Resource scheduling method and device, electronic equipment, storage medium and edge cloud system

Abstract

The application provides a resource scheduling method and device, electronic equipment, storage medium and edge cloud system. The resource scheduling method comprises: in the case of identifying the resource requirement of a target workload, determining whether there is an edge node in the edge cloud cluster that meets the resource requirement; in the case that there is an edge node in the edge cloud cluster that meets the resource requirement, deploying the target workload on the edge node that meets the resource requirement; in the case that there is no edge node in the edge cloud cluster that meets the resource requirement at a specified time, or according to at least one of the resource requirement, the resource state of the edge cloud cluster, the running monitoring information of the edge cloud cluster and the current traffic corresponding to the target workload, performing global scheduling on the load deployment state of the edge cloud cluster. The technical scheme of the application can improve the resource utilization rate of the edge cloud cluster.

Description

Technical Field

[0001] This application relates to the field of edge computing technology, and in particular to a resource scheduling method, apparatus, electronic device, storage medium and edge cloud system. Background Technology

[0002] After years of development in virtualization and container technologies, servers in edge computing data centers are becoming increasingly larger, and pooling technologies for various resources (such as computing, storage, and network resources) are becoming more mature. Meanwhile, workloads (applications) are becoming increasingly miniaturized; a single workload may require only a small amount of computing, storage, and network resources. Therefore, workload and server scheduling is becoming increasingly important. This necessitates the use of a resource scheduling system to allocate a server to run a workload when it requires certain resources. However, existing resource scheduling systems can only passively meet resource demands, resulting in poor scheduling efficiency and an inability to handle complex scheduling requirements. Summary of the Invention

[0003] This application provides a resource scheduling method, apparatus, electronic device, storage medium, and edge cloud system to address the problems existing in the prior art.

[0004] In a first aspect, embodiments of this application provide a resource scheduling method, including:

[0005] Once the resource requirements of the target workload are identified, determine whether there are edge nodes in the edge cloud cluster that meet the resource requirements;

[0006] If there are edge nodes in the edge cloud cluster that meet the resource requirements, the target workload will be deployed on the edge nodes that meet the resource requirements.

[0007] If, at a specified time, there are no edge nodes in the edge cloud cluster that meet the resource requirements, the load deployment status of the edge cloud cluster is globally scheduled based on at least one of the resource requirements, the resource status of the edge cloud cluster, the operation monitoring information of the edge cloud cluster, and the current traffic corresponding to the workload; if there are no edge nodes in the edge cloud cluster that meet the resource requirements, the global scheduling includes resource scheduling based on the resource requirements of the target workload.

[0008] Secondly, embodiments of this application provide a resource scheduling apparatus, including:

[0009] The node matching module is used to determine whether there are edge nodes in the edge cloud cluster that meet the resource requirements of the target workload when the resource requirements are identified.

[0010] The first scheduling module is used to deploy the target workload to the edge node that meets the resource requirements when there is an edge node in the edge cloud cluster that meets the resource requirements.

[0011] The second scheduling module is used to perform global scheduling of the load deployment status of the edge cloud cluster based on at least one of the following: the resource requirements, the resource status of the edge cloud cluster, the operation monitoring information of the edge cloud cluster, and the current traffic corresponding to the target workload, when there are no edge nodes in the edge cloud cluster that meet the resource requirements at a specified time or when there are no edge nodes in the edge cloud cluster that meet the resource requirements. When global scheduling includes scheduling of the target workload, global scheduling includes resource scheduling based on the resource requirements of the target workload.

[0012] Thirdly, embodiments of this application provide an electronic device, including a memory, a processor, and a computer program stored in the memory, wherein the processor implements the method provided in any embodiment of this application when executing the computer program.

[0013] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the method provided in any embodiment of this application.

[0014] Compared with the prior art, this application has the following advantages:

[0015] According to the technical solution provided in the embodiments of this application, in addition to matching and scheduling workloads and edge nodes based on resource requirements, a global scheduling operation on the load deployment status of the edge cloud cluster can be performed periodically or triggered as a supplement. In the global scheduling, it can be implemented based on at least one of the following: the resource requirements of the workload, the resource status of the edge cloud cluster, the operation monitoring information of the edge cloud cluster, and the current traffic corresponding to the target workload. This allows for scheduling of the overall load deployment status on each edge node of the edge cloud cluster, finding more suitable edge nodes for each workload, including the target workload, dynamically adjusting the resource occupancy of each edge node, making the resource usage of each edge node more balanced, and improving the overall resource utilization of the edge cloud cluster.

[0016] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application, it can be implemented according to the contents of the specification. In order to make the above and other objects, features and advantages of this application more obvious and understandable, specific embodiments of this application are given below. Attached Figure Description

[0017] In the accompanying drawings, unless otherwise specified, the same reference numerals throughout the various drawings denote the same or similar parts or elements. These drawings are not necessarily drawn to scale. It should be understood that these drawings depict only some embodiments according to this application and should not be construed as limiting the scope of this application.

[0018] Figure 1 This is a schematic diagram of the network architecture of the edge cloud cluster in the embodiments of this application;

[0019] Figure 2 A flowchart illustrating a resource scheduling method provided in an embodiment of this application;

[0020] Figure 3 This is a partial flowchart illustrating another resource scheduling method provided in an embodiment of this application;

[0021] Figure 4 This is a schematic diagram of the structural framework of a resource scheduling device provided in an embodiment of this application;

[0022] Figure 5 A schematic diagram of the structural framework of an electronic device provided in this application embodiment; and

[0023] Figure 6 This is a schematic diagram of the structural framework of an edge cloud system provided in an embodiment of this application. Detailed Implementation

[0024] In the following description, only certain exemplary embodiments are briefly described. As those skilled in the art will recognize, the described embodiments can be modified in various ways without departing from the concept or scope of this application. Therefore, the drawings and description are considered to be exemplary in nature and not restrictive.

