A cross-cluster resource scheduling method and device

By transmitting the application files to be run instead of the mirror files in cross-cluster resource scheduling, the problems of excessive network bandwidth and slow transmission speed are solved, and scheduling efficiency is improved.

CN113886058BActive Publication Date: 2024-11-19CHINA UNITED NETWORK COMM GRP CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202010624216.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-07-01
Publication Date
2024-11-19
Estimated Expiration
2040-07-01

AI Technical Summary

Technical Problem

In cross-cluster resource scheduling, the transmission of mirror files occupies too much network bandwidth, slow transmission speed, and low scheduling efficiency.

Method used

Transfer files to run applications between clusters, rather than mirror files. The files to be run are obtained and sent through management nodes, and the resource support is determined by using the scheduling server to achieve cross-cluster resource scheduling.

Benefits of technology

It reduces network bandwidth usage during cross-cluster resource scheduling, improves transmission speed and scheduling efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN113886058B_ABST
    Figure CN113886058B_ABST
Patent Text Reader

Abstract

The present application discloses a cross-cluster resource scheduling method and device, which relates to the field of communications and solves the problem that the transmission of image files between clusters occupies too much network bandwidth, has slow transmission speed, and low scheduling efficiency during the existing cross-cluster resource scheduling. The cross-cluster resource scheduling method includes: obtaining the file of the application to be run; when it is determined that the remaining resources in the first cluster cannot support the operation of the application to be run, sending a resource scheduling request to the scheduling server; receiving a first identifier sent by the scheduling server; according to the first identifier, sending the file of the application to be run to the management node in the second cluster, so that the management node in the second cluster runs the application to be run according to the file of the application to be run. The cross-cluster resource scheduling method provided by the present application can reduce the network bandwidth occupied during cross-cluster resource scheduling, increase the transmission speed, and improve the scheduling efficiency because the files of the application to be run are transmitted between clusters.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of communications, and in particular to a cross-cluster resource scheduling method and device. Background Art

[0002] In order to improve data processing capabilities and data reliability, more and more applications (such as machine learning models) are deployed in clusters in a single data center (DC). As the running time increases, the applications in the cluster require more and more resources. It is possible that the resources in a single data center (that is, the resources provided by the cluster in a DC) cannot meet the application requirements, which may lead to application failure.

[0003] In the prior art, the above problem is solved by scheduling between clusters of different DCs in units of image files (such as image files in docker format). For example: when a device in cluster 1 determines that it cannot meet the needs of application 1, the device requests resources from the scheduling server; when the scheduling server determines that one or more devices in cluster 2 meet the conditions (i.e., can provide resources for the application), it sends the address of the device that meets the conditions to the device; the device sends the image file of application 1 to the device that meets the conditions, so that the device that meets the conditions can run application 1 according to the image file.

[0004] If cluster 1 and cluster 2 do not belong to the same local area network, the data transmission rate between cluster 1 and cluster 2 will be relatively low. The storage space occupied by the image file is usually relatively large. In this case, the above method will cause the image file to occupy too much network bandwidth, slow transmission speed, and low scheduling efficiency. Summary of the invention

[0005] The present application provides a cross-cluster resource scheduling method and device, which are used to solve the problems of excessive network bandwidth occupied by image files transmitted between clusters, slow transmission speed and low scheduling efficiency during existing cross-cluster resource scheduling.

[0006] In order to achieve the above objectives, this application adopts the following technical solutions:

[0007] In a first aspect, the present application provides a cross-cluster resource scheduling method, which is applied to a management node in a first cluster, and the method includes:

[0008] After the management node in the first cluster obtains the file of the application to be run, the management node determines whether the remaining resources in the first cluster can support the running of the application to be run. When it is determined that the remaining resources in the first cluster cannot support the running of the application to be run, the management node sends a resource scheduling request to the scheduling server for requesting to determine that the remaining resources support the running of the application to be run. The scheduling server sends the identifier of the management node in the second cluster that meets the resource scheduling request to the management node. The management node sends the file of the application to be run to the management node in the second cluster based on the first identifier, and the management node in the second cluster runs the application to be run based on the file of the application to be run.

