WO2015184612A1

WO2015184612A1 - Resource scheduling method and apparatus

Info

Publication number: WO2015184612A1
Application number: PCT/CN2014/079233
Authority: WO
Inventors: 雍文远
Original assignee: 华为技术有限公司
Priority date: 2014-06-05
Filing date: 2014-06-05
Publication date: 2015-12-10
Also published as: CN105519075A

Abstract

A resource scheduling method and apparatus. The implementation of the method comprises: after a virtual network function (VNF) receives a resource allocation request, acquiring a current VNF load growth trend; if the current VNF load growth trend is a rising trend, selecting at least one VM from selectable VMs to allocate a resource, to make load of the selectable VMs balanced. By selecting a VM to allocate a resource in a case in which network load is in a growth trend to make load of VMs balanced, load of the VMs can be kept at a relatively reasonable load level, so that it is ensured that a request can be processed in time; in a case in which a quantity of requests bursts, the VMs can all share new resource requirements, so that a VM is prevented from an overload phenomenon, request processing efficiency is improved, and a time delay is reduced, thereby improving quality of service; and impact resistance of a network and system security are improved.

Description

Resource scheduling method, and device

Technical field

The present invention relates to the field of communications technologies, and in particular, to a resource scheduling method and apparatus. Background technique

On a cloud computing platform, there are virtual machines (VMs) in the system host. Each VM has certain computing power, memory, and network capabilities.

Each application (APP, Application) runs on a set of virtual machines and provides services by processing a series of external requests. Usually, the resource scheduling module distributes requests from the network to multiple VMs, and multiple VMs perform a large amount of processing. The number of requests that each VM can process in a unit time is certain, so if processing is required per unit time, The number of requests is large, so there will be more VMs.

The virtualization of the network function refers to the function of the communication network node (ie, the node in the communication network) running on the VM of the cloud platform as the APP software, and providing the communication network node. The communication network node may be a Network Controller (RNC), a Serving GPRS Support Node (SGSN), or the like. Are functions of the virtual network node of a network called virtualization function (VNF, Virtual Network Function) ₀ Since VNF APP by software running on the virtual machine implementation, and thus will usually VNF considered identical APP meaning.

As the number of requests processed by each application changes continuously, the load of application processing will continue to change. When the request for processing of the application increases, and the currently used VM cannot be processed in time, then one or more VMs need to be added to the application to process more requests, so that the user's request can be processed in time, and the processing resource is increased. The process of adding VMs to satisfy more processing requests is Scale Out; in contrast, when the number of requests that need to be processed by the application drops, the application does not need so much processing resources to deal with, which requires release. Some processing resources (release VM), the released VM can be used for other applications, which can improve the utilization of processing resources and reduce the cost of use, the process is scale in.

In the Scale Out phase, the above solution allocates VM resources for processing requests, or directly increases The VM is then allocated in the added VM, or any VM with available resources is selected for allocation; therefore, the VM with higher load is prone to overload of processing capability, resulting in the request being unable to be processed or the processing delay is increased, and the quality of service is low. When the number of requests increases suddenly, the ability to withstand network attacks is low, and the system security is low. Summary of the invention

The embodiments of the present invention provide a resource scheduling method, and a device, which are used to improve network impact resistance and system security; improve request processing efficiency and reduce delay, thereby improving service quality.

An embodiment of the present invention provides a resource scheduling method, including:

After the virtualized network function receives the resource allocation request, the VNF obtains the current VNF load growth trend;

If the current VNF load growth trend is an upward trend, at least one VM allocation resource is selected in the optional VM to balance the optional VM load.

In conjunction with the implementation of the aspect, in a first possible implementation, the selecting, by the VM, the at least one VM allocates resources, and the balancing of the optional VM load includes:

According to the load of the VM, the VM allocation resources are selected according to the load of the VM from low to high; or, according to the VM load, the VM with the lowest load in the optional VM is selected.

In a second possible implementation manner, the method further includes: if the current VNF load growth trend is a downward trend, selecting at least one VM allocation resource in the optional VM, so that The load tends to concentrate in the optional VM.