[0025] One resource scheduling method in related technologies allocates servers that can meet the resource requirements of a specific workload. If no server that can meet the resource requirements is found, resource scheduling cannot be performed. This is a relatively passive resource scheduling method and also a local scheduling method. The workload can be an application running on servers, computing devices, etc., that consumes resources such as CPU (Central Processing Unit), memory, hard disk, and network I / O (Input / Output).

[0026] In practical resource scheduling applications, different time points and scenarios are often involved. For example, at time point a, there are n (n > 1) servers, each with sufficient resources to meet the resource requirements of the workload to be scheduled. At time point b, there are m (m > 2) servers, but only 2 servers meet the resource requirements of the workload to be scheduled. Scheduling needs to be implemented at different time points based on the current resource supply situation. Furthermore, at different time points, workloads that were originally consuming resources may no longer be consuming resources for various reasons. Over time, dynamic changes may occur, such as server fragmentation. Different workloads on the same server may have high CPU requirements, while workloads on other servers may have low CPU requirements, or there may be no workloads on other servers at all. To make workloads run more efficiently, it is necessary to adjust the servers hosting the workloads at different time points based on the overall resource occupancy and supply situation to achieve specific scheduling goals, such as more balanced resource usage, lower server power consumption, and more stable workload operation.

[0027] This application provides a novel resource scheduling scheme that enables more flexible and efficient scheduling of workloads deployed on servers, achieving scheduling goals such as more balanced resource utilization, lower server power consumption, and more stable workload operation. The applicable scenarios for this technical solution are described below:

[0028] The resource scheduling scheme in this application can be used for resource scheduling of edge cloud clusters. In contrast to the concept of a central cloud (a data center with large-scale resources and concentrated location), an edge cloud cluster is a data center built on edge infrastructure and located closer to the edge, possessing computing, storage, networking, and security capabilities. Edge infrastructure may include edge devices such as wireless data centers (or clusters), core network equipment, base stations, gateways, servers, computing devices, and storage devices, as well as the corresponding network operating environment.

[0029] like Figure 1As shown, an edge cloud cluster can include multiple edge nodes. Each edge node can be a single edge device at an edge location, or a collection of multiple edge devices at an edge location, such as a data center or other collection of edge devices located in the same geographical area. Edge devices on different edge nodes, or different edge devices within the same edge node, can be of the same type or different types. Edge nodes can provide at least one resource, such as computing resources, storage resources, and network resources. Computing resources can include CPUs, GPUs (Graphics Processing Units), etc.; storage resources can include memory, hard drives, etc.; and network resources can include bandwidth. The resource scheduling scheme of this application can schedule resources based on the aforementioned resources provided by the edge nodes. Edge cloud clusters can be applied to network service architectures such as ENS (Edge Node Service) and CDN (Content Delivery Network), and can be applied to various practical application scenarios such as e-commerce, logistics, and the Internet of Things, providing cloud computing services to users in various scenarios by running workloads.

[0030] To facilitate understanding of the technical solutions of the embodiments of this application, the relevant technologies of the embodiments of this application are described below. The following relevant technologies are optional solutions and can be combined with the technical solutions of the embodiments of this application in any way, and all of them fall within the protection scope of the embodiments of this application.

[0031] This application provides a resource scheduling method, such as... Figure 2 As shown, the resource scheduling method 200 may include the following steps S201 to S203:

[0032] S201, if the resource requirements of the target workload are identified, determine whether there are edge nodes in the edge cloud cluster that meet the resource requirements; if there are edge nodes in the edge cloud cluster that meet the above resource requirements, proceed to step S202; if there are no edge nodes in the edge cloud cluster that meet the above resource requirements, proceed to step S203.

[0033] A target workload can be a workload already deployed on an edge node and consuming the resources of that edge node. In an edge cloud cluster, when multiple workloads are distributed across multiple edge nodes, each workload can serve as a target workload when it has resource requirements. The resource requirements of a target workload can include the amount of resources it needs, such as how many CPUs, how much memory, and how much bandwidth are required to deploy it. The network architecture of the edge cloud cluster can be referred to the previous description and will not be repeated here.

[0034] S202 deploys the target workload on edge nodes that meet the resource requirements of the target workload.

[0035] In one example, the target workload can be deployed on an edge node that provides the amount of resources required for the target workload. For example, if the target workload requires 4 CPUs and 512 GB of memory, and edge node A provides 8 CPUs and 512 GB of memory, then the target workload can be deployed on edge node A.

[0036] S203, based on at least one of the following: resource requirements, resource status of the edge cloud cluster, operation monitoring information of the edge cloud cluster, and current traffic corresponding to the target workload, perform global scheduling of the load deployment status of the edge cloud cluster.

[0037] Step S203 can also be executed at a specified time. The specified time can be set according to actual needs. For example, time points with fixed intervals can be used as the specified time, such as time points with 1-hour intervals, such as 1 o'clock, 2 o'clock, 3 o'clock, etc. Alternatively, time points with non-fixed intervals can be used as the specified time, such as 1 o'clock, 2 o'clock, 4 o'clock, etc. If the edge cloud cluster does not have edge nodes that meet the above resource requirements, it could be that the edge cloud cluster does not have edge nodes that can provide the amount of resources required for the target workload. For example, if the target workload requires 4 CPUs and 512GB of memory, the edge nodes in the target network may not be able to provide 4 CPUs and 512GB of memory. The current traffic corresponding to the target workload can be determined by the access volume of the user's terminal.

