CN116932417B

CN116932417B - Performance tuning method and device

Info

Publication number: CN116932417B
Application number: CN202311203910.5A
Authority: CN
Inventors: 陈丹杰; 诸葛晓蕾
Original assignee: New H3C Information Technologies Co Ltd
Current assignee: New H3C Information Technologies Co Ltd
Priority date: 2023-09-18
Filing date: 2023-09-18
Publication date: 2023-12-22
Anticipated expiration: 2043-09-18
Also published as: CN116932417A

Abstract

The embodiment of the application provides a performance tuning method and device, which relate to the technical field of communication, and the method comprises the following steps: operating the test model according to the default configuration of the server, taking configuration items used when the test model is operated as configuration items to be tuned, and recording configuration parameters of each configuration item to be tuned; in the process of running the test model, monitoring the designated performance data of the server, and determining at least one tuning parameter of a designated configuration item based on the designated performance data, wherein the designated configuration item belongs to a configuration item to be tuned; according to at least one tuning parameter of the designated configuration items and preset tuning rules, the configuration parameters of each configuration item to be tuned are respectively adjusted to obtain at least one tuning scheme, wherein the tuning scheme comprises the configuration parameters corresponding to each configuration item to be tuned; and (3) respectively operating the test model according to each tuning scheme to obtain a performance test result corresponding to each tuning scheme. Thus, the performance tuning efficiency of the server is improved.

Description

Performance tuning method and device

Technical Field

The present disclosure relates to the field of communications technologies, and in particular, to a performance tuning method and apparatus.

Background

In the server market, clients have higher and higher requirements on server performance, and server performance is more and more emphasized. In the scenario of server performance competition, each manufacturer may run the same test model on the server using the same model data, and tune the configuration of the server based on the running result, so as to make the server achieve higher performance as much as possible.

In the process of tuning the performance of a server by each manufacturer, the model principle and model data of the test model are usually required to be manually analyzed, so that configuration items of the server on which the test model depends are determined, tuning schemes are designated for the configuration items, then the test model is operated on the server based on the tuning schemes, the performance test result is manually analyzed after the operation is completed, the tuning schemes are re-prepared, and then the test is performed based on the re-designated tuning schemes. Through multiple tests within a specified time period, a tuning scheme which enables better server performance can be obtained.

However, as the complexity of the test model increases, more configuration items are required to participate in tuning, the efficiency of manually analyzing and specifying the tuning scheme is lower, a great deal of labor cost and time cost are required to be input, and the tuning efficiency is lower.

Disclosure of Invention

In view of this, the embodiments of the present application provide a performance tuning method and apparatus to improve the performance tuning efficiency of a server. The specific technical scheme is as follows:

in a first aspect, an embodiment of the present application provides a performance tuning method, including:

operating a test model according to default configuration of a server, taking configuration items used when the test model is operated as configuration items to be tuned, and recording configuration parameters of each configuration item to be tuned;

monitoring specified performance data of the server in the process of running the test model, and determining at least one tuning parameter of a specified configuration item based on the specified performance data, wherein the specified configuration item belongs to the to-be-tuned configuration item;

according to at least one tuning parameter of the designated configuration items and preset tuning rules, the configuration parameters of each configuration item to be tuned are respectively adjusted to obtain at least one tuning scheme, wherein the tuning scheme comprises the configuration parameters corresponding to each configuration item to be tuned;

and respectively operating the test model according to each tuning scheme to obtain a test result corresponding to each tuning scheme.

In one possible implementation, after the running the test model according to the default configuration of the server, the method further includes:

Acquiring a performance baseline obtained by running the test model;

after the test model is operated according to each tuning scheme respectively to obtain the performance test result corresponding to each tuning scheme, the method further comprises the following steps:

comparing a performance test result corresponding to each tuning scheme with the performance baseline, adjusting the tuning scheme based on the comparison result, and operating the test model according to the adjusted tuning scheme to obtain a performance test result corresponding to the adjusted tuning scheme;

after each adjustment of the tuning scheme is carried out, the test model is rerun based on the adjusted tuning scheme, the latest obtained performance test result is compared with the previous performance test result, and the tuning scheme is iteratively adjusted based on the comparison result until a preset stop condition is reached.

In a possible implementation manner, the adjusting the configuration parameters of each configuration item to be tuned according to the at least one tuning parameter of the specified configuration item and the preset tuning rule to obtain at least one tuning scheme includes:

according to a preset arrangement sequence among the to-be-tuned configuration items, the configuration parameters of each to-be-tuned configuration item are sequentially tuned according to at least one tuning parameter of the designated configuration item and a preset tuning rule to obtain at least one tuning scheme.

In a possible implementation manner, according to a preset arrangement sequence between the to-be-tuned configuration items, according to at least one tuning parameter and a preset tuning rule of the specified configuration items, the configuration parameters of each to-be-tuned configuration item are sequentially tuned to obtain at least one tuning scheme, including:

taking the first to-be-tuned optimal configuration item as the current to-be-tuned optimal configuration item according to the preset arrangement sequence;

adjusting the configuration parameters of the current configuration item to be tuned according to at least one tuning parameter of the designated configuration item and a preset tuning rule;

taking the second to-be-tuned optimal configuration item as the current to-be-tuned optimal configuration item according to the preset arrangement sequence;

adjusting the configuration parameters of the current configuration item to be adjusted according to the configuration parameters adjusted by the last configuration item to be adjusted, at least one adjustment parameter of the designated configuration item and a preset adjustment rule;

and taking the next to-be-tuned configuration item as the current to-be-tuned configuration item according to the preset arrangement sequence, and returning to the step of adjusting the configuration parameters of the current to-be-tuned configuration item according to the configuration parameters adjusted by the last to-be-tuned configuration item, the at least one tuning parameter of the designated configuration item and the preset tuning rule until the last to-be-tuned configuration item is adjusted according to the preset arrangement sequence, so as to obtain at least one tuning scheme.