In conjunction with the second possible implementation of the aspect, in a third possible implementation, the selecting at least one VM allocates resources in the optional VM, so that the load tends to be concentrated in the optional VM. Includes:

According to the load of the VM, the VM allocation resources are selected according to the load of the VM from high to low; or, according to the load of the VM, the VM with the highest load among the optional VMs is selected.

In the fourth possible implementation manner, the obtaining the current VNF load growth trend includes: an implementation on the one hand, a first, a second, or a third possible implementation on the one hand: Obtain resources; historical statistics of % consumption, and determine the current VNF load growth trend based on the historical statistical information.

In conjunction with the fourth possible implementation on the one hand, in the fifth possible implementation, before receiving the resource allocation request, the method further includes:

The resource consumption at each time point is counted, and the time period in which the VNF load growth trend is an upward trend and the time period in which the VNF load growth trend is a downward trend are determined according to the statistically obtained resource consumption at each time point.

A second aspect of the present invention provides a resource scheduling apparatus, including:

a request receiving unit, configured to receive a resource allocation request;

a trend obtaining unit, configured to acquire a current VNF load growth trend after the request receiving unit receives the resource allocation request;

The scheduling control unit is configured to select at least one VM allocation resource in the optional VM to balance the optional VM load if the current VNF load growth trend acquired by the trend obtaining unit is an upward trend.

In combination with the implementation of the two aspects, in a first possible implementation manner, the scheduling control unit is configured to perform the sorting according to the load of the VM, and select the VM to allocate resources according to the load of the VM from low to high; or, according to the VM The load is sorted, and the VM with the lowest load in the optional VM is selected.

In combination with the implementation of the second aspect, in a second possible implementation manner, the scheduling control unit is further configured to: if the current VNF load growth trend obtained by the trend acquiring unit is a downward trend, At least one VM is selected from the VM to allocate resources such that the load tends to concentrate in the optional VM.

In conjunction with the second possible implementation of the second aspect, in a third possible implementation, the scheduling control unit is configured to perform the sorting according to the load of the VM, and select the VM to allocate resources according to the load of the VM from high to low. Or, sort according to the load of the VM, and select the VM with the highest load among the optional VMs.

In combination with the implementation of the two aspects, the first, the second, or the third possible implementation of the second aspect, in the fourth possible implementation, the trend obtaining unit is configured to obtain historical statistics of the resource consumption. And determining a current VNF load growth trend according to the historical statistical information. With reference to the fourth possible implementation of the two aspects, in a fifth possible implementation, the resource scheduling apparatus further includes:

A statistical unit for counting resource consumption at each time point;

The trend calculation unit is configured to determine, according to the resource consumption amount of each time point obtained by the statistical unit, a time period in which the VNF load growth trend is an upward trend, and a time period in which the VNF load growth trend is a downward trend.

It can be seen from the above technical solutions that the embodiments of the present invention have the following advantages: By selecting a VM to allocate resources and balancing the VMs in a network load growth trend, the load of each VM can be maintained. A reasonable load level ensures that requests can be processed in a timely manner, and in the case of a sudden increase in the number of requests, each VM can share the newly added resource requirements, avoiding the overload of the VM, improving the request processing efficiency and reducing the delay, thereby improving Quality of service; improved network impact resistance and system security. DRAWINGS

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following drawings will be briefly described in the description of the embodiments, and it is obvious that the drawings in the following description are only some embodiments of the present invention, Those skilled in the art can also obtain other drawings based on these drawings without paying any inventive labor.

