Server load detection and traffic distribution system

Abstract

The invention discloses a server load detection and flow distribution system, which relates to the technical field of servers and solves the problem that the load parameters given by the system are not accurate and are likely to affect the subsequent network task processing process if the server is not load tested. The invention uses simulated request data to perform performance testing on the server, and the simulated request data contains a plurality of data with different request amounts. The simulated request data is transmitted to the server, and load testing is preferentially performed on different servers. The simulated data is used to perform data load analysis and processing to confirm the read and write rates of the corresponding servers, and corresponding waveforms are established according to specific values. In the process of determining the optimal load amount, not only the maximum read and write rate is considered, but also the trend of the data is taken into consideration. The determination process is more comprehensive, so that the confirmed optimal load amount is more accurate and the error is smaller, which is convenient for subsequent flow distribution processing.

Description

Technical Field

[0001] The present invention relates to the technical field of servers, and in particular to a server load detection and flow distribution system. Background Art

[0002] A server is a type of computer that runs faster, has a higher load, and is more expensive than an ordinary computer. A server provides computing or application services to other clients (such as PCs, smartphones, ATMs and other terminals, and even large devices such as train systems) in the network.

[0003] The application with patent publication number CN111835653A relates to the field of network traffic control technology, and in particular to a control method and system for network traffic load balancing, the method comprising: configuring the traffic proportion and hop number of each link, and monitoring the status of each link; if the status of each link is normal, and the hop number of each link is the same, the traffic is distributed to each link in normal state according to the traffic proportion; if there is a link in abnormal state, the link is removed, and if the hop number of the remaining links in normal state is the same, the traffic is distributed to each link in normal state according to the traffic proportion; when it is detected that the link has returned to normal state, the link is added, and if the hop number of the links in normal state is the same, the traffic is distributed to each link in normal state according to the traffic proportion. The application has the following beneficial effects: when the network traffic increases, the response speed of the server is still not affected; when a network link fails, the network will not be interrupted, thereby ensuring the user experience.

[0004] During the process of load detection and traffic distribution, when the internal read and write rate of the corresponding server slows down, it is marked as a load state. Later, when the network task traffic is distributed, tasks are distributed according to the servers in the load state or normal state. However, this method still has the following shortcomings that need to be improved:

[0005] The server has not been load tested. The load parameters given by the system are not accurate and may affect the subsequent network task processing process.

[0006] In the process of allocating network tasks, the node tasks are not divided evenly according to the total number of servers, which makes it easy for subsequent processing data to be confused. Summary of the invention

[0007] In view of the deficiencies in the prior art, the present invention provides a server load detection and traffic distribution system, which solves the problem that the load parameters given by the system are not accurate and easily affect the subsequent network task processing process if the server is not load tested.

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions: a server load detection and flow distribution system, comprising:

[0009] The database stores a large amount of simulation request data, and the simulation request data is preset data, which is input and stored in advance by the operator;

[0010] The performance testing unit retrieves simulated request data from the database and performs performance testing on different types of servers. The specific methods are as follows:

[0011] Gradually increase the amount of data requested in the simulated request data, test the server, and record the read and write rates during the test;

[0012] The data request amount parameter is used as the horizontal coordinate axis, and the recorded read and write rate parameter is used as the vertical coordinate axis to construct a two-dimensional coordinate system, and the test parameters generated during the test are filled into the two-dimensional coordinate system, and then the points corresponding to several test parameters are connected to construct a performance test waveform chart, and the constructed performance test waveform chart is transmitted to the load parameter analysis unit;

[0013] The load parameter analysis unit receives the performance test waveform and analyzes the performance test waveform in the following ways:

[0014] Mark the turning points in the performance test waveform. The waveform lines at both ends of the turning points have different trends. The waveform lines between the confirmed turning points are marked as lines to be analyzed.