In a possible implementation manner, the adjusting the configuration parameters of the current to-be-tuned configuration item according to at least one tuning parameter of the specified configuration item and a preset tuning rule includes:

if the current to-be-tuned optimal configuration item belongs to the appointed configuration item, adjusting the configuration parameters of the current to-be-tuned optimal configuration item to at least one tuning parameter corresponding to the current to-be-tuned optimal configuration item;

and if the current to-be-tuned optimal configuration item does not belong to the appointed configuration item, adjusting the configuration parameters of the current to-be-tuned optimal configuration item according to a preset tuning rule corresponding to the current to-be-tuned optimal configuration item.

In one possible implementation manner, the adjusting the configuration parameters of the current configuration item to be tuned according to the configuration parameters adjusted by the previous configuration item to be tuned, the at least one tuning parameter of the specified configuration item, and the preset tuning rule includes:

if the current to-be-tuned optimal configuration item does not belong to the appointed configuration item and the configuration parameter of the last to-be-tuned optimal configuration item is one, the configuration parameter of the current to-be-tuned optimal configuration item is adjusted according to a preset tuning rule corresponding to the current tuning optimal configuration item;

If the current to-be-tuned configuration item does not belong to the designated configuration item and the configuration parameters of the last to-be-tuned configuration item are multiple, respectively carrying out primary adjustment on the configuration parameters of the current to-be-tuned configuration item based on each configuration parameter of the last to-be-tuned configuration item according to a preset tuning rule corresponding to the current to-be-tuned configuration item to obtain multiple configuration parameters of the current to-be-tuned configuration item.

In a second aspect, an embodiment of the present application provides a performance tuning apparatus, including:

the running module is used for running the test model according to the default configuration of the server, taking the configuration items used when the test model is run as the configuration items to be tuned, and recording the configuration parameters of each configuration item to be tuned;

the monitoring module is used for monitoring the appointed performance data of the server in the process of running the test model, and determining at least one tuning parameter of an appointed configuration item based on the appointed performance data, wherein the appointed configuration item belongs to the configuration item to be tuned;

the adjusting module is used for respectively adjusting the configuration parameters of each configuration item to be adjusted according to at least one adjustment parameter of the designated configuration item and a preset adjustment rule to obtain at least one adjustment scheme, wherein the adjustment scheme comprises the configuration parameters corresponding to each configuration item to be adjusted;

The operation module is further used for operating the test model according to each tuning scheme respectively to obtain a test result corresponding to each tuning scheme.

In one possible implementation manner, the apparatus further includes an acquisition module:

the acquisition module is used for acquiring a performance baseline obtained by running the test model;

the adjusting module is further configured to compare a performance test result corresponding to each tuning scheme with the performance baseline, adjust the tuning scheme based on the comparison result, and operate the test model according to the adjusted tuning scheme to obtain a performance test result corresponding to the adjusted tuning scheme; and after each adjustment of the tuning scheme, rerun the test model based on the adjusted tuning scheme, comparing the latest performance test result with the previous performance test result, and iteratively adjusting the tuning scheme based on the comparison result until reaching a preset stop condition.

In one possible implementation manner, the adjusting module is specifically configured to:

In a third aspect, an embodiment of the present application further provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, where the processor, the communication interface, and the memory complete communication with each other through the communication bus;

a memory for storing a computer program;

and a processor, configured to implement the method according to the first aspect when executing the program stored in the memory.

In a fourth aspect, embodiments of the present application further provide a computer readable storage medium, in which a computer program is stored, the computer program implementing the method according to the first aspect when being executed by a processor.

In a fifth aspect, embodiments of the present application also provide a computer program product comprising instructions which, when run on a computer, cause the computer to perform the method of the first aspect described above.

By adopting the technical scheme, the server configuration items used in running the test model can be used as the configuration items to be tuned, the configuration parameters of each configuration quantity to be tuned are recorded, the appointed performance data of the server are monitored, and tuning parameters of the appointed configuration items are determined based on the appointed performance data. The tuning parameters are obtained according to specified performance data in the actual running process of the test model, and the tuning parameters can reflect the better tuning direction of the performance of the server to a certain extent, so that the tuning scheme is determined according to the tuning parameters of the specified configuration items, and the performance tuning efficiency can be improved. After at least one tuning scheme is obtained, the test model can be operated according to each tuning scheme, and a performance test result corresponding to each tuning scheme can be obtained. Therefore, performance test results corresponding to various tuning schemes can be obtained efficiently. Therefore, compared with the process of manually making a tuning scheme and continuously optimizing the tuning scheme, the process of making the tuning scheme in the embodiment of the application does not need manual participation, reduces labor cost and can greatly improve tuning efficiency.

Of course, not all of the above-described advantages need be achieved simultaneously in practicing any one of the products or methods of the present application.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is a flowchart of a performance tuning method according to an embodiment of the present application;

FIG. 2 is an exemplary schematic diagram of a tuning queue provided in an embodiment of the present application;

FIG. 3 is an exemplary schematic diagram of a method for determining a tuning scheme according to an embodiment of the present application;

fig. 4 is a schematic software structure of an electronic device according to an embodiment of the present application;

FIG. 5 is an exemplary schematic diagram of a monitoring unit function provided in an embodiment of the present application;

fig. 6 is a schematic structural diagram of a performance tuning device according to an embodiment of the present application;

fig. 7 is a block diagram of an electronic device for implementing a performance tuning method according to an embodiment of the present application.