[0038] The workload deployment status can include at least one of the following: the deployment location of each workload in the edge cloud cluster, i.e., which edge node it is specifically deployed on; the amount of resources reserved by each edge node for the workloads deployed therein; the amount of resources occupied by the workloads; the number of replicas of the workloads; etc. The aforementioned global scheduling can be a scheduling method that adjusts the workload deployment status within a preset edge cloud cluster. Adjustments to the workload deployment status can include at least one of the following: adjusting the deployment location of each workload; adjusting the reservations in each edge node; adjusting the number of replicas of the workloads; or evictring abnormal workloads on certain edge nodes. Adjusting the deployment location of a workload can be switching the workload's deployment location from the current edge node to another edge node, or switching the workload's deployment location from the currently reserved resources in the current edge node to other reserved resources in the current edge node.

[0039] If no edge nodes in the edge cloud cluster meet the above resource requirements, the global scheduling mentioned above may include resource scheduling based on the resource requirements of the target workload. At a specified time, the global scheduling may include resource scheduling based on the resource requirements of the target workload. For example, it may supplement the non-global scheduling (i.e., the scheduling in step S202) by adjusting the reserved resources or replica count of the target workload, or by executing other scheduling that does not conflict with the non-global scheduling. At a specified time, the global scheduling may also exclude resource scheduling based on the resource requirements of the target workload, but include scheduling based on the resource requirements of other workloads besides the target workload.

[0040] Based on step S203 above, a global scheduling operation is performed on the load deployment status of the edge cloud cluster according to at least one of the resource requirements, the resource status of the edge cloud cluster, the operation monitoring information of the edge cloud cluster, and the current traffic corresponding to the target workload. The global scheduling operation can be executed on a schedule or triggered when there are no edge nodes in the edge cloud cluster that meet the above resource requirements (i.e., when scheduling cannot be performed in the scheduling method described in step S202).

[0041] When executed on a scheduled basis, it can supplement the passive scheduling method in step 202. It can be executed based on at least one of the following: resource requirements, resource status of the edge cloud cluster, operation monitoring information of the edge cloud cluster, and current traffic corresponding to the target workload. It can actively perform global scheduling on the overall load deployment status of each edge node of the edge cloud cluster, find more suitable edge nodes for each workload, including the target workload, dynamically adjust the resource occupancy of each edge node, make the resource use of each edge node more balanced, and improve the resource utilization of each edge node. When executed by trigger, it can supplement the local scheduling method in step S202. It can comprehensively consider the influencing factors of resource scheduling based on at least one of the following: resource requirements, resource status of the edge cloud cluster, operation monitoring information of the edge cloud cluster, and current traffic corresponding to the target workload. In this way, it can perform global scheduling on the overall load deployment status of each edge node of the edge cloud cluster, find more suitable edge nodes for each workload, including the target workload, make the resource use of each edge node more balanced, and improve the resource utilization of each edge node.

[0042] Based on the above step S203, the resource scheduling method provided in this application embodiment can also reduce fragmented resources (or resource fragments) in each edge node and improve the operational stability of the workload.

[0043] In step S202 above, when deploying the target workload to an edge node that meets the resource requirements, the edge node to be deployed can be any edge node within the entire edge cloud cluster, or any edge node within a specified range of nodes in the edge cloud cluster. In one example, the specified range of nodes can be the range of nodes formed by edge nodes in the same geographical area, such as the range of nodes formed by edge nodes in the same data center or the range of nodes formed by edge nodes in the same city.

[0044] In one implementation, step S202 can be implemented by a passive scheduler in the resource scheduling system, and step S203 can be implemented by an active scheduler in the resource scheduling system. The passive scheduler and the active scheduler can be distributed on different devices (e.g., servers) or integrated on the same device. Based on the above method provided by the embodiments of this application, the passive scheduler and the active scheduler can work together to improve the efficiency and accuracy of resource scheduling and make the resource use of each edge node more balanced.

[0045] In one implementation (hereinafter referred to as Implementation 1), the resource status of the edge cloud cluster may include: the resource occupancy status of the workload in the edge cloud cluster and the resource supply status of each edge node in the edge cloud cluster.

[0046] Correspondingly, in step S203 above, the load deployment status of the edge cloud cluster is globally scheduled based on the resource requirements of the target workload, the resource status of the edge cloud cluster, the operation monitoring information of the edge cloud cluster, and the traffic corresponding to the workload. This includes at least one of the following scheduling processes: adjusting the reserved resources for the workload based on the resource occupancy status; adjusting the number of replicas of the workload based on the current traffic; and adjusting the deployment status of the workload on each edge node in the edge cloud cluster based on at least one of the resource requirements of the target workload, the resource supply status of each edge node, and the operation monitoring information of the edge cloud cluster. If there are no edge nodes in the edge cloud cluster that meet the resource requirements, the adjusted workloads may include the target workload. At a specified time, the adjusted workloads may or may not include the target workload.

[0047] Based on implementation method one, the global scheduling method in step S203 above can be implemented through a variety of scheduling processing methods, which can improve the flexibility of scheduling various workloads of the edge cloud cluster and help improve the overall efficiency of resource scheduling.

[0048] In one implementation (hereinafter referred to as Implementation Two), the aforementioned resource occupancy status may include: a first resource occupancy rate of the workload in the current edge node. In one example, the aforementioned resource occupancy status may include the first resource occupancy rate of each workload (including the target workload) in the current edge node. Correspondingly, in Implementation One, adjusting the reserved resources of the workload according to the resource occupancy status may include: for a given workload, determining whether the first resource occupancy rate corresponding to the workload is greater than a preset first occupancy rate threshold; if the first resource occupancy rate is greater than the first occupancy rate threshold, increasing the reserved resources of the workload, i.e., resource expansion.

