based on dual-park dual-active, and its overall process is as follows: Figure 1 As shown, including:

[0047] S1. In a single park, benchmark different service types, monitor and record the performance data of each service type;

[0048] Since the software and hardware configurations, services, etc. of each park in the dual campus are the same and mirror each other, you only need to choose one of the parks for testing.

[0049] The specific testing process is as follows Figure 2 As shown, including:

[0050] 1. Benchmark test settings:

[0051] Deploy system services in a single park and conduct benchmark tests for different service types. Test content includes but is not limited to service throughput, response time, and usage of server CPU, memory, disk and other resources. Write performance test scripts based on the test tasks planned in the benchmark test.

[0052] 2. Performance data collection:

[0053] During the benchmark test, key performance indicators of each service type are monitored and recorded, including response time, TPS, CPU utilization, memory occupancy, etc. In this embodiment, the open source tool jmeter is used to run the performance test script to obtain performance data.

[0054] 3. Performance data analysis:

[0055] By analyzing the performance data, the performance of each service type in a single park can be obtained, and then a performance support capability report for different services under specific resource configurations can be generated.

[0056] S2. Combine the performance data of the single-park benchmark test and the comprehensive attenuation factor model to estimate the terminal management and control performance of the dual-park.

[0057] In this step, the dual-park dual-active system architecture is as follows Figure 3 As shown, the container cluster within each park is equipped with an internal database, application services and an internal gateway. The application services of each park are connected to the external database respectively, and each park is connected to the DNS intelligent resolution device through the external gateway.

[0058] Based on the above architecture, the terminal management and control performance of the dual-park can be estimated by combining the performance data obtained from the benchmark test in step S1 and the comprehensive attenuation factor model.

[0059] The comprehensive attenuation factor model includes a network delay attenuation factor model Fdelay, a data synchronization attenuation factor model Fsync, and a resource competition attenuation factor model Fresource.

[0060] 1. The network delay attenuation factor model Fdelay is calculated based on the impact of network delay between two campuses.

[0061] Fdelay = 1-(x × D);

[0062] D: average network delay between two campuses;

[0063] x: The ratio of network throughput loss caused by latency.

[0064] Formula description: Assuming that the average delay is D milliseconds, each increase of 1 millisecond in delay will result in a decrease in throughput by x, so we get Fdelay = 1-(x×D).

[0065] The average network delay D is measured as follows:

[0066] (1.1) Use the ping command to obtain preliminary latency data (for example: ping campus ip-c 100);

[0067] (1.2) Use iperf to simulate business traffic and analyze the delay of actual data flow (for example: one campus runs iperf-s; the other campus runs iperf-c campus ip-ub 0-t 100);

[0068] (1.3) Based on the data values ​​obtained by ping and iperf, take the average of the two as the average network delay D.

[0069] The loss in network throughput due to latency can be measured as follows:

[0070] (1-1) Use ifconfig to query the network card to query the theoretical network throughput B (this value is fixed and not affected by environmental factors);

[0071] (1-2) The actual network throughput T is calculated by the average network delay D, and the formula is: T = window size / average network delay D. The window size refers to the TCP sliding window size, which is usually 64KB and generally does not need to be changed. Even if this value is changed, it is a fixed value;

[0072] (1-3) Calculate the loss ratio of network throughput loss due to delay. The formula is: x = 1-(T / B).

[0073] 2. The data synchronization attenuation factor model Fsync is calculated based on the data synchronization frequency and overhead between the two parks.

[0074] Fsync = 1 - (S × P);

[0075] S: frequency of data synchronization between two parks; (the frequency of data synchronization between parks is fixed, so S is a fixed value);

[0076] P: performance overhead caused by each synchronization;

[0077] Formula description: Assuming that the synchronization frequency is S and the performance loss of each synchronization overhead is P, then Fsync = 1-(S×P).

[0078] The performance overhead P caused by each synchronization is measured as follows:

[0079] (2.1) By viewing the data synchronization log, obtain the time t when data synchronization starts and the time e when data synchronization ends;

[0080] (2.2) Obtain the time w spent on processing the data synchronization task by viewing the data synchronization log;

[0081] (2.3) Calculate the performance overhead P caused by each synchronization, that is, P = (et) / (w+(et)).