Detailed Description

The following description of the embodiments of the present application will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all, of the embodiments of the present application. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are within the scope of the present disclosure.

The embodiment of the application provides a performance tuning method, which can be applied to electronic equipment, and the electronic equipment can be a terminal or a server as an example. The method as shown in fig. 1 comprises the following steps:

s101, operating a test model according to default configuration of a server, taking configuration items used when the test model is operated as configuration items to be tuned, and recording configuration parameters of each configuration item to be tuned.

The default configuration refers to a server configuration when the server configuration is not adjusted.

The test model is used to test server performance. The test model may be an artificial intelligence (Artificial Intelligence, AI) model. For example, the test model may be a medical AI model.

As an example, the configuration items to be tuned may include at least one of a central processor (Central Processing Unit, CPU), a network card, a memory, and a graphics processor (Graphics Processing Unit, GPU), and may also include other configuration items of the server. The configuration items used by the server in running the test model are related to the function of the test model, which is not limited in this application.

The to-be-tuned configuration items and the configuration parameters of each to-be-tuned configuration item may be recorded in the form of key-value pairs. For example, the configuration item to be tuned is the number of CPU cores, the number of CPU cores used in running the test model is 0 to 30, and the key is the number of CPU cores, and the value is 0 to 30.

S102, monitoring specified performance data of a server in the process of running the test model, and determining at least one tuning parameter of a specified configuration item based on the specified performance data, wherein the specified configuration item belongs to a to-be-tuned tuning configuration item.

The specified performance data specifies performance data of the server module corresponding to the configuration item when the server module runs the test model. The specified configuration items may be empirically preset.

Taking the designated configuration item as an example of the CPU core number, the electronic equipment can calculate the core number with high affinity through the utilization rate of each core of the CPU and the distribution condition of each core in the CPU, and then take the core number with high affinity as the tuning parameter of the CPU core number. The utilization rate of each core of the CPU and the distribution condition of each core in the CPU are the appointed performance data of the CPU core number, and the CPU is the server module corresponding to the CPU core number.

The affinity is used for representing the relation between the tuned module and the CPU core number in the system process or the server.

The method for calculating the affinity according to the embodiment of the present application is not particularly limited.

Assuming that the CPU includes 32 cores, wherein the usage rate of 15 cores numbered 0-14 is 100%, the usage rate of other cores is 30%, and the affinity of the cores with high usage rate is high, the affinity of 15 cores numbered 0-14 is high, and the electronic device can use the cores numbered 0-14 as the tuning parameters of the CPU core number, that is, the tuning parameters of the CPU core number are 0-14.

Taking the designated configuration item as a network card port delay parameter as an example, the electronic equipment can calculate the packet loss rate of the network card port through the calling condition of each network card and the packet receiving and transmitting rate of the network card port, then determine the network card port delay parameter according to the packet loss rate of the network card port, and take the network card port delay parameter as the tuning parameter of the network card port delay parameter.

The method for calculating the network card port delay parameter is not particularly limited. Further, the above-described specified configuration items and tuning parameters corresponding to the specified configuration items are merely examples, and the specified configuration items are not limited thereto in actual implementation.

S103, according to at least one tuning parameter of the designated configuration items and preset tuning rules, the configuration parameters of each configuration item to be tuned are respectively adjusted to obtain at least one tuning scheme, and the tuning scheme comprises the configuration parameters corresponding to each configuration item to be tuned.

Wherein the preset tuning rules depend on the working principle of the server. The preset tuning rules are preconfigured according to the influence of the to-be-tuned configuration items on the server.

And S104, respectively operating the test model according to each tuning scheme to obtain a performance test result corresponding to each tuning scheme.

After the electronic equipment control server runs the test model, performance test results corresponding to each tuning scheme can be output, so that a user can determine an optimal tuning scheme according to the performance test results corresponding to each tuning scheme.

By adopting the method, the server configuration items used in running the test model can be used as the configuration items to be tuned, the configuration parameters of each configuration quantity to be tuned are recorded, the appointed performance data of the server is monitored, and the tuning parameters of the appointed configuration items are determined based on the appointed performance data. The tuning parameters are obtained according to specified performance data in the actual running process of the test model, and the tuning parameters can reflect the better tuning direction of the performance of the server to a certain extent, so that the tuning scheme is determined according to the tuning parameters of the specified configuration items, and the performance tuning efficiency can be improved. After at least one tuning scheme is obtained, the test model can be operated according to each tuning scheme, and a performance test result corresponding to each tuning scheme can be obtained. Therefore, performance test results corresponding to various tuning schemes can be obtained efficiently. Therefore, compared with the process of manually making a tuning scheme and continuously optimizing the tuning scheme, the process of making the tuning scheme in the embodiment of the application does not need manual participation, reduces labor cost and can greatly improve tuning efficiency.

When determining the tuning scheme, the electronic device needs to adjust the configuration parameters of each to-be-tuned tuning configuration item according to the preset arrangement sequence. Based on this, S103, according to at least one tuning parameter of the designated configuration item and a preset tuning rule, the configuration parameters of each configuration item to be tuned are respectively tuned to obtain at least one tuning scheme, where the tuning scheme includes configuration parameters corresponding to each configuration item to be tuned, and includes:

according to the preset arrangement sequence among the to-be-tuned configuration items, the configuration parameters of each to-be-tuned configuration item are sequentially tuned according to at least one tuning parameter of the designated configuration item and a preset tuning rule, so as to obtain at least one tuning scheme.