[0015] Confirm the trend parameters of the line to be analyzed, obtain the coordinate parameters of the endpoints of the corresponding line to be analyzed, and mark the coordinate parameters of the front endpoint as (X1, Y1), and the coordinate parameters of the back endpoint as (X2, Y2). Use QS = (Y2-Y1) ÷ (X2-X1) to confirm the trend parameter QS of the line to be analyzed, and mark the trend parameters QS of different lines to be analyzed as QS i , where i represents different lines to be analyzed;

[0016] Analytical QS i Does it meet QS? i >0, if satisfied, the corresponding trend parameters and the line to be analyzed are calibrated as the demand parameters and demand lines; if not satisfied, no processing is performed;

[0017] From several demand lines, select the highest point of the internal endpoint and mark the read / write rate corresponding to the highest point as DX k , and then mark the demand parameter of the corresponding demand line as XQ k , where k represents different demand lines;

[0018] HD k =DX k ×C1+XQ k ×C2 gets the approved value HD of the corresponding demand line k , where C1 and C2 are preset fixed coefficient factors, and then from several approved values ​​HD k The maximum value is selected, and the demand line corresponding to the maximum value is confirmed. The corresponding maximum read and write rate is confirmed from the confirmed demand line, so as to confirm the corresponding request volume according to the maximum read and write rate, and this request volume is used as the optimal load of the corresponding server.

[0019] Preferably, the server traffic distribution system comprises:

[0020] A demand confirmation unit confirms the phase demand of the external network task and transmits the confirmed phase demand to the main distribution unit;

[0021] The main allocation unit, based on the confirmed stage demand and the confirmation of several servers participating in this network task, first confirms whether the optimal load in several servers meets the stage demand, and performs different system operations, specifically in the following ways:

[0022] The optimal loads of several servers participating in this network task are summed to obtain the combined parameter HB, and then the stage demand is marked as JD, and it is first determined whether HB satisfies HB≥JD:

[0023] If it is satisfied, it can be evenly distributed according to the optimal load of the corresponding server to complete the calculation of the overall network task;

[0024] If it is not satisfied, then confirm the total number of servers and mark them as GS. According to the task nodes within the network task, divide the network task into several node tasks, and simultaneously confirm the task requirements of different node tasks. From several node tasks, the remaining GS node tasks are subjected to secondary distribution processing. The task requirements of the remaining node tasks are summed up, and the sum parameter QH is confirmed. Analyze whether HB satisfies HB≥QH. If so, the GS node tasks that need to be secondary distributed are transferred to the secondary distribution unit. If not, GS node tasks are eliminated again until HB≥QH is reached. All proposed node tasks are transferred to the secondary distribution unit, and the remaining node tasks are calculated and processed.

[0025] Preferably, the secondary distribution unit receives the node tasks that need to be secondary distributed, restarts the server, and performs calculation processing on the node tasks in turn in an evenly distributed manner, and the processing process is as follows:

[0026] Prioritize extracting GS node tasks and transferring them to the corresponding servers for processing. After all servers have completed the processing, the second stage of processing will be carried out to process GS node tasks again. The processing will be carried out in sequence until the node tasks that need to be redistributed are completed.

[0027] Beneficial Effects

[0028] The present invention provides a server load detection and flow distribution system. Compared with the prior art, it has the following beneficial effects:

[0029] The present invention uses simulated request data to perform performance testing on a server, and the simulated request data contains data of several different request amounts. The simulated request data is transmitted to the server, and a performance test is performed. At the same time, the read and write rates are confirmed. Different servers are preferentially subjected to load testing. The simulated data is used to perform data load analysis and processing to confirm the read and write rates of the corresponding servers. Subsequently, a corresponding waveform is established according to specific values. In the process of determining the optimal load amount, not only the maximum read and write rate is considered, but also the trend of the data is taken into consideration. The determination process is more comprehensive, so that the confirmed optimal load amount is more accurate, the error is smaller, and the subsequent flow distribution process is convenient.