[0049] The first resource occupancy threshold represents the upper limit of the first resource occupancy rate of a workload in its current edge node. The specific value of the first resource occupancy threshold can be set according to actual needs, empirical values, etc., for example, it can be set to 90%, 80%, or other larger values. When the first resource occupancy rate of a workload is greater than the upper limit, it indicates that the workload is consuming too many node resources, and the currently reserved resources may not be sufficient to meet the workload's resource requirements. Therefore, it is necessary to increase the resources reserved for the workload to ensure its normal operation. If the matching method described in step S201 fails to find resources sufficient to meet the requirements of the target workload, the resources of the target workload can be expanded to improve the matching degree between the target workload and the current edge node.

[0050] In another implementation (hereinafter referred to as implementation three), in the above implementation one, adjusting the reserved resources of the workload according to the resource occupancy status may include: for a workload, determining whether the first resource occupancy rate corresponding to the workload is less than a preset second occupancy rate threshold, and if the first resource occupancy rate is less than the second occupancy rate threshold, reducing the reserved resources of the workload, i.e., resource shrinking.

[0051] The second resource utilization threshold can be lower than the first resource utilization threshold. The second threshold represents the lower limit of the first resource utilization of a workload in its current edge node. The specific value of the second threshold can be set according to actual needs, experience, etc., for example, it can be set to 10%, 20%, or other smaller values. When the first resource utilization of a workload is lower than the lower limit, it means that the workload is using fewer resources than is currently reserving, which will cause some resource waste. The resources reserved for this workload need to be reduced to minimize waste and free up resources for other workloads that require them, thus balancing the resources reserved for each workload.

[0052] In another implementation, when adjusting the reserved resources for the workload based on resource occupancy status, the relationship between a first resource occupancy rate and a first occupancy rate threshold, or between the first resource occupancy rate and a second occupancy rate threshold, can be compared to simultaneously determine whether to expand or shrink resources. Resource expansion and shrinking can also reduce fragmented resources, facilitating more rational resource scheduling. In one example, resource expansion and shrinking can be achieved through vertical elastic scaling.

[0053] In one example, resource scaling for the target workload can be achieved in two ways. One is to directly scale up the resources for the target workload. This method is applicable when there are excess resources available for scaling up in the edge node where the target workload resides, such as other unused resources in the edge node. The other method is to first scale down the resources of other workloads in the edge node where the target workload resides, where the first resource utilization rate is less than a second utilization rate threshold, freeing up the corresponding resources, and then scale up the resources for the target workload. This method is applicable when there are insufficient resources available for scaling up in the edge node where the target workload resides. The method for scaling down the resources of other workloads is similar to the method for scaling down the resources of the target workload, and will not be elaborated here.

[0054] In this application embodiment, the reserved resources for the workload can be in the form of virtual machines or containers. In one example, multiple virtual machines or containers can be partitioned in an edge node, and the workload runs in one of the partitioned virtual machines or containers, which is the reserved resource for the workload. Based on the resource reservation method of virtual machines or containers, the technical solution of this application is applicable to both IaaS (Infrastructure as a Service) and container-based resource scheduling scenarios, exhibiting good versatility and eliminating the need to design separate resource scheduling schemes for the two scenarios.

[0055] In one implementation (hereinafter referred to as Implementation Four), in Implementation One above, adjusting the number of replicas of a workload based on the current traffic corresponding to the workload includes: for a workload, determining whether the difference between the current traffic corresponding to the workload and the historical traffic corresponding to the workload is less than a preset first traffic threshold; if the difference is less than the first traffic threshold, reducing the number of replicas of the workload. Historical traffic can be the traffic at the historical time point most recent to the current time point; if the difference is greater than the first traffic threshold, the current state can be maintained or other processing can be performed, such as judgment and processing based on a second traffic threshold, which will be described later.

[0056] The first traffic threshold can be negative and can be used to measure the degree to which the current traffic corresponding to a workload has decreased compared to historical traffic. For a workload, if the difference between the current traffic and historical traffic is less than the first traffic threshold, it indicates a significant reduction in current traffic compared to historical traffic, and the resource requirements are correspondingly reduced. In this case, the number of replicas for the workload can be reduced, thereby reducing the amount of resources required by the workload. If the matching method described in step S201 fails to find a match that meets the resource requirements of the target workload, and if the current traffic corresponding to the target workload has also decreased, the number of replicas can be reduced to decrease the resource requirements of the target workload, thereby improving the matching degree between the target workload and the current edge node. In one example, the adjustment of the number of replicas can be achieved through horizontal elastic scaling.

[0057] In one implementation (hereinafter referred to as Implementation Five), when expanding the resource capacity of the edge node where a workload is currently located, i.e., increasing the reserved resources of the workload, adjusting the number of replicas of the workload according to the current traffic corresponding to the workload may include: determining whether the difference between the current traffic corresponding to the workload and the historical traffic corresponding to the workload is greater than a preset second traffic threshold; if the difference is greater than the second traffic threshold, the number of replicas of the workload may be increased; if the difference is less than the second traffic threshold, the current state may be maintained or other processing may be performed, such as judgment and processing based on a first traffic threshold, and the judgment and processing process may refer to the preceding content.

[0058] The second traffic threshold can be a positive number and can be used to measure how much the current traffic for a workload has increased compared to historical traffic. For a workload, if the difference between the current traffic and historical traffic is greater than the second traffic threshold, it indicates that the current traffic has increased significantly compared to historical traffic, and the resource requirements have also increased accordingly. In this case, the number of workload replicas can be increased to meet the current traffic needs.

[0059] In one implementation (hereinafter referred to as Implementation Six), the resource supply status of each edge node may include a second resource utilization rate of each edge node occupied by workloads, such as the ratio of the amount of resources occupied by each workload on the edge node to the total amount of resources provided by the edge node.

[0060] Correspondingly, in the first embodiment described above, adjusting the deployment status of each edge node in the edge cloud cluster based on at least one of the resource requirements of the target workload, the resource supply status of each edge node, and the operation monitoring information of the edge cloud cluster may include: for each edge node, determining whether the second resource occupancy rate of the edge node is greater than a third occupancy rate threshold; if the second resource occupancy rate is greater than the third occupancy rate threshold, evict some workloads from the edge node; if the second resource occupancy rate is greater than the third occupancy rate threshold, maintain the current state.