[0009] It can be seen that in the cross-cluster resource scheduling method provided by the present application, what is transmitted between clusters is the file of the application to be run. Compared with the image file of the application to be run transmitted in the prior art, the storage space occupied by the file of the application to be run is relatively small, which effectively reduces the network bandwidth occupied during cross-cluster resource scheduling, improves the transmission speed, and improves the scheduling efficiency.

[0010] In a second aspect, the present application provides a cross-cluster resource scheduling device, which is applied to a management node in a first cluster. The device includes: an acquisition module, a determination module, a request module, a receiving module, and a sending module.

[0011] The acquisition module is used to acquire the file of the application to be run. The determination module is used to determine whether the remaining resources in the first cluster can support the operation of the application to be run obtained by the acquisition module. The request module is used to send a resource scheduling request to the scheduling server when the determination module determines that the remaining resources in the first cluster cannot support the operation of the application to be run. The resource scheduling request is used to request to determine the cluster whose remaining resources support the operation of the application to be run. The receiving module is used to receive a first identifier sent by the scheduling server, and the first identifier is an identifier of a management node in the second cluster that meets the resource scheduling request. The sending module is used to send the file of the application to be run to the management node in the second cluster according to the first identifier received by the receiving module, so that the management node in the second cluster runs the application to be run according to the file of the application to be run.

[0012] In a third aspect, the present application provides a cross-cluster resource scheduling device, the cross-cluster resource scheduling device comprising a memory and a processor. The memory and the processor are coupled. The memory is used to store a computer program code, the computer program code comprising a computer instruction. When the processor executes the computer instruction, the cross-cluster resource scheduling device performs the cross-cluster resource scheduling method as described in the first aspect and any possible design thereof.

[0013] In a fourth aspect, the present application provides a computer-readable storage medium, which stores instructions. When the computer-readable storage medium runs on a cross-cluster resource scheduling device, the device executes the cross-cluster resource scheduling method described in the first aspect and any possible design method thereof.

[0014] In a fifth aspect, the present application provides a computer program product, which includes computer instructions. When the computer instructions are executed on a cross-cluster resource scheduling device, the cross-cluster resource scheduling device executes the cross-cluster resource scheduling method as described in the first aspect and any possible design method thereof.

[0015] For the specific description of the second to fifth aspects and their various implementations in the present application, reference may be made to the detailed description in the first aspect and its various implementations; and for the beneficial effects of the second to fifth aspects and their various implementations, reference may be made to the beneficial effects analysis in the first aspect and its various implementations, which will not be repeated here.

[0016] These and other aspects of the present application will become more apparent from the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0018] Figure 1 It is a schematic diagram of a cross-cluster resource scheduling system in the prior art;

[0019] Figure 2 This is a schematic diagram of a cross-cluster resource scheduling system in an embodiment of the present application;

[0020] Figure 3 A schematic diagram of the hardware structure of a cross-cluster resource scheduling device provided in an embodiment of the present application;

[0021] Figure 4 A schematic diagram of a cross-cluster resource scheduling method provided in an embodiment of the present application;

[0022] Figure 5 A schematic diagram of the structure of a cross-cluster resource scheduling device provided in an embodiment of the present application Figure 1 ;

[0023] Figure 6 A schematic diagram of the structure of a cross-cluster resource scheduling device provided in an embodiment of the present application Figure 2 . DETAILED DESCRIPTION

[0024] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0025] The terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, unless otherwise specified, "plurality" means two or more.

[0026] As the running time increases, the resources in a single data center will not be able to meet the application needs, and it is necessary to schedule between clusters in different DCs to improve data processing capabilities and data reliability.