[0030] Subsequently, the network tasks are evenly divided according to the optimal load of different servers and the specific capacity of the network tasks. During the equalization process, the remaining node tasks corresponding to the number of servers are processed subsequently or twice. This distribution and processing method is used to evenly divide the traffic within the network task, and during the equalization process, it is ensured that each server can handle the corresponding network task, which will not cause confusion in the subsequent processing data and lead to erroneous results. In this way, the overall processing accuracy of the entire network task can be improved, and the overall effect of traffic distribution can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is a schematic diagram of the principle framework of the present invention. DETAILED DESCRIPTION

[0032] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0033] Embodiment 1

[0034] See also Figure 1, the present application provides a server load detection system, including a database, a performance test unit and a load parameter analysis unit, wherein the database, the performance test unit and the load parameter analysis unit are electrically connected in sequence from an output end to an input end;

[0035] The database stores a large amount of simulation request data internally, and the simulation request data is preset data, which is input and stored in advance by the operator. Specifically, the simulation request data are parameter data processed during the actual processing of the server, and the simulation request data includes different types of parameter data;

[0036] The performance testing unit retrieves simulated request data from the database, performs performance testing on different types of servers, and monitors the read and write rates of the servers in real time, constructs a performance test waveform graph belonging to the corresponding server according to the specific request amount and read and write rate of the simulated request data, and transmits the constructed performance test waveform graph to the load parameter analysis unit, wherein the specific method of construction is:

[0037] Gradually increase the amount of data requested in the simulated request data, test the server, and record the read and write rates during the test;

[0038] A two-dimensional coordinate system is constructed by taking the data request volume parameter as the horizontal coordinate axis and the recorded read and write rate parameter as the vertical coordinate axis. The test parameters generated during the test are filled into the two-dimensional coordinate system. Then, the points corresponding to several test parameters are connected to construct a performance test waveform graph, and the constructed performance test waveform graph is transmitted to the load parameter analysis unit.

[0039] Specifically, there are several data with different request amounts in the simulated request data, and the simulated request data is transmitted to the server, and a performance test is performed, and the read and write rates are confirmed. During the performance test, the read and write rates can fully demonstrate the performance of the server. If the read and write rates slow down, it can indirectly show the load situation of the corresponding server.

[0040] The load parameter analysis unit receives and analyzes the performance test waveform, and confirms the optimal load of different servers by confirming the trend parameters and fluctuation points within the waveform, wherein the specific determination method is:

[0041] Mark the turning points in the performance test waveform. Specifically, the turning point is the point where the turning point occurs. The waveform lines at both ends of the turning point have different trends. The waveform lines between the confirmed turning points are marked as the lines to be analyzed.

[0042] Confirm the trend parameters of the line to be analyzed, obtain the coordinate parameters of the endpoints of the corresponding line to be analyzed, and mark the coordinate parameters of the front endpoint as (X1, Y1), and the coordinate parameters of the back endpoint as (X2, Y2). Use QS = (Y2-Y1) ÷ (X2-X1) to confirm the trend parameter QS of the line to be analyzed, and mark the trend parameters QS of different lines to be analyzed as QS i , where i represents different lines to be analyzed;

[0043] Analytical QS i Does it meet QS? i >0, if satisfied, the corresponding trend parameters and the line to be analyzed are calibrated as the demand parameters and demand lines; if not satisfied, no processing is performed;

[0044] From several demand lines, select the highest point of the internal endpoint and mark the read / write rate corresponding to the highest point as DX k , and then mark the demand parameter of the corresponding demand line as XQ k , where k represents different demand lines;

[0045] HD k =DX k ×C1+XQ k ×C2 gets the approved value HD of the corresponding demand line k , where C1 and C2 are preset fixed coefficient factors, and then from several approved values ​​HD k The maximum value is selected, and the demand line corresponding to the maximum value is confirmed. The corresponding maximum read and write rate is confirmed from the confirmed demand line, so as to confirm the corresponding request volume according to the maximum read and write rate, and this request volume is used as the optimal load of the corresponding server.