1 is a schematic flowchart of a method according to an embodiment of the present invention;

2 is a schematic diagram of a network load cycle change according to an embodiment of the present invention;

3 is a schematic diagram of resource allocation in a VM according to an embodiment of the present invention;

4 is a schematic structural diagram of a resource scheduling apparatus according to an embodiment of the present invention;

FIG. 5 is a schematic structural diagram of a resource scheduling apparatus according to an embodiment of the present invention;

6 is a schematic structural diagram of a resource scheduling apparatus according to an embodiment of the present invention;

FIG. 7 is a schematic structural diagram of a resource scheduling apparatus according to an embodiment of the present invention;

FIG. 8 is a schematic structural diagram of a server used as a resource scheduling apparatus according to an embodiment of the present invention. detailed description

In order to make the objects, technical solutions and advantages of the present invention more clear, the present invention will be described below with reference to the accompanying drawings. The invention is further described in detail, and it is apparent that the described embodiments are only a part of the embodiments of the invention, rather than all embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present invention without creative efforts are within the scope of the present invention.

An embodiment of the present invention provides a resource scheduling method, as shown in FIG. 1 , including:

After receiving the resource allocation request, the current VNF load growth trend is obtained. In the embodiment of the present invention, the resource allocation request is a request for requesting allocation of resources, and the request is used as a request for triggering resource allocation, and its expression form. It may be: requesting access, requesting to establish a new session, and the like, and any request for allocating processing resources in the VM, and the specific form of the present invention is not limited in the embodiment of the present invention. The resource allocation request can be used as a condition for triggering the current VNF load growth trend. In the case where the VNF load growth trend is historical statistical information, the VNF load growth trend can be obtained in advance; and the current VNF load growth trend is triggered. After the condition, the current VNF load growth trend is determined based on historical statistical information.

The above-mentioned VNF load growth trend is a parameter for indicating the change of the VM resource consumption of the entire system; the VNF load growth trend in the embodiment of the present invention can be represented by a graph, for example: a corresponding graph of time and load, this curve The graph will show a trend of rising or falling. It can be predicted by the current time whether the current VNF load growth trend is rising or falling. The VNF load growth trend in the embodiment of the present invention can also be expressed by the control value, and the setting load is high. The value and the load are low. When the current load is higher than the load, the current VNF load growth trend is considered to be an upward trend. When the current load is lower than the load low value, the current VNF load growth trend is considered to be a downward trend. The parameter used to indicate the change of the VM resource consumption of the entire system may also have other representations, which is not limited by the embodiment of the present invention.

102: If the current VNF load growth trend is an upward trend, select at least one VM allocation resource in the optional VM to balance the above optional VM load.

It can be understood that, when there is no optional VM, the solution of the embodiment of the present invention adds a VM and allocates resources in the newly added VM. In extreme cases, when a VM cannot be added, a system overload will occur.

In the embodiment of the present invention, when the network load is increasing, the VM is allocated resources, and the load balancing of each VM is performed, so that the load of each VM can be maintained at a relatively reasonable load level, thereby ensuring that the request can be processed in time. And in the case that the number of requests suddenly increases, each VM can share the newly added resource requirements, avoiding the overload phenomenon of the VM, and improving the processing efficiency of the request. Time, thereby improving service quality; improving network impact resistance and system security. The embodiment of the present invention further provides an optional implementation manner of selecting at least one VM allocation resource in the optional VM to balance the optional VM load, as follows: Selecting at least one VM allocation in the optional VM. Resources that enable the above optional VM load balancing include:

In the above, regarding the implementation of selecting at least one VM allocation resource in the optional VM to balance the above-mentioned optional VM load, there may be other alternatives, for example: sorting according to the load of the VM, selecting a VM with a lower load. As an optional VM set, when there is a resource allocation request, arbitrarily select a VM to allocate resources from the optional VM set; or, set a load threshold, and use a VM below the load threshold as an optional VM set. When there is a resource allocation request, the VM allocation resource is arbitrarily selected from the optional VM set; therefore, the above implementation of selecting at least one VM allocation resource in the optional VM to make the above-mentioned optional VM load balancing is preferred The implementations should not be construed as limiting the uniqueness of the embodiments of the invention.

The VNF load growth trend is a downward trend as compared with the VNF load growth trend. The embodiment of the present invention further provides a resource scheduling solution in the case where the VNF load growth trend is a downward trend, as follows: Also includes:

If the current VNF load growth trend is a downward trend, at least one VM allocation resource is selected in the optional VM, so that the load tends to be concentrated in the above optional VM.