[0061] The third resource utilization threshold can be used to measure the degree to which the resources of an edge node are occupied. The specific value of the third resource utilization threshold can be set according to actual needs, experience, etc., for example, it can be set to 90%, 80%, or other larger values. When the second resource utilization rate of an edge node is greater than the third resource utilization threshold, it indicates that the resources of the edge node are being over-utilized. In order to avoid the edge node from running at full load, some resources need to be released. This can be achieved by driving some of the workload on the edge node.

[0062] In one example, for the edge node where the target workload is currently located, if its second resource utilization rate is greater than the third utilization rate threshold, other workloads can be evicted first to ensure the operation of the target workload that currently requires resources. In another example, for the edge node where the target workload is currently located, if its second resource utilization rate is greater than the third utilization rate threshold, according to the priority order of each workload, if other workloads have lower priority, they can be evicted first; if the edge node where the target workload is currently located has lower priority, the target workload can be evicted first. In yet another example, for other edge nodes besides the edge node where the target workload is currently located, if their second resource utilization rate is greater than the third utilization rate threshold, lower priority workloads can be evicted according to the priority order of each workload.

[0063] In one implementation (hereinafter referred to as Implementation Seven), resource requirements may include priority information of the target workload, such as the priority order of the target workload among various workloads. In one example, resource requirements may include priority information for each workload (including the target workload). Priorities can be preset according to the type and function of the workload; for example, the priority of the management program can be set to be higher than that of the business program.

[0064] Correspondingly, in the first embodiment described above, as Figure 3As shown, the deployment status of each workload on each edge node in the edge cloud cluster is adjusted based on at least one of the resource requirements of the target workload, the resource supply status of each edge node, and the operation monitoring information of the edge cloud cluster. This includes: S301, determining whether the priority of the target workload is higher than the priority of the workloads in the candidate edge nodes based on the priority information of the target workload; if the priority of the target workload is higher than the priority of the workloads in the candidate edge nodes, proceeding to step S302; otherwise, maintaining the current state; S302, evictting at least some workloads from the candidate edge nodes; S303, deploying the target workload on the evictped candidate edge nodes.

[0065] A candidate edge node can be an edge node other than the one currently hosting the target workload. The total resources of this edge node may be sufficient to meet the resource requirements of the target workload, but the remaining resources may be insufficient due to other workloads occupying them. In one example, the target workload requires 256GB of memory. Edge node B has a total memory of 512GB, which is sufficient for the target workload. However, because other workloads are running on edge node B, occupying more than half of its memory, the remaining memory on edge node B is insufficient to meet the target workload's memory requirements. Therefore, edge node B can be considered a candidate edge node.

[0066] Implementation method seven is applicable when neither the edge node where the target workload is currently located nor other edge nodes have sufficient remaining resources for the target workload. In this case, implementation method seven allows the higher-priority target workload to preempt resources from other lower-priority workloads, thus prioritizing the operation of the higher-priority target workload. In one example, the target workload on edge node A is a management program. A service program is deployed on an edge node B, which is located outside the management program's edge node. The service program's node resources are sufficient to meet the management program's resource requirements. Therefore, the service program can be evicted from edge node B to free up resources for the management program on edge node A to migrate and use.

[0067] In one example, both Implementation 6 and Implementation 7 can be used as specific implementations of rescheduling, which means readjusting the deployment of workloads on resources under certain conditions.

[0068] In one implementation (hereinafter referred to as implementation eight), the resource scheduling method 200 may further include: when the resource requirement of the evicted workload is identified, determining whether there is an edge node in the edge cloud cluster that meets the resource requirement and has not evicted the workload; and if there is an edge node in the edge cloud cluster that meets the resource requirement and has not evicted the workload, deploying the workload to the edge node.

[0069] For evicted workloads, such as those evicted according to Implementation Six and Implementation Seven, if their resource requirements are identified, they can be deployed on edge nodes that have not been evicted according to Implementation Eight, avoiding deployment on edge nodes that have evicted the workload, thereby avoiding resource conflicts with other workloads again during rescheduling and deployment.

[0070] In one implementation (hereinafter referred to as Implementation Nine), the deployment status of each workload and each edge node in the edge cloud cluster is adjusted according to at least one of the resource requirements of the target workload, the resource supply status of each edge node, and the operation monitoring information of the edge cloud cluster. This includes: determining whether there are abnormal edge nodes in the edge cloud cluster based on the operation monitoring information of the edge cloud cluster; for edge nodes with abnormalities, at least some workloads are removed from the edge node, or at least some workloads in the edge node are migrated to other edge nodes.

[0071] In Implementation Nine, the abnormal situation may include at least one of the following: a fault (e.g., hard drive failure), a high risk of failure, or multiple restarts of the edge node. For edge nodes that are faulty or at high risk of failure, evicting or migrating at least part of the workload can reduce the power consumption of the edge node and prevent it from causing a fault or a larger failure. In one example, a faulty edge node may be marked as tainted, and no new workloads will be deployed on the marked edge node. Implementation Nine may be combined with any of the foregoing implementations.

[0072] In one implementation (hereinafter referred to as Implementation Ten), the resource requirements of the target workload may include at least one constraint of affinity conditions and distributed deployment conditions. The affinity condition may refer to the need to deploy the target workload on a specific type of edge node, such as a CPU-intensive edge node or a memory-intensive edge node. The distributed deployment condition may refer to distributing multiple target workloads across different edge nodes when there are multiple target workloads, avoiding aggregation on a single edge node. This prevents anomalies on a single edge node from affecting the operation of too many target workloads. The distributed deployment condition is applicable to SMS sending scenarios, where the workload used for sending SMS messages is deployed across different edge nodes, which can better ensure the normal operation of the SMS sending function.