The preset arrangement sequence is preset according to the working principle of the server. For example, the configuration items to be tuned include configuration items related to a CPU, configuration items related to a memory, and configuration items related to a network card, and according to the working principle of the server, the configuration items related to the CPU, the configuration items related to the memory, and the configuration items related to the network card may be sequentially adjusted.

For a plurality of configuration items related to the CPU, the arrangement sequence is also preset, for example, the CPU core number, the CPU over-frequency, the CPU power consumption, the CPU thread and the like can be sequentially adjusted.

As shown in fig. 2, fig. 2 exemplarily shows a CPU tuning queue, a memory tuning queue, and a network card tuning queue. The CPU tuning queue comprises an internal module 1-N, the memory tuning queue comprises an internal module 1-N, and the network card tuning queue comprises an internal module 1-N.

It should be noted that, each internal module in fig. 2 is configured to adjust a configuration parameter of a configuration item to be tuned. The internal modules 1-N in the CPU tuning queue are used for adjusting configuration parameters of configuration items related to the CPU; the internal modules 1-N in the memory tuning queue are used for adjusting configuration parameters of the memory-related configuration items; the internal modules 1-N in the network card tuning queue are used for adjusting configuration parameters of configuration items related to the network card.

The internal module 1 of the CPU tuning queue adjusts the configuration parameters of the to-be-tuned configuration items corresponding to the internal module 1 according to at least one tuning parameter of the designated configuration items and a preset tuning rule, and then transmits the adjusted configuration parameters of the to-be-tuned configuration items to the internal module 2 in a key value pair mode;

after the internal module 2 adjusts the configuration parameters of the configuration items to be adjusted corresponding to the internal module 2 according to at least one adjustment parameter of the designated configuration items and a preset adjustment rule, the configuration items to be adjusted corresponding to the internal module 2 and the adjusted configuration parameters, and the configuration items to be adjusted corresponding to the internal module 1 and the adjusted configuration parameters are downwards transmitted to the internal module 3 in the form of key value pairs;

After the internal module 3 adjusts the configuration parameters of the to-be-adjusted configuration items corresponding to the internal module 3 according to at least one adjustment parameter of the designated configuration items and a preset adjustment rule, the to-be-adjusted configuration items corresponding to the internal module 3 and the adjusted configuration parameters, the to-be-adjusted configuration items corresponding to the internal module 2 and the adjusted configuration parameters, and the to-be-adjusted configuration items corresponding to the internal module 1 and the adjusted configuration parameters are transmitted downwards in a key value pair mode until the internal module N adjusts the configuration parameters of the to-be-adjusted configuration items corresponding to the internal module N, and then the to-be-adjusted configuration items corresponding to each internal module in the CPU adjustment queue and the adjusted configuration parameters are transmitted to the internal module 1 in the memory adjustment queue in a key value pair mode;

the memory tuning queue and the network card tuning queue also adjust configuration parameters in a similar manner until the internal module N in the network card tuning queue adjusts the configuration parameters of the configuration items to be tuned corresponding to the internal module N, and then takes all key value pairs output by the internal module N in the network card tuning queue as a complete tuning scheme and outputs the tuning scheme.

The following describes how the electronic device determines the tuning scheme using the preset ranking order:

Step 1, taking the first to-be-tuned optimal configuration item as the current to-be-tuned optimal configuration item according to a preset arrangement sequence.

And 2, adjusting the configuration parameters of the current configuration item to be adjusted according to at least one adjustment parameter of the designated configuration item and a preset adjustment rule.

The following two conditions exist for adjusting the current to-be-tuned optimal configuration item:

in case 1, if the current to-be-tuned configuration item belongs to the designated configuration item, the configuration parameters of the current to-be-tuned configuration item are adjusted to at least one tuning parameter corresponding to the current to-be-tuned configuration item.

The electronic device can adjust the configuration parameters of the current to-be-adjusted configuration item to at least one tuning parameter corresponding to the current to-be-adjusted configuration item because the current to-be-adjusted configuration item is the first configuration item in all to-be-adjusted configuration items, and whether the current to-be-adjusted configuration item is influenced by other to-be-adjusted configuration items is not required to be considered.

For example, the current to-be-tuned optimal configuration item belongs to a designated configuration item, tuning parameters of the designated configuration item are X and Y, and the electronic device adjusts the configuration parameters of the current to-be-tuned optimal configuration item to X and Y respectively, that is, the current to-be-tuned optimal configuration item corresponds to two configuration parameters, one is that the configuration parameter of the current to-be-tuned optimal configuration item takes a value of X, and the other is that the configuration parameter of the current to-be-tuned optimal configuration item takes a value of Y.

Therefore, the tuning parameters are obtained by the electronic equipment based on the specified performance data, so that the tuning parameters can reflect the tuning direction of better server performance, the configuration parameters of the current to-be-tuned configuration items can be adjusted to the tuning parameters, the server performance is quickly improved to a higher level, and the efficiency of tuning the server performance is improved.

And 2, if the current to-be-tuned optimal configuration item does not belong to the designated configuration item, adjusting the configuration parameters of the current to-be-tuned optimal configuration item according to a preset tuning rule corresponding to the current to-be-tuned optimal configuration item.

