Server architecture and resource scheduling method

Abstract

The invention provides a server framework and a resource scheduling method.The server framework comprises a server body, a cover plate and a mainboard, a mounting base is arranged in the server body, a module assembly is arranged in the mounting base, a clamping groove is formed in the side face of the module assembly, a guide rail is arranged in the mounting base, a movable groove is formed in the guide rail, and a spring is arranged in the movable groove; a transmission rod is arranged at the end of the spring, a rotating shaft is arranged in the movable groove, the resource demand prediction system, the hot plug system, the request distribution system, the resource distribution system and the monitoring system are adopted, the future resource demand dynamic state of the server body can be grasped in real time, a manager can prepare an adaptive pluggable module assembly in advance, and the management efficiency is improved. According to the technical scheme, under the condition that the operation of the server main body is not interrupted, the plugging operation of the module assembly is flexibly carried out, so that the maintenance convenience is greatly improved, and the server main body is ensured to always meet the constantly changing service requirements.

Description

Technical Field

[0001] The present invention relates to the technical field of servers, and in particular to a server architecture and a resource scheduling method. Background Art

[0002] The patent application with Chinese patent number "CN102081431A" discloses a server architecture, including two partitions arranged between the side panels, multiple slide grooves arranged on the side panels and the partitions, at least one adapter board module and multiple mainboards. The partition has at least one positioning column. The mainboards are slid in the corresponding slide grooves respectively. The adapter board module includes a support member and an adapter board, and the support member has a slot engaged with the positioning column to position the adapter board module on the partition. The adapter board includes two first gold finger connection interfaces. The mainboard includes a second gold finger connection interface, and the first gold finger connection interface is respectively adapted to engage with the second gold finger connection interface, so that each adapter board is connected to two mainboards correspondingly. Through the modular design of the adapter board, the assembly operation of the adapter board can be made smoother and simpler. This server architecture also connects the adapter board module and the mainboard through the gold finger interface, and the mainboard can be slid in the shell, avoiding the inconvenience caused by cable connection.

[0003] The patent application with Chinese patent number "CN116991783A" discloses a multi-node server architecture, which includes a server chassis, a central processing unit node pluggably installed in the server chassis, a data exchange node pluggably installed in the server chassis, a graphics processing unit node pluggably installed in the server chassis, and a mid-backplane module installed in the server chassis, wherein the mid-backplane module is provided with a first connector, a second connector and a third connector for signal interconnection, the central processing unit node is plugged with the first connector, the data exchange node is plugged with the second connector, and the graphics processing unit node is plugged with the third connector. The present invention can minimize the use of data cables, conveniently and quickly perform the separate disassembly and maintenance operations of each node, realize the flexible configuration of each node, and ensure the heat dissipation performance at the same time.

[0004] The above patent documents and prior art have the following technical problems when used:

[0005] In the actual application of the server architecture in the above-mentioned patent document, the hardware components of the server do not adopt an independent modular design. This fixed-configuration hardware module shows obvious lack of flexibility in the face of increasingly changing business needs, and does not support the direct hot-swap function of the overall module. It cannot adapt to the data processing needs when the business load changes dynamically. It is difficult for administrators to quickly insert or unplug the corresponding hardware modules according to the actual needs of the server without interrupting the operation of the server, so as to achieve flexible allocation of server resources. Summary of the invention

[0006] Technical issues solved

[0007] In view of the deficiencies of the prior art, the present invention provides a server architecture and method based on module expansion, which solves the following problems:

[0008] In the actual application of server architecture, fixed-configuration hardware modules lack independent modular design and direct hot-swap functions, resulting in insufficient flexibility when facing changing business needs, and it is difficult to achieve flexible allocation of resources without interrupting server operation.