[0073] Correspondingly, in the first embodiment described above, adjusting the deployment status of each workload on each edge node in the edge cloud cluster based on at least one of the resource requirements of the target workload, the resource supply status of each edge node, and the operational monitoring information of the edge cloud cluster can include: adjusting the deployment location of the target workload to an edge node that satisfies at least one of the above constraints. For example, the deployment location of a target workload with high CPU requirements can be adjusted to a CPU-intensive edge node, and the deployment location of a target workload with high memory requirements can be adjusted to a memory-intensive edge node. Multiple target workloads can be distributed across different edge nodes, thereby enabling more precise matching and scheduling of target workloads and edge nodes. The tenth embodiment can be combined with any of the embodiments described above.

[0074] In one implementation (hereinafter referred to as Implementation Eleven), the resource scheduling method 200 may further include: when there are no edge nodes in the edge cloud cluster that meet the resource requirements of the target workload at a specified time or when there are no edge nodes in the edge cloud cluster that meet the resource requirements of the target workload, detecting whether there are abnormal workloads in each workload; and when there are abnormal workloads, expelling the abnormal workloads from the edge nodes where the abnormal workloads are located.

[0075] Abnormal workloads can include at least one of the following: workloads that have reached the end of their lifecycle, and workloads that cannot provide the corresponding service capabilities. The processing described in Implementation Method 11 above can clear abnormal workloads, freeing up more resources on the edge nodes for other workloads that require them, thus improving the resource utilization of the edge nodes.

[0076] Based on the same technical concept, embodiments of this application also provide a resource scheduling device, such as... Figure 4 As shown, the resource scheduling device 400 includes: a node matching module 401, a first scheduling module 402, and a second scheduling module 403.

[0077] The node matching module 401 is used to determine whether there are edge nodes in the edge cloud cluster that meet the resource requirements when the resource requirements of the target workload are identified; the first scheduling module 402 is used to deploy the target workload to the edge nodes that meet the resource requirements when there are edge nodes in the edge cloud cluster that meet the resource requirements; the second scheduling module 403 is used to perform global scheduling of the load deployment status of the edge cloud cluster based on at least one of the resource requirements, the resource status of the edge cloud cluster, the operation monitoring information of the edge cloud cluster, and the current traffic corresponding to the target workload when there are no edge nodes in the edge cloud cluster that meet the resource requirements at a specified time or when there are no edge nodes in the edge cloud cluster that meet the resource requirements.

[0078] If no edge nodes in the edge cloud cluster meet the above resource requirements, the global scheduling mentioned above may include resource scheduling based on the resource requirements of the target workload. At a specified time, the global scheduling may include scheduling of the target workload, for example, supplementing the non-global scheduling (i.e., the scheduling in step S202) by adjusting the reserved resources or replica count of the target workload, or executing other scheduling that does not conflict with the non-global scheduling. At a specified time, the global scheduling may also exclude resource scheduling based on the resource requirements of the target workload and instead perform resource scheduling based on the resource requirements of other workloads besides the target workload.

[0079] In one example, the first scheduling module 402 can act as a passive scheduler, and the second scheduling module 403 can act as an active scheduler.

[0080] In one implementation, the resource status of the edge cloud cluster may include: the resource occupancy status of the workload in the edge cloud cluster and the resource supply status of each edge node in the edge cloud cluster. Correspondingly, the second scheduling module 403 may include at least one of the first adjustment unit, the second adjustment unit, and the third adjustment unit.

[0081] The first adjustment unit can be used to adjust the reserved resources for the workload based on the resource occupancy status. The second adjustment unit can be used to adjust the number of replicas for the workload based on the current traffic. The third adjustment unit can be used to adjust the deployment status of each edge node in the edge cloud cluster based on at least one of the resource requirements of the target workload, the resource supply status of each edge node, and the operation monitoring information of the edge cloud cluster.

[0082] In one implementation, the aforementioned resource occupancy status may include: a first resource occupancy rate of the workload in the current edge node. Correspondingly, the first adjustment unit may be used to: for a workload, determine whether the first resource occupancy rate corresponding to the workload is greater than a preset first occupancy rate threshold; if the first resource occupancy rate is greater than the first occupancy rate threshold, increase the reserved resources of the workload, i.e., expand resource capacity.

[0083] In one implementation, the second adjustment unit can be used to: for a workload, determine whether the difference between the current traffic corresponding to the workload and the historical traffic corresponding to the workload is less than a preset first traffic threshold; if the difference is less than the first traffic threshold, reduce the number of replicas of the target workload. Historical traffic can be the traffic at the most recent historical time point; if the difference is greater than the first traffic threshold, the current state can be maintained or other processing can be performed. The first traffic threshold is a negative number.

[0084] In one implementation, for a workload, when resource expansion is performed on the edge node where the workload is currently located, i.e., the reserved resources for the workload are increased, the second adjustment unit can be used to: determine whether the difference between the current traffic corresponding to the workload and the historical traffic corresponding to the workload is greater than a preset second traffic threshold; if the difference is greater than the second traffic threshold, the number of replicas of the workload can be increased; if the difference is less than the second traffic threshold, the current state can be maintained or other processing can be performed. The second traffic threshold is a positive number.

[0085] In one implementation, the resource supply status of each edge node may include a second resource occupancy rate of each edge node occupied by workloads. Correspondingly, the third adjustment unit may be used to: for each edge node, determine whether the second resource occupancy rate of the edge node is greater than a third occupancy rate threshold; if the second resource occupancy rate is greater than the third occupancy rate threshold, evict some workloads from the edge node; if the second resource occupancy rate is greater than the third occupancy rate threshold, maintain the current state.