[0027] In the prior art, when cross-cluster resource scheduling is performed, scheduling is performed between clusters of different DCs in units of mirror files. For example, Figure 1 As shown, the cross-cluster resource scheduling system includes: cluster 1 (including device a and device b), cluster 2 (including device c and device d) and a scheduling server. Cluster 1 and cluster 2 are connected to the scheduling server, and the transmission between cluster 1 and cluster 2 is through the network. The resource scheduling process between cluster 1 and cluster 2 can be: 1. When a device in cluster 1 (such as device a) determines that cluster 1 cannot meet the resource requirements of application 1, the device a sends a resource request to the scheduling server. 2. The scheduling server sends a query request to cluster 2 to query whether the resources meet the requirements of application 1. 3. When it is determined that one or more devices in cluster 2 (such as device d) meet the conditions, device d returns a confirmation to the scheduling server. 4. The scheduling server obtains the address of device d and sends the address of device d to device a. 5. Device a sends the docker format image file of application 1 to device d according to the address of device d. In this way, device d can run application 1 according to the image file.

[0028] Since the image file in Docker format includes the operating system for running application 1 and the dependent libraries of application 1, the storage space occupied by the image file is usually large. If the above cluster 1 and cluster 2 do not belong to the same local area network, the image file will occupy too much network bandwidth, the transmission speed will be slow, and the scheduling efficiency will be low.

[0029] In response to the above problems, an embodiment of the present application provides a cross-cluster resource scheduling method. When it is determined that the remaining resources in the first cluster cannot support the operation of the application to be run, the management node in the first cluster sends the file of the application to be run to the management node in the second cluster that meets the resource scheduling request. Compared with the prior art, in which the image file of the application to be run is transmitted between clusters during cross-cluster resource scheduling, the present application transmits the file of the application to be run between clusters. Since the storage space occupied by the file of the application to be run is relatively small, it can reduce the network bandwidth occupied during cross-cluster resource scheduling, increase the transmission speed, and improve the scheduling efficiency.

[0030] The cross-cluster resource scheduling method provided in the embodiment of the present application is applied to a cross-cluster resource scheduling system. Figure 2 The architecture of the cross-cluster resource scheduling system is shown in FIG. Figure 2 As shown, the cross-cluster resource scheduling system includes: cluster 101, cluster 102, and scheduling server 103. Cluster 101 and cluster 102 are connected to scheduling server 103, and information is transmitted between cluster 101 and cluster 102 through a network. Of course, in actual applications, the cross-cluster resource scheduling system is not limited to including the above-mentioned devices, but may also include other devices, such as a third cluster.

[0031] A cluster is a computer system that is connected by a group of loosely integrated computer software and / or hardware to work together to complete computing tasks. In a sense, they can be regarded as a single computer. The individual computers in a cluster are usually called nodes, usually connected by a local area network, but there are other possible connection methods. Cluster computers are usually used to improve the computing speed and / or reliability of a single computer. In general, cluster computers have a much higher performance-price ratio than single computers, such as workstations or supercomputers.

[0032] Any of the above clusters includes a management node and multiple computing nodes. The management node (also called a processing process / processing thread) is used to manage the resource allocation of each computing node in the cluster and to communicate with the scheduling server 103, such as requesting the scheduling server 103 to schedule resources in other clusters for it.

[0033] The management node can be implemented by a device in the cluster, or it can be a functional module in a device in the cluster, or it can be implemented by at least two devices in the cluster (in this case, the management node may include multiple sub-modules, and different sub-modules are located on different devices). The embodiments of the present application do not make specific limitations on this.

[0034] The basic hardware structures of the management node, computing node and scheduling server 103 are similar, including Figure 3 The components included in the communication device shown are described below. Figure 3Taking the communication device shown as an example, the hardware structure of the management node, the computing node and the scheduling server 103 is introduced.

[0035] like Figure 3 As shown, the communication device may include at least one processor 301 , a communication line 302 , a memory 303 , and a communication interface 304 .

[0036] Specifically, the processor 301 is used to execute the computer-executable instructions stored in the memory 303, so as to implement the steps or actions of the terminal.

[0037] The processor 301 may be a chip. For example, it may be a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a system on chip (SoC), a central processor unit (CPU), a network processor (NP), a digital signal processor (DSP), a microcontroller unit (MCU), a programmable logic device (PLD), or other integrated chips.

[0038] The communication line 302 is used to transmit information between the processor 301 and the memory 303 .

[0039] The memory 303 is used to store and execute computer-executable instructions, and the execution is controlled by the processor 301 .