[0009] Technical Solution

[0010] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0011] On the one hand, he provided a server architecture, including a server body, a cover plate and a mainboard, a mounting seat is provided inside the server body, a module assembly is provided inside the mounting seat, a card slot is opened on the side of the module assembly, a guide rail is provided inside the mounting seat, a movable groove is provided inside the guide rail, a spring is provided inside the movable groove, a transmission rod is provided at the end of the spring, a rotating shaft is provided inside the movable groove, a movable rod is rotatably connected to the top of the rotating shaft, a card block is provided at the end of the movable rod, a CPU is provided on the surface of the mainboard, and a memory is provided on the surface of the mainboard.

[0012] On the other hand, a resource scheduling method based on the above server architecture is also provided, which comprises the following steps:

[0013] Sp1: Collecting the current resource status information of the server body through the main control system, and sending the current resource status information to the resource demand prediction system;

[0014] Sp2: The resource demand prediction system predicts resource demand based on the current resource status information and historical resource status information of the server body using the trained resource prediction model, feeds back the prediction results to the main control system, and selects the module components that need to be plugged in and out based on the prediction results;

[0015] Sp3: Insert the selected module assembly into the corresponding slot along the guide rail inside the mounting seat. When the module assembly is fully inserted, the locking mechanism is triggered to automatically lock the module assembly, and a signal that the module assembly has been inserted and information on whether the hardware connection status of the module assembly is normal are sent to the main control system;

[0016] Sp4: The request allocation system automatically obtains the module information of all inserted module components from the main control system, and sets the task processing weight of each module component according to the module information of all inserted module components in the server body, and allocates different task requests to module components with different task processing weights;

[0017] Sp5: The resource allocation system dynamically allocates the resources of the server body to different module components.

[0018] Beneficial Effects

[0019] Compared with the prior art, the present invention has the following beneficial effects:

[0020] The present invention adopts a resource demand prediction system, a request allocation system, a resource allocation system and a monitoring system. Through the resource demand prediction system, the future resource demand dynamics of the server body can be grasped in real time, so that management personnel can prepare adaptive pluggable module components in advance. Once the demand emerges, function expansion and resource allocation can be achieved through pluggable module components, ensuring that the server body can also operate stably during business peak periods, effectively avoiding the risk of service performance degradation or interruption caused by resource shortages. At the same time, the request allocation system and the resource allocation system work together to ensure that these module components can be seamlessly expanded within the server body, thereby significantly enhancing the upgrade and maintenance performance of the server body, realizing flexible allocation and efficient utilization of server body resources, and ensuring that the server body can always meet the ever-changing business needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 is an axonometric view of the present invention;

[0022] Figure 2 The figure is a disassembled diagram of the server main body and module components of the present invention;

[0023] Figure 3 The internal structure of the present invention Figure 1 ;

[0024] Figure 4 The internal structure of the present invention Figure 2 ;

[0025] Figure 5 It is a structural diagram of the module assembly and the guide rail of the present invention;

[0026] Figure 6 For the present invention Figure 5 A magnified view of middle;

[0027] Figure 7 This is a diagram of the internal system structure of the server body of the present invention. DETAILED DESCRIPTION

[0028] 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. Specific embodiment one:

[0030] like Figure 1-6 As shown, this embodiment provides a server architecture, including a server body 1, a cover plate 2 and a main board 7. The main board 7 is a core component of the server body 1, which connects and coordinates the work of various components, provides slots and interfaces for key components such as a CPU 8 and a memory 9, and realizes data processing and transmission. A CPU 8 is provided on the surface of the main board 7. The CPU 8 is the central processing unit of the server body 1 and is responsible for executing instructions and processing data. A memory 9 is provided on the surface of the main board 7. The memory 9 is used to store data and instructions that the CPU 8 needs to process.

[0031] The server body 1 is provided with a cooling fan 21 to reduce the temperature inside the server body 1 to prevent damage to components caused by overheating, and the server body 1 is provided with a power module 10 to provide a stable power supply for the server body 1;

[0032] Furthermore, a display screen 20 is provided on the surface of the server body 1. Through the setting of the display screen 20, performance indicators such as the memory 9 usage rate and business volume indicators of the server body 1 can be displayed. When the algorithm predicts that the memory 9 demand will increase, the performance indicators of the server body 1 and the predicted demand growth will be displayed on the display screen 20.