[0086] In one implementation, resource requirements may include priority information of the target workload. Correspondingly, the third adjustment unit may be used to: determine whether the priority of the target workload is higher than the priority of the workloads in the candidate edge nodes based on the priority information of the target workload; if the priority of the target workload is higher than the priority of the workloads in the candidate edge nodes, evict at least some workloads from the candidate edge nodes and deploy the target workload to the evictped candidate edge nodes.

[0087] In one implementation, the third adjustment unit can be used to: determine whether there are abnormal edge nodes in the edge cloud cluster based on the operation monitoring information of the edge cloud cluster; for the abnormal edge nodes, evict at least some of the workload in the edge node, or migrate at least some of the workload in the edge node to other edge nodes.

[0088] In one implementation, the first scheduling module 402 can also be used to: when the resource requirement of the evicted workload is identified, determine whether there is an edge node in the edge cloud cluster that meets the resource requirement and has not evicted the workload; if there is an edge node in the edge cloud cluster that meets the resource requirement and has not evicted the workload, deploy the workload to the edge node.

[0089] In one implementation, the resource requirements of the target workload may include at least one constraint: affinity conditions and distributed deployment conditions. Correspondingly, the third adjustment unit may be used to adjust the deployment location of the target workload to an edge node that satisfies at least one of the above constraints.

[0090] In one implementation, the third scheduling module 403 may further include an eviction unit, which can be used to: detect whether there is an abnormal workload in each workload when there is no edge node in the edge cloud cluster that meets the resource requirements of the target workload at a specified time or when there is no edge node in the edge cloud cluster that meets the resource requirements of the target workload; and eviction of the abnormal workload from the edge node where the abnormal workload is located when there is an abnormal workload.

[0091] The functions of each module and unit in the resource scheduling device provided in this application embodiment can be found in the corresponding description in the above method, and they have corresponding beneficial effects, which will not be repeated here.

[0092] Based on the same technological concept, such as Figure 5 As shown, the electronic device includes a memory 501 and a processor 502. The memory 501 stores a computer program that can run on the processor 502. When the processor 502 executes the computer program, it implements the method described in the above embodiments. The number of memories 501 and processors 502 can be one or more.

[0093] The electronic device also includes:

[0094] Communication interface 503 is used to communicate with external devices and perform data exchange and transmission.

[0095] If the memory 501, processor 502, and communication interface 503 are implemented independently, they can be interconnected via a bus to communicate with each other. This bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 5 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0096] Optionally, in a specific implementation, if the memory 501, processor 502, and communication interface 503 are integrated on a single chip, then the memory 501, processor 502, and communication interface 503 can communicate with each other through an internal interface.

[0097] Based on the same technical concept, embodiments of this application also provide an edge cloud system, such as... Figure 6 As shown, the edge cloud system 600 may include a resource scheduling system 601 and an edge cloud cluster 602. The resource scheduling system 601 is used to schedule resources in the edge cloud cluster 602, and the resource scheduling system 601 can be deployed on the electronic device provided in this embodiment. The edge cloud cluster 602 includes multiple edge nodes, which can communicate with the electronic device provided in this embodiment. A specific network architecture of the continuous cloud network 602 can be referred to... Figure 1 .

[0098] This application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method provided in this application.

[0099] This application also provides a chip, which includes a processor for calling and executing instructions stored in a memory, causing a communication device with the chip installed to perform the method provided in this application.

[0100] This application also provides a chip, including: an input interface, an output interface, a processor, and a memory. The input interface, output interface, processor, and memory are connected through an internal connection path. The processor is used to execute code in the memory. When the code is executed, the processor is used to execute the method provided in the application embodiment.

[0101] It should be understood that the aforementioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. General-purpose processors can be microprocessors or any conventional processor. It is worth noting that the processor can be a processor supporting Advanced Reduced Instruction Set Machines (ARM) architecture.

[0102] Further, optionally, the aforementioned memory may include read-only memory and random access memory. The memory may be volatile memory or non-volatile memory, or may include both. Non-volatile memory may include read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory may include random access memory (RAM), which serves as an external cache. By way of example, but not limitation, many forms of RAM are available. Examples include Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), and Direct Rambus RAM (DR RAM).

[0103] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions according to this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another.

[0104] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of those different embodiments or examples.

[0105] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "a plurality of" means two or more, unless otherwise explicitly specified.

[0106] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process. Furthermore, the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functionality involved.

[0107] The logic and / or steps described in the flowchart or otherwise herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus or device (such as a computer-based system, a processor-included system or other system that can fetch and execute instructions from, an instruction execution system, apparatus or device).

[0108] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. All or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware, the program being stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiments.

[0109] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. This storage medium can be a read-only memory, a disk, or an optical disk, etc.

[0110] The above description is merely an exemplary embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various variations or substitutions within the technical scope described in this application, and these should all be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A resource scheduling method, characterized in that, include: Once the resource requirements of the target workload are identified, determine whether there are edge nodes in the edge cloud cluster that meet the resource requirements; If there are edge nodes in the edge cloud cluster that meet the resource requirements, the target workload will be deployed on the edge nodes that meet the resource requirements. If, at a specified time or if no edge node in the edge cloud cluster meets the resource requirements, the load deployment status of the edge cloud cluster is globally scheduled based on at least one of the resource requirements, the resource status of the edge cloud cluster, the operational monitoring information of the edge cloud cluster, and the current traffic corresponding to the workload. If no edge node in the edge cloud cluster meets the resource requirements, the global scheduling includes resource scheduling based on the resource requirements of the target workload. The load deployment status includes multiple states shown below: the deployment location of each workload in the edge cloud cluster, the amount of resources reserved by each edge node for the workload deployed therein, and the number of workload replicas. The global scheduling includes multiple adjustments to the load deployment status of the edge cloud cluster shown below: adjusting the deployment location of each workload, adjusting the reserved resources in each edge node, and adjusting the number of workload replicas.