The preset tuning rules are preset according to the influence of the current to-be-tuned configuration items on the server, and each to-be-tuned configuration item corresponds to one preset tuning rule.

The configuration parameters include numeric parameters and non-numeric parameters, and the step sizes may be different for different numeric parameters when the adjustment is made. Different preset tuning rules can be set for different numerical parameters.

For example, the configuration parameters of the Memory input/output (Memory Input Output, memory IO) of the to-be-tuned configuration item in the performance baseline are 85, the configuration parameters of the Process (Process) of the to-be-tuned configuration item are 8, wherein the step size of the to-be-tuned configuration item Memory IO is 1, and the step size of the to-be-tuned configuration item Process is 2.

The electronic device may adjust the configuration parameters of the Memory IO of the to-be-tuned configuration item to 86, adjust the configuration parameters of the Process of the to-be-tuned configuration item to 10, and then rerun the test model based on the tuned tuning scheme, if the performance test result of the tuned tuning scheme is better than the performance baseline, the configuration parameters of the Memory IO of the to-be-tuned configuration item and the configuration parameters of the Process of the to-be-tuned configuration item may be continuously increased until the latest obtained performance test result is smaller than the previous performance test result.

If the performance test result of the adjusted tuning scheme is inferior to the performance baseline, the configuration parameters of Memory IO of the to-be-tuned configuration item can be reduced to 84, the configuration parameters of the to-be-tuned configuration item Process are reduced to 6, and then the test model is rerun based on the adjusted tuning scheme until the latest obtained performance test result is greater than the previous performance test result.

If the current to-be-tuned optimal configuration item does not belong to the designated configuration item, the configuration parameters of the current to-be-tuned optimal configuration item can be quickly adjusted by utilizing the preset tuning rule corresponding to the current to-be-tuned optimal configuration item.

And 3, taking the second to-be-tuned optimal configuration item as the current to-be-tuned optimal configuration item according to the preset arrangement sequence.

And 4, adjusting the configuration parameters of the current configuration item to be adjusted according to the configuration parameters adjusted by the last configuration item to be adjusted, at least one adjustment parameter of the designated configuration item and a preset adjustment rule.

It should be noted that, if the current to-be-tuned configuration item belongs to the designated configuration item, the configuration parameters of the current to-be-tuned configuration item are adjusted to at least one tuning parameter corresponding to the current to-be-tuned configuration item.

It can be understood that when the configuration parameters of the current to-be-tuned configuration item are adjusted, the configuration parameters of the current to-be-tuned configuration item are adjusted to at least one tuning parameter corresponding to the current to-be-tuned configuration item by combining the working principle of the server and the configuration parameters adjusted by the previous to-be-tuned configuration item.

If the current to-be-tuned optimal configuration item does not belong to the designated configuration item and the configuration parameter of the last to-be-tuned optimal configuration item is one, the configuration parameter of the current to-be-tuned optimal configuration item is adjusted according to a preset tuning rule corresponding to the current to-be-tuned optimal configuration item.

If the current to-be-tuned configuration item does not belong to the designated configuration item and the configuration parameters of the last to-be-tuned configuration item are multiple, respectively carrying out one-time adjustment on the configuration parameters of the current to-be-tuned configuration item based on each configuration parameter of the last to-be-tuned configuration item according to the preset tuning rule corresponding to the current to-be-tuned configuration item to obtain multiple configuration parameters of the current to-be-tuned configuration item.

For example, the current to-be-tuned configuration item does not belong to the designated configuration item, the configuration parameters of the last to-be-tuned configuration item are 3, and the electronic device needs to adjust the configuration parameters of the current to-be-tuned configuration item at one time based on the preset tuning rule corresponding to the current to-be-tuned configuration item and each configuration parameter of the last to-be-tuned configuration item, so that the electronic device can obtain multiple configuration parameters of the current to-be-tuned configuration item.

And 5, taking the next to-be-tuned configuration item as the current to-be-tuned configuration item according to a preset arrangement sequence, and returning to the step of adjusting the configuration parameters of the current to-be-tuned configuration item according to the configuration parameters adjusted by the last to-be-tuned configuration item, at least one tuning parameter of the designated configuration item and a preset tuning rule until the last to-be-tuned configuration item is adjusted according to the preset arrangement sequence, so as to obtain at least one tuning scheme.

It can be understood that if a specified configuration item including multiple configuration parameters exists in all to-be-tuned configuration items corresponding to the test model, or the electronic device obtains multiple configuration parameters of the current to-be-tuned configuration item based on the configuration parameters adjusted by the last to-be-tuned configuration item and the preset tuning rules corresponding to the current to-be-tuned configuration item, multiple tuning schemes can be obtained based on the method for determining the tuning scheme provided by the embodiment of the present application.

As shown in fig. 3, fig. 3 schematically illustrates that the internal module 1 in the CPU tuning queue in fig. 2 is a binding core module, the internal module 2 is a basic input output System (Basic Input Output System, BIOS) configuration module, and the internal module 3 is an Operating System (OS) configuration module. It should be noted that, the internal module in the embodiment of the present application is merely an example, and in an actual implementation, the internal module in the CPU tuning queue is not limited thereto.

The monitoring unit is used for monitoring various performance data of the server when the test model is operated.

The monitoring unit may provide affinity suggestions for the test model and the CPU core number. The affinity suggestion is the tuning parameter corresponding to the CPU core number of the configuration item to be tuned.