[0040] The memory 303 may be independent and connected to the processor via the communication line 302. The memory 303 may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), and enhanced synchronous dynamic random access memory (ESDRAM). It should be noted that memory of the systems and devices described herein is intended to comprise, without being limited to, these and any other suitable types of memory.

[0041] The communication interface 304 is used to communicate with other devices or communication networks, where the communication network can be Ethernet, radio access network (RAN), or wireless local area network (WLAN).

[0042] It should be pointed out that Figure 3 The structure shown in the figure does not constitute a limitation on the communication device. Figure 3 In addition to the components shown, the communication device may include more or fewer components than shown, or combine certain components, or arrange the components differently.

[0043] Combine the following Figure 2 The cross-cluster resource scheduling system and Figure 3 The communication device shown describes the cross-cluster resource scheduling method provided in an embodiment of the present application.

[0044] like Figure 4 As shown, the cross-cluster resource scheduling method provided in the embodiment of the present application includes:

[0045] S401: A management node in a first cluster obtains a file of an application to be run.

[0046] Specifically, the application to be run is an application that needs to be run in the first cluster. The embodiment of the present application does not limit the format and embodiment of the application to be run. Exemplarily, the application to be run is a machine learning model, and the file of the application to be run is the model itself of the machine learning model.

[0047] The first cluster here may be the cluster 101 described above, or may be the cluster 102.

[0048] Optionally, the method for the management node in the first cluster to obtain the file of the application to be run may be: receiving the file of the application to be run sent by the terminal.

[0049] Exemplarily, when the application to be run is a machine learning model, the developer of the machine learning model uploads the machine learning model to the first cluster through the terminal.

[0050] Optionally, the method for the management node in the first cluster to obtain the file of the application to be run may also be: receiving the file of the application to be run sent by the management node in any other cluster except the first cluster.

[0051] Exemplarily, when the first cluster is cluster 101, the management node in cluster 101 can receive the file of the application to be run sent by the management node in cluster 102. Similarly, when the first cluster is cluster 102, the management node in cluster 102 can receive the file of the application to be run sent by the management node in cluster 101.

[0052] Optionally, after obtaining the file of the application to be run, the management node in the first cluster may save the file of the application to be run.

[0053] S402: The management node in the first cluster determines whether the remaining resources in the first cluster can support the running of the application to be run.

[0054] From the above description, it can be seen that the management node is used to manage the resource allocation of each computing node in the cluster. In other words, the management node can determine the remaining resources of each computing node in the cluster and determine the remaining resources of the cluster.

[0055] After obtaining the file of the application to be run, the management node in the first cluster can determine the size of the resources required to run the file of the application to be run. In order to ensure the normal operation of the application to be run, the management node in the first cluster needs to first determine whether the remaining resources in the cluster (i.e., the first cluster) can support the operation of the application to be run.

[0056] Specifically, the management node in the first cluster determines whether the remaining resources in the first cluster can support the running of the application to be run according to the size of the resources required for running the application to be run and the size of the remaining resources in the first cluster.

[0057] If the size of the remaining resources in the first cluster is greater than or equal to the size of the resources required to run the application to be run, it means that the remaining resources in the first cluster can support the running of the application to be run. In this case, S403-S404 are continued after S402. If the size of the remaining resources in the first cluster is less than the size of the resources required to run the application to be run, it means that the remaining resources in the first cluster cannot support the running of the application to be run. In this case, S405-S409 are continued after S402.

[0058] S403: The management node in the first cluster generates an image file of the application to be run according to the file of the application to be run.

[0059] Exemplarily, when the application to be run is a machine learning model, the management node in the first cluster analyzes the dependent libraries required by the machine learning model, generates a docker image file for the machine learning model, and installs the dependent libraries required by the machine learning model in the image file.

[0060] Optionally, after generating the image file of the application to be run, the management node in the first cluster may save the generated image file of the application to be run.

[0061] S404: The management node in the first cluster runs the application to be run in the first cluster according to the image file.

[0062] S405 . The management node in the first cluster sends a resource scheduling request to the scheduling server 103 . The resource scheduling request is used to request a cluster to determine the remaining resources that support the running of the application to be run.