[0033] A mounting seat 5 is provided inside the server body 1, through which the module assembly 3 is fixed and an electrical connection interface is provided, the module assembly 3 is provided inside the mounting seat 5, a card slot 12 is provided on the side of the module assembly 3, and a guide rail 4 is provided inside the mounting seat 5, through which the module assembly 3 is guided to slide and position in the mounting seat 5, guide grooves 6 are provided on both sides of the module assembly 3, and the shape and position of the guide grooves 6 are adapted to the guide rails 4, and the guide grooves 6 cooperate with the guide rails 4 to guide the sliding direction of the module assembly 3.

[0034] Furthermore, a movable groove 13 is provided inside the guide rail 4, and a spring 14 is provided inside the movable groove 13. The setting of the spring 14 facilitates the restoration of the transmission rod 15 to its original position. A transmission rod 15 is provided at the end of the spring 14, and a moving block 18 is provided at the end of the transmission rod 15, and the end of the moving block 18 is connected through the end of the guide rail 4. A person changes the position of the transmission rod 15 by moving the moving block 18, thereby stretching the spring 14. At the same time, a rotating shaft 17 is provided inside the movable groove 13, and a movable rod 16 is rotatably connected to the top of the rotating shaft 17, one end of the movable rod 16 is rotatably connected to the surface of the transmission rod 15, and the other end of the movable rod 16 is connected through the side of the guide rail 4.

[0035] Through the setting of the movable rod 16, when the transmission rod 15 moves, the transmission rod 15 is rotatably connected to the end of the movable rod 16, thereby driving the movable rod 16 to rotate, so that the other end of the movable rod 16 rotates into the movable groove 13, so that the block 19 rotates out of the groove 12.

[0036] Furthermore, a block 19 is provided at the end of the movable rod 16, the shape and position of the slot 12 are adapted to the block 19, and the slot 12 and the block 19 are arranged in a linear array in several groups (such as three groups). When the block 19 and the slot 12 are engaged, the module assembly 3 can be locked. At the same time, the spring 14, the transmission rod 15, the movable rod 16, the rotating shaft 17, the moving block 18 and the block 19 together constitute a locking mechanism, which is used to automatically lock the module assembly 3 after it is installed in place. The elastic force of the spring 14 pushes the transmission rod 15 and the movable rod 16, so that the block 19 is inserted into the slot 12, thereby achieving firm fixation of the module assembly 3.

[0037] Furthermore, contacts are provided inside the mounting seat 5 to contact corresponding contacts on the module assembly 3 to achieve electrical connection. Specifically, a connecting line 11 is provided on the side of the mounting seat 5, and the end of the connecting line 11 is electrically connected to the main board 7 to achieve transmission of electrical signals. An independent shared data wiring interface is integrated on the main board 7, and the interface adopts high-speed, high-bandwidth connection technology. The shared data wiring interface is connected to the mounting seat 5 corresponding to each module assembly 3 through the wiring layer inside the main board 7 through the connecting line 11, thereby realizing rapid transmission and sharing of data between different module assemblies 3.

[0038] The above-mentioned interface design has good compatibility and scalability, and can support data interaction between module components 3 of different types and specifications. The connecting line 11 adopts a high-speed, low-latency cable as a data transmission medium, such as a copper cable or an optical fiber. After the connecting line 11 is reasonably wired inside the server body 1, each module component 3 can be connected to the shared data wiring interface on the main board 7. At the same time, in order to improve the reliability and maintainability of the wiring, the connecting line 11 adopts a modular design, and each mounting seat 5 has an independent plug and socket, which is convenient for replacement or maintenance when necessary.