[0046] In actual application, different servers have different usage time and their corresponding internal components have different aging conditions, which will cause the maximum load that the servers can withstand to change even though they are the same type of servers. Therefore, it is preferred to perform load testing on different servers and use simulated data to perform data load analysis and processing to confirm the read and write rates of the corresponding servers. Subsequently, the corresponding waveform is established based on the specific values. In the process of determining the optimal load, not only the maximum read and write rate is considered, but also the trend of the data. The determination process is more comprehensive, making the confirmed optimal load more accurate and with smaller errors, which is convenient for subsequent traffic distribution processing.

[0047] Embodiment 2

[0048] In the specific implementation process, this embodiment is different from the first embodiment in that:

[0049] During the use of the server load detection system, there is also a server traffic distribution system adapted thereto, including a demand confirmation unit, a main distribution unit and a secondary distribution unit, wherein the demand confirmation unit is electrically connected to the input end of the main distribution unit, and the main distribution unit is electrically connected to the input end of the secondary distribution unit;

[0050] The demand confirmation unit confirms the stage demand of the external network task and transmits the confirmed stage demand to the main allocation unit. Specifically, first, before the corresponding network task is executed, there is a confirmed demand inside, and the system can identify such demand by itself;

[0051] The main allocation unit, based on the confirmed stage demand and the confirmation of the several servers participating in the network task, preferentially confirms whether the optimal load in the several servers meets the stage demand, and performs different system operations, wherein the specific execution method is:

[0052] The optimal loads of several servers participating in this network task are summed to obtain the combined parameter HB, and then the stage demand is marked as JD, and it is first determined whether HB satisfies HB≥JD:

[0053] If it is satisfied, it can be evenly distributed according to the optimal load of the corresponding server to complete the calculation of the overall network task;

[0054] If it is not satisfied, then confirm the total number of servers and mark them as GS. According to the task nodes within the network task, divide the network task into several node tasks, and simultaneously confirm the task requirements of different node tasks. From several node tasks, the remaining GS node tasks are subjected to secondary distribution processing. The task requirements of the remaining node tasks are summed up, and the sum parameter QH is confirmed. Analyze whether HB satisfies HB≥QH. If so, the GS node tasks that need to be secondary distributed are transferred to the secondary distribution unit. If not, GS node tasks are eliminated again until HB≥QH is reached. All proposed node tasks are transferred to the secondary distribution unit, and the remaining node tasks are calculated and processed.

[0055] The secondary distribution unit receives the node tasks that need to be secondary distributed, restarts the server, and performs calculation and processing on the node tasks in turn in an evenly distributed manner. The processing process is as follows:

[0056] Prioritize extracting GS node tasks, transferring them to the corresponding servers for processing. After all servers have completed processing, the second stage of processing is carried out to process GS node tasks again, and the processing is carried out in sequence until the node tasks that need to be re-allocated are completed.

[0057] Specifically, this distribution processing method is used to evenly distribute the traffic within the network task, and during the distribution process, it is ensured that each server can handle the corresponding network task, which will not cause confusion in the subsequent processing data and lead to erroneous results. In this way, the overall processing accuracy of the entire network task can be improved, and the overall effect of traffic distribution can be improved.