[0063] Specifically, the resource scheduling request includes the size of resources required to run the application to be run, and the scheduling server 103 determines the cluster whose remaining resources support the running of the application to be run according to the size of resources required to run the application to be run.

[0064] Optionally, when the size of remaining resources of a certain cluster is greater than or equal to the size of resources required by the application to be run, the cluster is a cluster whose remaining resources support the running of the application to be run.

[0065] S406: The management node in the first cluster receives the first identifier sent by the scheduling server 103, where the first identifier is the identifier of the management node in the second cluster that meets the resource scheduling request.

[0066] The second cluster is any cluster in the cross-cluster resource scheduling system that is different from the first cluster.

[0067] Exemplarily, the first identifier may be an application programming interface (API) address of a management node in the second cluster, or may be a number of a management node in the second cluster.

[0068] S407: The management node in the first cluster sends the file of the application to be run to the management node in the second cluster according to the first identifier, so that the management node in the second cluster runs the application to be run according to the file of the application to be run.

[0069] S408: After receiving the file of the application to be run, the management node in the second cluster generates an image file of the application to be run according to the file of the application to be run.

[0070] S409: The management node in the second cluster runs the application to be run in the second cluster according to the image file.

[0071] Compared with the prior art in which image files of applications to be run are transmitted between clusters during cross-cluster resource scheduling, the cross-cluster resource scheduling method provided in the present application transmits files of applications to be run between clusters. Since the storage space occupied by the files of applications to be run is relatively small, the network bandwidth occupied during cross-cluster resource scheduling can be reduced, the transmission speed can be increased, and the scheduling efficiency can be improved.

[0072] The above mainly introduces the solution provided by the embodiment of the present application from the perspective of the method. In order to realize the above functions, it includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should easily realize that, in combination with the units and algorithm steps of each example described in the embodiments disclosed herein, the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present application.

[0073] The embodiment of the present application further provides a cross-cluster resource scheduling device. The cross-cluster resource scheduling device can be applied to the management node in the first cluster, and can also be applied to the management node in the second cluster.

[0074] like Figure 5 FIG. 5 is a schematic diagram of a cross-cluster resource scheduling device 500 provided in an embodiment of the present application. The cross-cluster resource scheduling device 500 is used to execute Figure 4The cross-cluster resource scheduling method shown in the figure. The cross-cluster resource scheduling device 500 may include an acquisition module 501 , a determination module 502 , a request module 503 , a receiving module 504 , and a sending module 505 .

[0075] The acquisition module 501 is used to acquire the file of the application to be run. Figure 4 , the acquisition module 501 can be used to execute step S401.

[0076] The determination module 502 is used to determine whether the remaining resources in the first cluster can support the operation of the application to be run. Figure 4 , the determination module 502 can be used to execute step S402.

[0077] The request module 503 is used to send a resource scheduling request to the scheduling server 103 when the determination module determines that the remaining resources in the first cluster cannot support the operation of the application to be run. The resource scheduling request is used to request a cluster whose remaining resources support the operation of the application to be run. Figure 4 , the request module 503 can be used to execute step S405.

[0078] The receiving module 504 is configured to receive a first identifier sent by the scheduling server 103, where the first identifier is an identifier of a management node in the second cluster that satisfies the resource scheduling request. Figure 4 , the receiving module 504 can be used to execute step S406.

[0079] The sending module 505 is configured to send the file of the application to be run to the management node in the second cluster according to the first identifier, so that the management node in the second cluster runs the application to be run according to the file of the application to be run. Figure 4 , the sending module 505 can be used to execute step S407.

[0080] like Figure 6 As shown, an embodiment of the present application provides a cross-cluster resource scheduling device 500, which may also include: a generating module 506 and an operating module 507.

[0081] The generating module 506 is configured to generate an image file of the application to be run according to the file of the application to be run when the determining module 502 determines that the remaining resources in the first cluster support the running of the application to be run. Figure 4 , the generation module 506 can be used to execute step S403.

[0082] The running module 507 is used to run the application to be run according to the image file generated by the generating module. Figure 4 , the generation module 506 can be used to execute step S404.