[0039] When performing a hot-swap operation, the operator needs to accurately align the guide groove 6 of the module assembly 3 with the guide rail 4, and smoothly push it into the mounting seat 5 along the guide rail 4. During this process, the module assembly 3 will naturally apply a thrust to the block 19, prompting the movable rod 16 to flexibly rotate around the rotating shaft 17, thereby causing the block 19 to rotate and retract into the movable groove 13. At the same time, when the movable rod 16 rotates, its other end will rotate on the transmission rod 15. This action will link the transmission rod 15 and the moving block 18 to move toward the outside of the guide rail 4, thereby stretching the spring 14 and accumulating potential energy. Once the module assembly 3 is successfully installed, thanks to the ingenious design of the slot 12, the spring 14 will release its accumulated potential energy during the process of restoring its original state, pushing the transmission rod 15 back to its original position. This action will drive the movable rod 16 to rotate on the rotating shaft 17 again, thereby causing the block 19 to accurately rotate and snap into the slot 12, thereby achieving a firm fixation of the module assembly 3.

[0040] If the module assembly 3 needs to be pulled out, the operator only needs to simply pull the moving block 18 outward to drive the transmission rod 15 to move outward and stretch the spring 14. At the same time, the movement of the transmission rod 15 will link the movable rod 16 to rotate on the rotating shaft 17, thereby causing the clamping block 19 to rotate from the clamping slot 12 and retract into the movable slot 13. At this time, the operator can easily remove the module assembly 3 from the mounting base 5. Specific embodiment 2:

[0042] like Figure 7 As shown, the difference between this embodiment and the first embodiment is that the server body 1 includes a main control system 100, a resource demand prediction system 200, a request allocation system 300, a resource allocation system 400 and a monitoring system 500, wherein:

[0043] The main control system 100 includes a main board 7, a CPU 8 and a memory 9, and is used to control the resource demand prediction system, the request allocation system and the resource allocation system as a whole. The main control system is also connected to the monitoring system to obtain the current resource status information of the server body 1; the resource status information includes but is not limited to one or more of the CPU 8 usage rate, memory 9 occupancy, storage I / O rate, network bandwidth utilization, number of concurrent users, number of running applications, business transaction frequency, etc.;

[0044] The resource demand prediction system 200, which is based on a prediction algorithm of machine learning, predicts resource demand according to the current resource status information and historical resource status information of the server body 1, and determines the module components 3 that need to be plugged in and out according to the prediction results, so that the server management personnel can plan resource allocation in advance to avoid service performance degradation or interruption caused by insufficient resources during business peak periods;

[0045] A request allocation system 300, which is used to automatically obtain module information of all inserted module components 3 in the server body 1 through an algorithm, and set the task processing weight of each module component 3 according to the module information, and allocate different task requests to module components 3 with different weights according to the task requests received from various applications of the server body 1 and the type and priority of the task requests, so as to ensure that the server body 1 can efficiently use the resources of the pluggable module components 3, replenish resources by timely calling the pluggable module components 3, and maintain the normal operation of the business; wherein the module information includes one or more of the module type, module specification, module performance parameters and module physical location information;

[0046] The resource allocation system 400 is used to dynamically allocate the resources of the server body 1 to different module components 3 according to the task requests from various applications of the server body 1, so as to adapt to different workload requirements and avoid excessive concentration of resources on certain module components 3 to cause performance bottlenecks, thereby maintaining the stability of the overall performance of the server body 1;

[0047] The monitoring system 500 is used to monitor the working status of each module component 3 in real time through sensors, including the performance indicators and connection status of the module component 3, and to display the current resource status information of the server body 1 and the performance indicators and connection status of the module component 3 through the display screen 20. Specific embodiment three:

[0049] This embodiment provides a resource scheduling method implemented by the server architecture described in Embodiment 1 or 2, which includes the following steps:

[0050] Sp1: Determine the demand of the server body 1, obtain the current resource status information of the server body 1 through the main control system, and send the current resource status information to the resource demand prediction system; the current resource status information includes one or more of the real-time utilization rate of various resources, the type of business being run, and the load condition;