[0058] Some of the data in the above formulas are dimensionless and numerically calculated. Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0059] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A server load detection system, characterized in that: include: The database stores a large amount of simulation request data, and the simulation request data is preset data, which is input and stored in advance by the operator; The performance test unit retrieves simulated request data from the database, performs performance tests on different types of servers, and monitors the read and write rates of the servers in real time. It constructs a performance test waveform chart belonging to the corresponding server according to the specific request volume and read and write rate of the simulated request data, and transmits the constructed performance test waveform chart to the load parameter analysis unit. The specific method is as follows: Gradually increase the amount of data requested in the simulated request data, test the server, and record the read and write rates during the test; The data request amount parameter is used as the horizontal coordinate axis, and the recorded read and write rate parameter is used as the vertical coordinate axis to construct a two-dimensional coordinate system, and the test parameters generated during the test are filled into the two-dimensional coordinate system, and then the points corresponding to several test parameters are connected to construct a performance test waveform chart, and the constructed performance test waveform chart is transmitted to the load parameter analysis unit; The load parameter analysis unit receives and analyzes the performance test waveform, and confirms the optimal load of different servers by confirming the trend parameters and fluctuation points within the waveform. The specific method is as follows: Mark the turning points in the performance test waveform. The waveform lines at both ends of the turning points have different trends. The waveform lines between the confirmed turning points are marked as lines to be analyzed. Confirm the trend parameters of the line to be analyzed, obtain the coordinate parameters of the endpoints of the corresponding line to be analyzed, and mark the coordinate parameters of the front endpoint as (X1, Y1), and the coordinate parameters of the back endpoint as (X2, Y2). Use QS = (Y2-Y1) ÷ (X2-X1) to confirm the trend parameter QS of the line to be analyzed, and mark the trend parameters QS of different lines to be analyzed as QS i , where i represents different lines to be analyzed; Analytical QS i Does it meet QS? i >0, if satisfied, the corresponding trend parameters and the line to be analyzed are calibrated as the demand parameters and demand lines; if not satisfied, no processing is performed; From several demand lines, select the highest point of the internal endpoint and mark the read / write rate corresponding to the highest point as DX k , and then mark the demand parameter of the corresponding demand line as XQ k , where k represents different demand lines; HD k =DX k ×C1+XQ k ×C2 gets the approved value HD of the corresponding demand line k , where C1 and C2 are preset fixed coefficient factors, and then from several approved values ​​HD k The maximum value is selected, and the demand line corresponding to the maximum value is confirmed. The corresponding maximum read and write rate is confirmed from the confirmed demand line, so as to confirm the corresponding request volume according to the maximum read and write rate, and this request volume is used as the optimal load of the corresponding server.

2. A server traffic distribution system, which is adapted to the server load detection system of claim 1, characterized in that: include: A demand confirmation unit confirms the phase demand of the external network task and transmits the confirmed phase demand to the main distribution unit; The main allocation unit confirms the stage demand according to the confirmed stage demand and several servers participating in the network task, gives priority to confirming whether the optimal load in several servers meets the stage demand, and executes different system operations.

3. The server traffic distribution system according to claim 2, characterized in that: The specific method of executing different system operations by the main distribution unit is as follows: The optimal loads of several servers participating in this network task are summed to obtain the combined parameter HB, and then the stage demand is marked as JD, and it is first determined whether HB satisfies HB≥JD: If it is satisfied, it can be evenly distributed according to the optimal load of the corresponding server to complete the calculation of the overall network task; If it is not satisfied, then confirm the total number of servers and mark them as GS. According to the task nodes within the network task, divide the network task into several node tasks, and simultaneously confirm the task requirements of different node tasks. From several node tasks, the remaining GS node tasks are subjected to secondary distribution processing. The task requirements of the remaining node tasks are summed up, and the sum parameter QH is confirmed. Analyze whether HB satisfies HB≥QH. If so, the GS node tasks that need to be secondary distributed are transferred to the secondary distribution unit. If not, GS node tasks are eliminated again until HB≥QH is reached. All proposed node tasks are transferred to the secondary distribution unit, and the remaining node tasks are calculated and processed.

4. The server traffic distribution system according to claim 3, characterized in that: The secondary distribution unit receives the node tasks that need to be secondary distributed, restarts the server, and performs calculation and processing on the node tasks in turn in an evenly distributed manner. The processing process is as follows: Prioritize extracting GS node tasks and transferring them to the corresponding servers for processing. After all servers have completed processing, the second stage of processing will be carried out to process GS node tasks again. The processing will be carried out in sequence until the node tasks that need to be redistributed are completed.

Images