The monitoring unit may further provide tuning parameters of the to-be-tuned configuration item corresponding to the BIOS configuration module and tuning parameters of the to-be-tuned configuration item corresponding to the OS configuration module, which are not shown in fig. 3.

The tuning parameters of the binding core module are X, Y and Z. The tuning parameters of the BIOS configuration module are configuration A, configuration B and configuration C, and the tuning parameters of the OS configuration module are scheme A, scheme B and scheme C.

The core binding module respectively adjusts the configuration parameters of the to-be-adjusted configuration items corresponding to the core binding module into X, Y and Z, namely three configuration parameters of the to-be-adjusted configuration items corresponding to the core binding module, wherein the first configuration parameter is X, the second configuration parameter is Y, and the third configuration parameter is Z.

The binding core module respectively transmits the X, Y and Z and the to-be-tuned configuration items corresponding to the binding core module to the BIOS configuration module in a key value pair mode, and the BIOS configuration module respectively adjusts the configuration parameters of the to-be-tuned configuration items corresponding to the BIOS configuration module into a configuration A, a configuration B and a configuration C based on the preset tuning rules of the to-be-tuned configuration items corresponding to the BIOS configuration module.

The BIOS configuration module transmits the three configuration parameters after self adjustment and the to-be-adjusted configuration items corresponding to the BIOS configuration module, and the three configuration parameters after adjustment of the binding module and the to-be-adjusted configuration items corresponding to the binding module to the OS configuration module in a key value pair mode, and the OS configuration module adjusts the configuration parameters of the to-be-adjusted configuration items corresponding to the OS configuration module into a scheme A, a scheme B and a scheme C respectively based on the preset adjustment rules of the to-be-adjusted configuration items corresponding to the OS configuration module and each configuration parameter after adjustment of the BIOS configuration module.

And the same is repeated until all internal modules in the CPU tuning queue finish the adjustment of the configuration parameters of the configuration items to be tuned corresponding to the internal modules, and the CPU tuning queue outputs a tuning scheme aiming at the CPU. And the subsequent other tuning queues continue to adjust the configuration parameters until all the tuning queues finish the adjustment of the configuration parameters, and then output the configuration parameters participating in tuning to obtain a final tuning scheme.

Fig. 3 only shows a tuning scheme of the CPU tuning queue output, and in an actual implementation, a final tuning scheme further includes tuning schemes of other tuning queue outputs, which are not described herein.

By adopting the method, the configuration parameters of each configuration item to be tuned are adjusted one by one according to the preset tuning rule corresponding to each configuration item to be tuned and tuning parameters of the designated configuration item according to the preset arrangement sequence. The tuning scheme of the server running the high-complexity test model can be output in a short time. When the configuration parameters of other to-be-tuned optimal configuration items except the first to-be-tuned optimal configuration item are adjusted, the electronic equipment needs to integrate the configuration parameters of the current to-be-tuned optimal configuration item and the last to-be-tuned optimal configuration item, and the configuration parameters of the current to-be-tuned optimal configuration item are determined. Therefore, the determined tuning scheme can accord with the working principle of the server, and the problem that the server cannot normally operate the test model based on the tuning scheme is avoided. In addition, the configuration parameters of the configuration items to be tuned are adjusted one by one, the tuning bottleneck of the server performance can be accurately positioned, and the electronic equipment can acquire important configuration parameters affecting the server performance. And when the similar test model is operated again, the important configuration parameters are utilized to rapidly improve the performance of the server, and the performance tuning efficiency is improved.

After the electronic device obtains the performance test result of each tuning scheme, the electronic device can further adjust each tuning scheme, based on the adjustment, after the test model is operated according to the default configuration of the server, a performance baseline obtained by operating the test model can also be obtained, the electronic device further adjusts the tuning scheme according to the performance baseline, and after the test model is operated according to each tuning scheme respectively, the performance test result corresponding to each tuning scheme is obtained, the method further comprises:

and step A, comparing a performance test result corresponding to each tuning scheme with a performance baseline, adjusting the tuning scheme based on the comparison result, and operating a test model according to the adjusted tuning scheme to obtain a performance test result corresponding to the adjusted tuning scheme.

The electronic device can determine that the server performance needs to be continuously improved under the tuning scheme by comparing the performance baseline with the performance test result of the tuning scheme, and then adjusts the configuration parameters of each configuration item to be tuned in the tuning scheme according to the server performance needing to be improved.

And B, after each time of adjustment is carried out on the tuning scheme, the test model is rerun based on the adjusted tuning scheme, the latest performance test result is compared with the previous performance test result, and the tuning scheme is iteratively adjusted based on the comparison result until a preset stop condition is reached.

In the performance competition scene, the preset stopping condition may be preset tuning time or preset iteration times. The preset stop conditions are not particularly limited in the embodiment of the present application.

The performance competition is a server performance ranking competition, and specifically refers to model data of an official given test model, and the performance of the server running the test model is improved by formulating a server performance adjustment strategy within a preset duration.

It should be noted that, since there may be multiple situations in the values of the configuration parameters of the configuration item to be tuned in the server, the electronic device may perform multiple iterations on the numerical parameters in the tuning scheme.

After the performance base line is obtained by adopting the method, the performance to be improved exists by utilizing the tuning scheme which can be determined by the performance base line, so that the tuning scheme can be quickly tuned. And then, the tuning scheme is continuously and iteratively adjusted according to the performance test result of the adjusted tuning scheme and the performance test result of the last time, namely, the tuning scheme is continuously optimized, the performance of the server is improved to a great extent, and the performance tuning time can be reduced in an actual client scene and a performance competition scene.