[0082] 3. The resource competition attenuation factor model Fresource is estimated based on the impact of resource allocation and load balancing among parks.

[0083] Fresource=1-R;

[0084] R: performance loss ratio due to resource competition (such as CPU, network, etc.);

[0085] Formula description: Assuming that resource competition brings a performance loss ratio of R, then Fresource = 1-R.

[0086] The method to obtain R is as follows:

[0087] (3.1) Obtain the usage of CPU, network and other resources through the load monitoring platform, and distinguish between resource competition and non-resource competition. The distinction method includes:

[0088] When conducting a single-park test, different concurrency numbers are tested. The business throughput (tps) obtained from these concurrency numbers will show a Poisson distribution, that is, there is a maximum value, denoted as m, which is the business throughput when there is no resource competition. The resource utilization at this time is the best. Assuming that the resource utilization value at this time is u, then when the resource utilization value is greater than u, it is judged to be in resource competition.

[0089] (3.2) Through the single-park test data, the service throughput when resource contention and when there is no resource contention is obtained.

[0090] (3.3) Calculate R:

[0091] R=1-service throughput during resource contention / service throughput during non-resource contention.

[0092] 4. The comprehensive attenuation factor model Ftotal is used to estimate the performance of the dual-park, and is calculated using the following formula:

[0093] Ftotal=Fdelay×Fsync×Fresource;

[0094] The final dual-park performance estimation formula is:

[0095] Dual-park performance = single-park performance × N × Ftotal;

[0096] N is a multiple of the performance of a single park, which is determined based on the number of parks that need to be deployed. (Dual-park dual-active generally refers to two parks. However, it does not rule out the possibility of increasing capacity, that is, deploying another set of dual-park dual-active, or simply adding a single park, so it needs to be determined based on the actual number of parks deployed).

[0097] Actual tests are carried out in a dual-park environment to verify the accuracy of the estimation results. Each attenuation factor model can also be calibrated according to actual data to ensure the effectiveness of the estimation method.

[0098] The method proposed in this embodiment can obtain accurate dual-park dual-active terminal management and control performance estimation results while taking into account multiple factors such as network delay, synchronization overhead and resource contention.

[0099] Embodiment 2:

[0100] This embodiment proposes a terminal management and control performance test device based on dual-park dual-active, including:

[0101] Single-campus test module: Benchmark tests are performed on different service types in a single park, and the performance data of each service type is monitored and recorded;

[0102] Dual-park estimation module: Combines the performance data of the single-park benchmark test and the comprehensive attenuation factor model to estimate the terminal management and control performance of the dual-park. The estimation formula is: Dual-park performance = single-park performance × N × Ftotal; N is the multiple of the single-park performance, and Ftotal is the comprehensive attenuation factor model; Ftotal = Fdelay × Fsync × Fresource; Fdelay is the network delay attenuation factor model; Fsync is the data synchronization attenuation factor model; and Fresource is the resource competition attenuation factor model.

[0103] The dual-park estimation module includes a network delay attenuation factor model calculation unit, and the network delay attenuation factor model calculation unit includes:

[0104] Fdelay = 1-(x × D);

[0105] Where D represents the average network delay between the two campuses; x represents the loss ratio of network throughput caused by network delay.

[0106] The dual-park estimation module also includes a data synchronization attenuation factor model calculation unit, and the data synchronization attenuation factor model calculation unit includes:

[0107] Fsync = 1 - (S × P);

[0108] Among them, S represents the frequency of data synchronization between the two parks, and P represents the performance overhead caused by each synchronization.

[0109] The dual-park estimation module also includes a resource competition attenuation factor model calculation unit, and the resource competition attenuation factor model calculation unit includes:

[0110] Fresource=1-R;

[0111] Among them, R represents the performance loss ratio caused by resource competition;

[0112] R=1-(service throughput during resource contention / service throughput during non-resource contention).