[0083] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0084] In actual implementation, the acquisition module 501, the determination module 502, the request module 503, the receiving module 504, the sending module 505, the generation module 506 and the operation module 507 can be composed of Figure 3 The processor 301 shown calls the program code in the memory 303 to implement. The specific execution process can be referred to Figure 4 The description of the cross-cluster resource scheduling method shown in FIG. 1 is not repeated here.

[0085] Another embodiment of the present application further provides a computer-readable storage medium, which stores computer instructions. When the computer instructions are executed on a cross-cluster resource scheduling device, the cross-cluster resource scheduling device executes each step executed by the cross-cluster resource scheduling device in the method flow shown in the above method embodiment.

[0086] In another embodiment of the present application, a computer program product is also provided. The computer program product includes instructions. When the instructions are executed on a cross-cluster resource scheduling device, the cross-cluster resource scheduling device executes each step executed by the cross-cluster resource scheduling device in the method flow shown in the above method embodiment.

[0087] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0088] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0089] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0090] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0091] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0092] The above are only specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A cross-cluster resource scheduling method, characterized in that: Applied to a management node in a first cluster, the cross-cluster resource scheduling method includes: Get the file of the application to be run; When it is determined that the remaining resources in the first cluster cannot support the running of the application to be run, sending a resource scheduling request to a scheduling server, wherein the resource scheduling request is used to request a cluster whose remaining resources support the running of the application to be run; receiving a first identifier sent by the scheduling server, where the first identifier is an identifier of a management node in the second cluster that satisfies the resource scheduling request; Sending the file of the application to be run to the management node in the second cluster according to the first identifier, so that the management node in the second cluster runs the application to be run according to the file of the application to be run; When it is determined that the remaining resources in the first cluster support the operation of the application to be run, an image file of the application to be run is generated according to the file of the application to be run, including: when the application to be run is a machine learning model, the management node in the first cluster analyzes the dependent library required by the machine learning model, generates an image file in a docker format for the machine learning model, and installs the dependent library in the image file; According to the image file, the application to be run is run.

2. The method according to claim 1, characterized in that The step of obtaining the file of the application to be run includes: Receiving the file of the application to be run sent by the terminal; or, Receive the file of the application to be run sent by the management node in the third cluster.

3. A cross-cluster resource scheduling device, the device is applied to a management node in a first cluster, characterized in that: The device comprises: The acquisition module is used to obtain the files of the application to be run; A determination module, used to determine whether the remaining resources in the first cluster can support the running of the application to be run obtained by the acquisition module; a request module, configured to send a resource scheduling request to a scheduling server when the determining module determines that the remaining resources in the first cluster cannot support the operation of the application to be run, wherein the resource scheduling request is used to request a cluster whose remaining resources support the operation of the application to be run; A receiving module, configured to receive a first identifier sent by the scheduling server, where the first identifier is an identifier of a management node in the second cluster that satisfies the resource scheduling request; a sending module, configured to send the file of the application to be run obtained by the obtaining module to the management node in the second cluster according to the first identifier received by the receiving module, so that the management node in the second cluster runs the application to be run according to the file of the application to be run; a generation module, configured to generate an image file of the application to be run according to the file of the application to be run when the determination module determines that the remaining resources in the first cluster support the run of the application to be run, including: when the application to be run is a machine learning model, the management node in the first cluster analyzes the dependency library required by the machine learning model, generates an image file in a doc ker format for the machine learning model, and installs the dependency library in the image file; A running module is used to run the application to be run according to the image file generated by the generating module.

4. The device according to claim 3, characterized in that The acquisition module is specifically used for: Receiving the file of the application to be run sent by the terminal; or, Receive the file of the application to be run sent by the management node in the third cluster.

5. A cross-cluster resource scheduling device, characterized in that: The cross-cluster resource scheduling device includes a memory and a processor; the memory and the processor are coupled; the memory is used to store computer program code, and the computer program code includes computer instructions; when the processor executes the computer instructions, the cross-cluster resource scheduling device executes the cross-cluster resource scheduling method as described in any one of claims 1-2.

6. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and when the computer-readable storage medium runs on a cross-cluster resource scheduling device, the device executes the cross-cluster resource scheduling method according to any one of claims 1 to 2.