As shown in fig. 4, fig. 4 is a schematic software structure of an electronic device according to an embodiment of the present application, where the electronic device includes a test model input unit 401, a baseline operation unit 402, a monitoring unit 403, a distributed intelligent tuning unit 404, and an output operation unit 405.

The test model input unit 401 is used to input a software package of a test model and model data into a server.

The baseline operation unit 402 is configured to operate the input test model, and then output the performance baseline and the to-be-tuned optimal configuration item.

The monitoring unit 403 is configured to monitor various performance data of the server while the server runs the test model.

The monitoring unit 403 then determines tuning parameters for the specified configuration items based on the specified performance data, and passes the tuning parameters for the specified configuration items to the distributed intelligent tuning unit 404.

As shown in fig. 5, fig. 5 illustrates that the monitoring unit 403 may monitor the CPU usage rate by using Htop, monitor the network card usage rate by using Net-tools, and monitor the GPU usage rate by using Dmon, and in an actual implementation, the monitoring unit may also monitor other performance data of the server, which is not limited in particular in the embodiment of the present application. Wherein Htop is the instrument of monitoring CPU utilization ratio, net-tools is the instrument of monitoring network card utilization ratio, and Dmon is the instrument of monitoring GPU utilization ratio.

The monitoring unit 403 may monitor the usage rate condition of each core of the CPU and the distribution condition of cores with high usage rate in the CPU by calling Htop, calculate the affinity of the CPU core number, and then use the CPU core number with high affinity as the tuning parameter of the CPU core number. Specifically, the monitoring unit 403 may calculate the affinity of the number of cores of the CPU through the use rate of the cores and the distribution of cores with high use rate in the whole CPU. For example, the use rate of cores numbered 0 to 15 is 100%, the use rate of cores numbered 16 to 50 is 30%, and the monitoring unit 403 determines that the tuning parameter of the number of CPU cores is 0 to 15.

The monitoring unit 403 may monitor the call condition of each network card and the packet receiving and transmitting rate of the network card port by calling Net-tools, and when the packet loss rate of the network card port is higher than a preset threshold, the monitoring unit 403 may determine the interrupt signal parameter and the tuning parameter corresponding to the delay parameter of the network card port.

The monitoring unit 403 transmits the tuning parameters of the CPU core number and the tuning parameters of the network card port to the distributed intelligent tuning unit 404.

The distributed intelligent tuning unit 404 may tune the configuration parameters of each configuration item to be tuned based on the tuning parameters of the designated configuration item to obtain multiple tuning schemes, and then transmit the multiple tuning schemes to the output operation unit 405.

The output execution unit 405 executes the test model based on a plurality of tuning schemes, and outputs tuning schemes and performance test results. And the distributed intelligent tuning unit 404 may further perform scheme optimization on each tuning scheme after obtaining the performance test result corresponding to each tuning scheme, so as to continuously improve the server performance.

Based on the same concept, the embodiment of the present application provides a performance tuning apparatus, as shown in fig. 6, including:

the operation module 601 is configured to operate the test model according to a default configuration of the server, take configuration items used when the test model is operated as configuration items to be tuned, and record configuration parameters of each configuration item to be tuned;

the monitoring module 602 is configured to monitor specified performance data of the server during the running of the test model, determine at least one tuning parameter of a specified configuration item based on the specified performance data, where the specified configuration item belongs to a to-be-tuned configuration item;

the adjusting module 603 is configured to adjust the configuration parameters of each configuration item to be adjusted according to at least one adjustment parameter of the designated configuration item and a preset adjustment rule, so as to obtain at least one adjustment scheme, where the adjustment scheme includes configuration parameters corresponding to each configuration item to be adjusted;

The operation module 601 is further configured to operate the test model according to each tuning scheme, so as to obtain a test result corresponding to each tuning scheme.

Optionally, the apparatus further includes an acquisition module:

the adjusting module 603 is further configured to compare, for each tuning scheme, a performance test result corresponding to the tuning scheme with a performance baseline, adjust the tuning scheme based on the comparison result, and operate the test model according to the adjusted tuning scheme, so as to obtain a performance test result corresponding to the adjusted tuning scheme; and after each adjustment of the tuning scheme, rerun the test model based on the adjusted tuning scheme, comparing the latest performance test result with the previous performance test result, and iteratively adjusting the tuning scheme based on the comparison result until reaching a preset stop condition.

Optionally, the adjusting module 603 is specifically configured to:

taking the first to-be-tuned optimal configuration item as the current to-be-tuned optimal configuration item according to a preset arrangement sequence;

adjusting the configuration parameters of the current configuration item to be adjusted according to at least one adjustment parameter of the designated configuration item and a preset adjustment rule;

and taking the next to-be-tuned configuration item as the current to-be-tuned configuration item according to a preset arrangement sequence, and returning to the step of adjusting the configuration parameters of the current to-be-tuned configuration item according to the configuration parameters adjusted by the last to-be-tuned configuration item, at least one tuning parameter of the designated configuration item and a preset tuning rule until the adjustment of the last to-be-tuned configuration item is completed according to the preset arrangement sequence, so as to obtain at least one tuning scheme.

Optionally, the adjusting module 603 is specifically configured to:

If the current to-be-tuned optimal configuration item does not belong to the appointed configuration item, the configuration parameters of the current to-be-tuned optimal configuration item are adjusted according to the preset tuning rule corresponding to the current to-be-tuned optimal configuration item.