In the above embodiment, after the above implementations are used to make the load tend to be concentrated in the optional VM, it is convenient to select a VM that has idle or bears a small amount of traffic when the VM is released, thereby reducing the adverse impact on the service. therefore. The foregoing method may include: when releasing the VM, selecting to select a idle or a VM with a carrier service less than a certain threshold for release.

In the process of network load degradation, Scale in is needed and the VM resources are continuously released to ensure the most efficient use of resources. In the process of releasing the VM, if there are still users, calls, and sessions being processed on the VM to be released. Or the communication context, then directly releasing the VM will cause the communication of these users to be interrupted, and if all the users on the VM complete the communication process and then release the VM, they cannot be released to other VNF/APPs that need resources in time, resulting in a decrease in resource utilization. . this invention In an embodiment, when the VNF load growth trend is a downward trend, the at least one VM allocation resource is selected in the optional VM, so that the load tends to be concentrated in the optional VM, and the load is distributed as much as possible. On a part of the VM, the other part of the VM will be freed as much as possible, so that in Scale in, the free VM can be quickly released, and communication interruption can be avoided or reduced.

The embodiment of the present invention further provides an optional implementation manner of selecting at least one VM to allocate resources in the optional VM, so that the load tends to be concentrated in the optional VM, as follows: At least one VM allocates resources, so that the load tends to be concentrated in the above-mentioned optional VMs, including: sorting according to the load of the VM, selecting the VMs to allocate resources according to the load of the VM from high to low; or, sorting according to the load of the VM, Select the VM with the highest load in the optional VM to allocate resources.

Regarding the implementation of selecting at least one VM to allocate resources in the optional VM to make the load tend to be concentrated in the above-mentioned optional VM, there may be other alternatives, for example: sorting according to the load of the VM, selecting the load A higher optional VM is used as an optional VM set. When there is a resource allocation request, the VM is randomly selected from the optional VM set; or, a load threshold is set, which is higher than the load threshold. The VM is used as an optional VM set. When there is a resource allocation request, the VM is arbitrarily selected from the set of optional VMs. Therefore, the above-mentioned optional VM load is selected by selecting at least one VM allocation resource among the optional VMs. The implementation of the equalization, as a preferred implementation, should not be construed as limiting the uniqueness of the embodiments of the present invention.

In the embodiment of the present invention, the current VNF load growth trend may be determined based on the current load, or may be predicted based on the historical data. The embodiment of the present invention provides an optional implementation manner as follows: The VNF load growth trend includes: obtaining historical statistical information of resource consumption, and determining the current VNF load growth trend based on the above historical statistical information.

For a solution based on the prediction of the historical data, the embodiment of the present invention further provides a more specific implementation, as follows: Before receiving the resource allocation request, the method further includes:

The resource consumption at each time point is counted, and the time period in which the VNF load growth trend is an upward trend and the time period in which the VNF load growth trend is a downward trend are determined according to the statistically obtained resource consumption at each time point. According to the scheme for collecting historical data according to the embodiment of the present invention, the foregoing determining the current VNF load growth trend based on the historical statistical information includes: If the current time belongs to the time period in which the VNF load growth trend is an upward trend, it is determined that the current VNF load growth trend is an upward trend; if the current time belongs to a time period in which the VNF load growth trend is a downward trend, the current VNF load is determined. The growth trend is a downward trend.

The embodiment of the present invention comprehensively considers: The network activity is affected by the time of the crowd, and the network load will have periodic fluctuations every day, which is reflected in a certain network node or application, its processing load, the number of requests to be processed, The forwarding throughput also has a corresponding change in the period; the embodiment of the present invention divides the trend of the load into a number of "growth stages" and "reduction stages", and the time period based on the life rules of the crowd can be used to get rid of the load. The limit on the value used. In the embodiment of the present invention, "the resource consumption amount at each time point is counted, and the time period in which the VNF load growth trend is an upward trend is determined according to the statistically obtained resource consumption amount at each time point, and the VNF load growth trend is a downward trend. The time period "This part can be performed periodically. Generally speaking, it can be calculated periodically in days or in weekly cycles. For special dates, such as: statutory holidays, etc., statistics can be made separately. The details are as follows: Count the resource consumption at each time point within 4 days of the legal section, and determine the time period during which the VNF load growth trend is an upward trend based on the statistically obtained resource consumption at each time point, and the VNF load growth trend is a downward trend. period. This historical data will be used on the next legal holiday corresponding to it.