2. The resource scheduling method according to claim 1, characterized in that, The resource status of the edge cloud cluster includes: the resource occupancy status of the workload in the edge cloud cluster and the resource supply status of each edge node in the edge cloud cluster. The global scheduling of the load deployment status of the edge cloud cluster is based on at least one of the resource requirements, the resource status of the edge cloud cluster, the operation monitoring information of the edge cloud cluster, and the current traffic corresponding to the workload, including various adjustment processes shown below: Adjust the reserved resources for the workload based on the resource occupancy status; Adjust the number of workload replicas based on the current traffic; Based on at least one of the resource requirements, the resource supply status, and the operation monitoring information, the deployment status of the workload on each edge node in the edge cloud cluster is adjusted.

3. The resource scheduling method according to claim 2, characterized in that, The resource occupancy status includes: the first resource occupancy rate of the workload in the current edge node; The step of adjusting the reserved resources for the workload based on the resource occupancy status includes: For a workload, determine whether the first resource utilization rate corresponding to the workload is greater than a preset first utilization rate threshold. If the first resource utilization rate is greater than the first utilization rate threshold, increase the reserved resources for the workload.

4. The resource scheduling method according to claim 2, characterized in that, The adjustment of the number of workload replicas based on the current traffic includes: For a workload, determine whether the difference between the current traffic corresponding to the workload and the historical traffic corresponding to the workload is less than a preset first traffic threshold; the first traffic threshold is a negative number. If the difference is less than the first traffic threshold, the number of replicas of the workload is reduced.

5. The resource scheduling method according to claim 3, characterized in that, For a workload, given an increase in the reserved resources for that workload, adjusting the number of replicas for the workload based on the current traffic includes: Determine whether the difference between the current traffic of the workload and the historical traffic corresponding to the workload is greater than a preset second traffic threshold; the second traffic threshold is a positive number. If the difference is greater than the second traffic threshold, the number of replicas of the workload is increased.

6. The resource scheduling method according to claim 2, characterized in that, The resource supply status includes a second resource utilization rate of each edge node occupied by workload; The adjustment of the deployment status of the workload on each edge node in the edge cloud cluster based on the resource supply status includes: For each edge node, determine whether the second resource occupancy rate of that edge node is greater than the third occupancy rate threshold; If the second resource occupancy rate is greater than the third occupancy rate threshold, a portion of the workload is evicted from the edge node.

7. The resource scheduling method according to claim 2, characterized in that, The resource requirements include priority information for the target workload; The adjustment of the deployment status of the workload on each edge node in the edge cloud cluster according to the resource requirements includes: Based on the priority information, determine whether the priority of the target workload is higher than the priority of the workloads in the candidate edge nodes; If the priority of the target workload is higher than the priority of the workloads in the candidate edge nodes, at least a portion of the workloads are evictd from the candidate edge nodes. The target workload is deployed on the candidate edge nodes after being evicted.

8. The resource scheduling method according to claim 6 or 7, characterized in that, Also includes: If the resource requirements of the evicted workload are identified, determine whether there are edge nodes in the edge cloud cluster that meet the resource requirements and have not evicted the workload. If there is an edge node in the edge cloud cluster that meets the resource requirements and has not evicted the workload, then the workload shall be deployed on that edge node.

9. The resource scheduling method according to any one of claims 2-7, characterized in that, The step of adjusting the deployment status of the workload on each edge node in the edge cloud cluster based on the operational monitoring information includes: Based on the operational monitoring information, determine whether there are any abnormal edge nodes in the edge cloud cluster; For edge nodes that exhibit abnormalities, at least a portion of the workload is evicted from that edge node or at least a portion of the workload in that edge node is migrated to other edge nodes.

10. The resource scheduling method according to any one of claims 2-7, characterized in that, The resource requirements include at least one of the affinity conditions and distributed deployment conditions; The adjustment of the deployment status of the workload on each edge node in the edge cloud cluster according to the resource requirements includes: The deployment location of the target workload is adjusted to an edge node that satisfies at least one of the constraints.

11. The resource scheduling method according to any one of claims 2-7, characterized in that, Also includes: If there are no edge nodes in the edge cloud cluster that meet the resource requirements at the specified time, detect whether there are abnormal workloads in each workload; In the event of an abnormal workload, the abnormal workload is expelled from the edge node where it is located.

12. A resource scheduling device, characterized in that, include: The node matching module is used to determine whether there are edge nodes in the edge cloud cluster that meet the resource requirements of the target workload when the resource requirements are identified. The first scheduling module is used to deploy the target workload to the edge node that meets the resource requirements when there is an edge node in the edge cloud cluster that meets the resource requirements. The second scheduling module is used to perform global scheduling of the load deployment status of the edge cloud cluster based on at least one of the following: the resource requirements, the resource status of the edge cloud cluster, the operation monitoring information of the edge cloud cluster, and the current traffic corresponding to the target workload, when there are no edge nodes in the edge cloud cluster that meet the resource requirements at a specified time or when there are no edge nodes in the edge cloud cluster that meet the resource requirements. When there are no edge nodes in the edge cloud cluster that meet the resource requirements, the global scheduling includes scheduling of the target workload. The load deployment status includes multiple states shown below: the deployment location of each workload in the edge cloud cluster, the amount of resources reserved by each edge node for the workload deployed therein, and the number of replicas of the workload. The global scheduling includes multiple adjustments shown below: adjusting the deployment location of each workload, adjusting the reserved resources in each edge node, and adjusting the number of replicas of the workload.

13. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory, wherein the processor, when executing the computer program, implements the resource scheduling method according to any one of claims 1-11.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the resource scheduling method according to any one of claims 1-11.

Images