[0113] The terminal management and control performance testing device based on dual-park dual-active proposed in this embodiment can implement the terminal management and control performance testing method based on dual-park dual-active proposed in Example 1, and has the same technical effect as Example 1.

[0114] The above-mentioned embodiments are only preferred implementations of the present invention, and are only used to help understand the method and core ideas of the present application. The protection scope of the present invention is not limited to the above-mentioned embodiments. All technical solutions under the idea of ​​the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A terminal management and control performance testing method based on dual-park dual-active, characterized in that: include: S1. In a single park, benchmark different service types, monitor and record the performance data of each service type; S2. Combine the performance data of the benchmark test of a single park and the comprehensive attenuation factor model to estimate the terminal control performance of the dual park. The estimation formula is: dual park performance = single park performance × N × Ftotal; where N is the multiple relationship of the single park performance, and Ftotal is the comprehensive attenuation factor model; Ftotal=Fdelay×Fsync×Fresource; where Fdelay is the network delay attenuation factor model; Fsync is the data synchronization attenuation factor model; and Fresource is the resource competition attenuation factor model.

2. The terminal management and control performance testing method based on dual-park dual-active according to claim 1 is characterized in that: The network delay attenuation factor model Fdelay in step S2 includes: Fdelay = 1-(x × D); Where D represents the average network delay between the two campuses; x represents the loss ratio of network throughput caused by network delay.

3. The terminal management and control performance testing method based on dual-park dual-active according to claim 1 is characterized in that: The data synchronization attenuation factor model Fsync in step S2 includes: Fsync = 1 - (S × P); Among them, S represents the frequency of data synchronization between the two parks, and P represents the performance overhead caused by each synchronization.

4. The terminal management and control performance testing method based on dual-park dual-active according to claim 1 is characterized in that: The resource competition attenuation factor model Fresource in step S2 includes: Fresource=1-R; Among them, R represents the performance loss ratio caused by resource competition; R=1-(service throughput during resource contention / service throughput during non-resource contention).

5. A terminal management and control performance test device based on dual-park dual-active, characterized in that: include: Single-campus test module: Benchmark tests are performed on different service types in a single park, and the performance data of each service type is monitored and recorded; Dual-park estimation module: Combines the performance data of the single-park benchmark test and the comprehensive attenuation factor model to estimate the terminal management and control performance of the dual-park. The estimation formula is: Dual-park performance = single-park performance × N × Ftotal; N is the multiple of the single-park performance, and Ftotal is the comprehensive attenuation factor model; Ftotal = Fdelay × Fsync × Fresource; Fdelay is the network delay attenuation factor model; Fsync is the data synchronization attenuation factor model; and Fresource is the resource competition attenuation factor model.

6. The terminal management and control performance test device based on dual-park dual-active according to claim 5 is characterized in that: The dual-park estimation module includes a network delay attenuation factor model calculation unit, and the network delay attenuation factor model calculation unit includes: Fdelay = 1-(x × D); Where D represents the average network delay between the two campuses; x represents the loss ratio of network throughput caused by network delay.

7. The terminal management and control performance test device based on dual-park dual-active according to claim 5 is characterized in that: The dual-park estimation module includes a data synchronization attenuation factor model calculation unit, and the data synchronization attenuation factor model calculation unit includes: Fsync = 1 - (S × P); Among them, S represents the frequency of data synchronization between the two parks, and P represents the performance overhead caused by each synchronization.

8. The terminal management and control performance test device based on dual-park dual-active according to claim 5 is characterized in that: The dual-park estimation module includes a resource competition attenuation factor model calculation unit, and the resource competition attenuation factor model calculation unit includes: Fresource=1-R; Among them, R represents the performance loss ratio caused by resource competition; R=1-(service throughput during resource contention / service throughput during non-resource contention).

9. A computer-readable storage medium storing a computer program, characterized in that: The computer program is used to execute the terminal management and control performance testing method based on dual-park dual-active as described in any one of claims 1-4.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by the processor, the terminal management and control performance testing method based on dual-park dual-active is implemented as described in any one of claims 1 to 4.

Images