The following embodiment is used as an example to describe the resource scheduling scheme of the embodiment of the present invention by using the historical data of the statistics and the corresponding graph of the time and the load as follows.

Due to the influence of crowd time, the network activity will have periodic fluctuations every day. It is reflected in a certain network node or application. Its processing load, the number of requests to be processed, and the throughput of forwarding will also be corresponding. In the occurrence of periodic changes, we can divide the trend of load into several "growth stages" and "reduction stages", as shown in Figure 2: the abscissa is the time axis, the ordinate is the load, and the vertical dash is the load growth stage. The segmentation of the /load reduction phase indicates that the upward arrow identified below the time axis indicates the load growth phase and the downward arrow identified below the time axis indicates the load reduction phase.

During the network load growth phase, it is necessary to continuously add VMs to VNF/APP, expand the processing capability of VNF/APP, and perform Scale Out to ensure sufficient resources to meet communication requirements and ensure user experience. During the growth phase, VNF internally has resources. The usage policy can use the resource sharing policy (scheduling strategy one) that is evenly distributed to each VM. For example: When a new request is accessed or a new session is established, Select the VM with the lowest load among all VMs for resource allocation. In this way, each VM can maintain a reasonable load level to ensure a better user experience; in the case of a sudden increase in the number of requests (load shock), each VM can share the impact, so that the user's request can Timely processing to ensure user experience.

As shown in Figure 3, there are n VMs, and the entire cylinder represents all resources of the VM, and the slash part is

The resource that the VM has already occupied, the corresponding white part is an allocatable resource; when the resource of the VM exceeds a certain proportion, congestion may occur, which is called a congestion threshold. Assume that, as shown in FIG. 3, the resource usage rate of the VM is classified into classes. One is a class that approaches or reaches the congestion threshold, VM1, VM2; the second class is VM3, VM4... which also has resource allocation, and the three classes are VMn close to idle; there may be only one of them in practical applications. Alternatively, the number of the VMs is also arbitrary, and FIG. 3 is not to be construed as limiting the embodiments of the present invention.

According to the scheduling strategy one, VMn, or VMn, will be selected for resource allocation.

When the load drops, Scale in is needed and the VM resources are continuously released to ensure the most efficient use of resources. In the process of releasing the VM, if there is still a user, call, session or communication being processed on the VM to be released. Context, then directly releasing the VM will cause the communication of these users to be interrupted, resulting in a decline in the user experience; and if all users on the VM complete their communication process and then release the VM, the VM cannot be released to other VNF/APPs that need resources in time. This leads to a reduction in resource utilization. In the load reduction phase, the resource uses the centralized resource concentration strategy, that is, the policy that uses the opposite of the average sharing (scheduling strategy 2), the VMs are numbered in a certain order, and the new access request is distributed to the load below the overload threshold ( Or the cached threshold of the VM with the lowest number of VMs. This strategy allows the load to be distributed to a part of the VM as much as possible, while the other part of the VM is freed as much as possible, so that in Scale in, the free VM can be quickly released. Avoid or reduce the user experience degradation caused by user communication interruption.

According to the scheduling strategy one, VM3, or the VM in which VM3 is located, will be selected for resource allocation.

The point in time when the two policies are switched can be configured by the administrator of the VNF. Since the network activity is affected by the working hours and usage habits of the crowd, the daily load curve is basically repeated, so the maximum value of the load change can be automatically determined based on the statistical information of the past period of time (load When the point from the rising to the falling point and the minimum point (the point at which the load changes from falling to rising) is used as the time node of the system switching scheduling strategy, when the maximum point is reached every day (there may be more than one per day), The scheduling policy is changed to strategy two. When the minimum value is reached, the scheduling policy is changed to strategy one.