[0051] Sp2: The resource demand prediction system predicts the resource demand based on the current resource status information and historical resource status information of the server body 1 using the trained resource prediction model, and feeds the prediction result back to the main control system, and selects the module component 3 that needs to be plugged in and out based on the prediction result;

[0052] Sp3: When you need to unplug module component 3, just unplug the corresponding module component 3;

[0053] When the module assembly 3 needs to be inserted, the selected module assembly 3 is smoothly inserted into the corresponding slot along the guide rail 4 inside the mounting seat 5. When the module assembly 3 is fully inserted, the locking mechanism is triggered to automatically lock the module assembly 3, and a signal that the module assembly 3 has been inserted and information on whether the hardware connection status of the module assembly 3 is normal are sent to the main control system;

[0054] Sp4: The request allocation system automatically obtains the module information of all inserted module components 3 from the main control system, and sets the task processing weight of each module component 3 according to the module information of all inserted module components 3 in the server body 1, and allocates different task requests to module components 3 with different task processing weights;

[0055] The task processing weight of the module component 3 can be determined according to factors such as the resource capacity and performance parameters of each module component 3. For example, the larger the resource capacity and the better the performance parameters, the stronger the business data processing capability of the module component 3, the larger the corresponding task processing weight, and the more task requests will be assigned to it subsequently;

[0056] Sp5: The resource allocation system dynamically allocates the resources of the server body 1 to different module components 3 to reallocate and optimize the resources;

[0057] Therefore, this embodiment can grasp the resource demand dynamics of the server body 1 in real time through the resource demand prediction system, so that the management personnel can prepare the adaptive module components 3 in advance. Once the resource demand changes, the function expansion and resource allocation can be quickly realized by plugging and unplugging the corresponding module components 3, ensuring that the server body 1 can run stably even during business peak periods, and effectively avoiding the risk of service performance degradation or interruption caused by resource shortages.

[0058] At the same time, module component 3 is an independent modular design, which can be flexibly plugged in and out without interrupting the operation of server body 1, greatly improving the convenience of maintenance. At the same time, the request allocation system and the resource allocation system work together to ensure that module component 3 can be seamlessly expanded within server body 1, thereby significantly enhancing the upgrade and maintenance performance of server body 1, realizing flexible allocation and efficient utilization of server body 1 resources, and ensuring that server body 1 can always meet changing business needs. Specific embodiment four:

[0060] The difference between this embodiment and the third embodiment is that the resource demand prediction system uses the following method to obtain the resource demand prediction result:

[0061] Data preparation:

[0062] Collect historical resource status information of the server body 1, including but not limited to one or more of CPU 8 usage, memory 9 occupancy, storage I / O rate, network bandwidth utilization, number of concurrent users, number of running applications, business transaction frequency, etc., and the historical resource status information constitutes a training data set for the machine learning model;

[0063] Preprocess the data, including: data cleaning to remove outliers and fill in missing data with mean; feature extraction and conversion of raw data; and data standardization or normalization to make different features comparable and avoid features with larger values ​​dominating the model training process;

[0064] Model training, including: kernel function selection, parameter adjustment and iterative training, wherein the kernel function includes a linear kernel function Polynomial kernel function ( (where d is the degree of the polynomial) and the radial basis function (RBF, K(x i , x j )=exp(-γ||x i -x j || 2 ), where γ is a kernel parameter); the parameter adjustment includes the adjustment of the penalty parameter C (controlling the degree of penalty for training error, the larger the C, the heavier the penalty for error, and the higher the complexity of the model) and the kernel function parameter (such as γ in the RBF kernel, the larger the γ, the smaller the influence range of the support vector, and the more complex the model). Specifically, the optimal parameter combination can be found through methods such as cross-validation to improve the prediction performance of the model; the iterative training process includes for a given training data set (where x i is the input feature vector, y jis the corresponding resource usage target value), the objective function can be expressed as At the same time, the constraints Where ω is the weight vector of the hyperplane, b is the bias term, is the eigenvector after kernel function mapping, ε i and is a slack variable, which adjusts ω, b, ε through iterative training i and To make the objective function converge and obtain the optimal model;

[0065] Model evaluation, which includes using methods such as cross-validation to evaluate the performance of the model. For example, k-fold cross-validation can be used to evaluate the generalization ability of the model.