Optionally, the adjusting module 603 is specifically configured to:

if the current to-be-tuned optimal configuration item does not belong to the appointed configuration item and the configuration parameter of the last to-be-tuned optimal configuration item is one type, the configuration parameter of the current to-be-tuned optimal configuration item is adjusted according to a preset tuning rule corresponding to the current tuning optimal configuration item;

The embodiment of the present application further provides an electronic device, as shown in fig. 7, including a processor 701, a communication interface 702, a memory 703, and a communication bus 704, where the processor 701, the communication interface 702, and the memory 703 perform communication with each other through the communication bus 704,

A memory 703 for storing a computer program;

the processor 701 is configured to implement the steps of the performance tuning method when executing the program stored in the memory 703:

the communication bus mentioned above for the electronic devices may be a peripheral component interconnect standard (Peripheral Component Interconnect, PCI) bus or an extended industry standard architecture (Extended Industry Standard Architecture, EISA) bus, etc. The communication bus may be classified as an address bus, a data bus, a control bus, or the like. For ease of illustration, the figures are shown with only one bold line, but not with only one bus or one type of bus.

The communication interface is used for communication between the electronic device and other devices.

The Memory may include random access Memory (Random Access Memory, RAM) or may include Non-Volatile Memory (NVM), such as at least one disk Memory. Optionally, the memory may also be at least one memory device located remotely from the aforementioned processor.

The processor may be a general-purpose processor, including a central processing unit (Central Processing Unit, CPU), a network processor (Network Processor, NP), etc.; but also digital signal processors (Digital Signal Processing, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), field programmable gate arrays (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

In yet another embodiment provided herein, there is also provided a computer readable storage medium having stored therein a computer program which when executed by a processor implements the steps of any of the performance tuning methods described above.

In yet another embodiment provided herein, there is also provided a computer program product containing instructions that, when run on a computer, cause the computer to perform any of the performance tuning methods of the above embodiments.

In the above embodiments, it may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loaded and executed on a computer, produces a flow or function in accordance with embodiments of the present application, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable apparatus. The computer instructions may be stored in or transmitted from one computer-readable storage medium to another, for example, by wired (e.g., coaxial cable, optical fiber, digital Subscriber Line (DSL)), or wireless (e.g., infrared, wireless, microwave, etc.). The computer readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that contains an integration of one or more available media. The usable medium may be a magnetic medium (e.g., floppy Disk, hard Disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid State Disk (SSD)), etc.

It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

In this specification, each embodiment is described in a related manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments. In particular, for system embodiments, since they are substantially similar to method embodiments, the description is relatively simple, as relevant to see a section of the description of method embodiments.

The foregoing description is only of the preferred embodiments of the present application and is not intended to limit the scope of the present application. Any modifications, equivalent substitutions, improvements, etc. that are within the spirit and principles of the present application are intended to be included within the scope of the present application.

Claims

1. A method of performance tuning, comprising:

respectively operating the test model according to each tuning scheme to obtain a performance test result corresponding to each tuning scheme;

The adjusting the configuration parameters of each configuration item to be adjusted according to the at least one tuning parameter of the designated configuration item and the preset tuning rule to obtain at least one tuning scheme comprises the following steps:

according to a preset arrangement sequence among the to-be-tuned configuration items, according to at least one tuning parameter of the designated configuration items and a preset tuning rule, the configuration parameters of each to-be-tuned configuration item are sequentially tuned to obtain at least one tuning scheme;

according to a preset arrangement sequence among the to-be-tuned configuration items, the configuration parameters of each to-be-tuned configuration item are sequentially tuned according to at least one tuning parameter of the designated configuration item and a preset tuning rule to obtain at least one tuning scheme, and the method comprises the following steps:

adjusting the configuration parameters of the current configuration item to be tuned according to at least one tuning parameter of the designated configuration item and a preset tuning rule; the preset tuning rules are preconfigured according to the influence of the to-be-tuned configuration items on the server;

2. The method of claim 1, wherein after the running the test model according to the default configuration of the server, the method further comprises:

acquiring a performance baseline obtained by running the test model;

3. The method according to claim 1, wherein said adjusting the configuration parameters of the current configuration item to be tuned according to at least one tuning parameter of the specified configuration item and a preset tuning rule comprises:

4. A method according to claim 1 or 3, wherein said adjusting the configuration parameters of the current configuration item to be tuned according to the configuration parameters adjusted by the previous configuration item to be tuned, the at least one tuning parameter of the specified configuration item, and a preset tuning rule comprises:

if the current to-be-tuned optimal configuration item does not belong to the appointed configuration item and the configuration parameter of the last to-be-tuned optimal configuration item is one, adjusting the configuration parameter of the current to-be-tuned optimal configuration item according to a preset tuning rule corresponding to the current to-be-tuned optimal configuration item;

5. A performance tuning apparatus, comprising:

the operation module is further used for respectively operating the test model according to each tuning scheme to obtain a test result corresponding to each tuning scheme;

the adjusting module is specifically configured to:

6. The apparatus of claim 5, further comprising an acquisition module:

7. The apparatus of claim 5, wherein the adjustment module is specifically configured to:

8. The device according to claim 5 or 7, wherein the adjustment module is specifically configured to:

9. The electronic equipment is characterized by comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory are communicated with each other through the communication bus;

A memory for storing a computer program;

a processor for implementing the method of any of claims 1-4 when executing a program stored on a memory.

10. A machine-readable storage medium storing machine-executable instructions that, when invoked and executed by a processor, cause the processor to: the method of any one of claims 1-4 is implemented.