[0066] Result prediction includes inputting the current resource status information of the server body 1 into the model that has completed training and evaluation, and outputting the resource demand prediction results. For example, it predicts the changing trends of resource indicators such as CPU8 utilization and memory9 occupancy in the future (such as 1 hour), and provides decision-making basis for resource management of the server body 1, such as whether resource expansion or adjustment of resource allocation strategy is needed. Specific embodiment five:

[0068] The difference between this embodiment and the third embodiment is that the request allocation system sets the resource weights of different module components 3 by the following method:

[0069] According to the module information of all the inserted module components 3 in the server body 1, the task processing weight of each module component 3 is set; for example, if the module component 3 is a computing module, the weight can be determined based on the processor performance (core number × frequency), assuming that the number of processor cores of a computing module is 8 and the frequency is 3.0GHz, and the number of processor cores of a computing module is 4 and the frequency is 2.0GHz, then the corresponding task processing weights of the two can be set to 6 (8×3.0÷4) and 4 (4×2.0÷2) respectively; if the module component 3 is a storage module, the weight can be set according to the read and write speed and capacity; if the module component 3 is a network communication module, the weight is set according to the bandwidth and port processing capacity;

[0070] Receive task requests from various applications of the server body 1, and assign different task requests to module components 3 with different weights according to the type and priority of the task requests. For example, if there are 10 computing task requests, about 8 tasks are assigned to the module component 3 with a task processing weight of 6, and 2 tasks are assigned to the module component 3 with a task processing weight of 4;

[0071] At the same time, the request allocation system resets the task processing weight of the module component 3 periodically (e.g., every few minutes) or according to specific events (e.g., sudden changes in module load, insertion of new modules or failures). For example, if it is found that the load of a certain module component 3 is continuously too high (e.g., the CPU 8 usage rate of the computing module exceeds 80% for a long time), its task processing weight is appropriately reduced; if the performance of a certain module component 3 is improved (e.g., through software optimization or hardware upgrade), its task processing weight is increased accordingly. Specific embodiment six:

[0073] The difference between this embodiment and the third embodiment is that the resource allocation system dynamically allocates resources by the following method:

[0074] The module information of each plugged-in module component 3 is obtained, and the resource allocation coefficient of each module component 3 is generated according to the module information. For example, if there are three storage modules as the module component 3, and their storage capacities are C1, C2, and C3 respectively, then the total storage capacity C total =C1+C2+C3, for module i, its corresponding resource allocation coefficient

[0075] According to the task requests from various applications of the server main body 1, the resources of the server main body 1 are dynamically allocated to different module components 3 according to the resource allocation coefficient. For example, for a storage write request, the request size is SIze request , the resource allocation coefficient of a module component 3 is Proportion i , then the storage resource Allocation allocated to the module component 3 i =Size request ×proportion i .

[0076] As the server runs, the status of the module component 3 may change, such as module failure, performance degradation, or upgrade. The resource allocation system needs to monitor these changes in real time through the monitoring system, re-evaluate the resource capacity and performance of the module, and update the allocation ratio. For example, if part of the storage space of a storage module is damaged, the resource allocation system will recalculate the remaining available capacity and adjust its allocation ratio in the total storage resources.

[0077] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "comprising a reference structure" do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0078] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A server architecture, comprising a server body (1), a cover plate (2) and a main board (7), characterized in that: The server body (1) is provided with a mounting seat (5) inside, a module assembly (3) inside the mounting seat (5), a card slot (12) is provided on the side of the module assembly (3), a guide rail (4) is provided inside the mounting seat (5), a movable groove (13) is provided inside the guide rail (4), a spring (14) is provided inside the movable groove (13), a transmission rod (15) is provided at the end of the spring (14), a rotating shaft (17) is provided inside the movable groove (13), a movable rod (16) is rotatably connected to the top of the rotating shaft (17), a card block (19) is provided at the end of the movable rod (16), a CPU (8) is provided on the surface of the mainboard (7), and a memory (9) is provided on the surface of the mainboard (7).

2. The server architecture according to claim 1, characterized in that: Guide grooves (6) are provided on both sides of the module assembly (3), and the shape and position of the guide grooves (6) are compatible with the guide rails (4).

3. The server architecture according to claim 1, characterized in that: A moving block (18) is provided at the end of the transmission rod (15), and the end of the moving block (18) is connected through the end of the guide rail (4).

4. The server architecture according to claim 1, characterized in that: One end of the movable rod (16) is rotatably connected to the surface of the transmission rod (15), and the other end of the movable rod (16) is connected to the side surface of the guide rail (4) through the connection.

5. The server architecture according to claim 1, characterized in that: The shape and position of the clamping slot (12) are matched with the clamping block (19).

6. The server architecture according to claim 1, characterized in that: A contact point is provided inside the mounting seat (5), a connecting wire (11) is provided on the side of the mounting seat (5), and an end of the connecting wire (11) is electrically connected to the main board (7).

7. The server architecture according to claim 1, characterized in that: A cooling fan (21) is provided inside the server body (1), and a power supply module (10) is provided inside the server body (1).

8. The server architecture according to claim 1, characterized in that: The server body (1) includes a main control system, a resource demand prediction system, a request allocation system, a resource allocation system and a monitoring system, wherein: The main control system comprises a main board (7), a CPU (8) and a memory (9), and is used to control the resource demand prediction system, the request allocation system and the resource allocation system as a whole, and is connected to the monitoring system to obtain current resource status information of the server body (1); The resource demand prediction system is used to predict resource demand based on current resource status information and historical resource status information of the server body (1), and determine the module components (3) that need to be plugged in or out based on the prediction results; The request allocation system is used to automatically obtain module information of all module components (3) inserted in the server body (1), set the task processing weight of each module component (3) according to the module information, and allocate different task requests to module components (3) with different weights according to task requests received from various application programs of the server body (1); The resource allocation system is used to dynamically allocate resources of the server body (1) to different module components (3) according to task requests; The monitoring system is used to track the working status of the module assembly (3) in real time.

9. A resource scheduling method implemented based on the server architecture according to any one of claims 1 to 8, characterized in that: The steps include: Sp1: collecting the current resource status information of the server body (1) through the main control system, and sending the current resource status information to the resource demand prediction system; Sp2: The resource demand prediction system predicts resource demand based on the current resource status information and historical resource status information of the server body (1) using the trained resource prediction model, feeds back the prediction results to the main control system, and selects the module components that need to be plugged in and out based on the prediction results (3); Sp3: inserting the selected module assembly (3) into the corresponding slot along the guide rail (4) inside the mounting seat (5); when the module assembly (3) is fully inserted, triggering the locking mechanism to automatically lock the module assembly (3), and sending a signal to the main control system indicating that the module assembly (3) has been inserted and information on whether the hardware connection status of the module assembly (3) is normal; Sp4: The request allocation system automatically obtains the module information of all the inserted module components (3) from the main control system, and sets the task processing weight of each module component (3) according to the module information of all the inserted module components (3) in the server body (1), and allocates different task requests to the module components (3) with different task processing weights; Sp5: The resource allocation system dynamically allocates the resources of the server body (1) to different module components (3).

10. The resource scheduling method according to claim 9, characterized in that: The resource scheduling method also includes that during the execution of steps Sp1-Sp5, the monitoring system continuously monitors the working status of each module component (3).

Images