By using the corresponding resource scheduling policies in different scaling stages, resource utilization can be effectively improved, and service impairments can be avoided and reduced, and the user experience can be guaranteed.

The embodiment of the invention further provides a resource scheduling device, as shown in FIG. 4, including:

a request receiving unit 401, configured to receive a resource allocation request;

The trend obtaining unit 402 is configured to obtain a current VNF load growth trend;

The scheduling control unit 403 is configured to select at least one VM in the optional VM if the current VNF load growth trend acquired by the trend obtaining unit 402 is an upward trend after the request receiving unit 401 receives the resource allocation request. Allocate resources to balance the above optional VM load.

In the embodiment of the present invention, the resource allocation request is a request for requesting allocation of resources, and the request is used as a request for triggering resource allocation, and the expression may be: requesting access, requesting to establish a new session, and the like, in the VM. Any specific request for the allocation of the processing resource is not limited in the embodiment of the present invention.

In the embodiment of the present invention, when the network load is increasing, the VM is allocated resources, and the load balancing of each VM is performed, so that the load of each VM can be maintained at a relatively reasonable load level, thereby ensuring that the request can be processed in time. , and in the case of a sudden increase in the number of requests, each VM It can share the new resource requirements, avoid VM overload, improve request processing efficiency and reduce latency, thereby improving service quality; improving network impact resistance and system security.

The embodiment of the present invention further provides an optional implementation manner of selecting at least one VM allocation resource in the optional VM to balance the foregoing optional VM load, as follows: The scheduling control unit 403 is configured to load according to the VM. Sorting, selecting the VM allocation resources according to the load of the VM from low to high; or, sorting according to the VM load, and selecting the VM with the lowest load among the optional VMs.

The VNF load growth trend is a downward trend as compared with the VNF load growth trend. The embodiment of the present invention further provides a resource scheduling scheme in the case where the VNF load growth trend is a downward trend, as follows: The scheduling control unit 403 is further configured to: if the current VNF load growth trend acquired by the trend acquiring unit 402 is a downward trend, select at least one VM allocation resource in the optional VM, so that the load is in the foregoing The selected VM tends to concentrate.

After the above implementation is used to make the load tend to be concentrated in the above-mentioned optional VM, it is convenient to select a VM that is idle or bears a small amount of traffic when the VM is released, and the adverse effect on the service is reduced. Therefore. The above scheduling control device may further include:

The release control unit 501 is configured to select, when the VM is released, to select a VM that is idle or has a bearer service less than a certain threshold to release.

In the process of network load degradation, Scale in is needed and the VM resources are continuously released to ensure the most efficient use of resources. In the process of releasing the VM, if there are still users, calls, and sessions being processed on the VM to be released. Or communication context, then releasing the VM directly will cause these The communication of the user is interrupted, and if all the users on the VM complete the communication process and then release the VM, they cannot be released to other VNF/APPs that need resources in time, resulting in a decrease in resource utilization. In the embodiment of the present invention, when the VNF load growth trend is a downward trend, the at least one VM allocation resource is selected in the optional VM, so that the load tends to be concentrated in the optional VM. As far as possible, it is distributed to a part of the VM, and the other part of the VM is freed as much as possible, so that when Scale in, the free VM can be quickly released, and communication interruption can be avoided or reduced.

The embodiment of the present invention further provides an optional implementation manner of selecting at least one VM to allocate resources in the optional VM, so that the load tends to be concentrated in the optional VM, as follows: According to the load of the VM, the VM allocates resources according to the load of the VM from high to low; or, according to the load of the VM, selects the VM with the highest load among the optional VMs.

In the embodiment of the present invention, the current VNF load growth trend may be determined based on the current load, or may be based on the historical data. The optional implementation manner is as follows: The trend obtaining unit 402 , used to obtain historical statistical information of resource consumption, and determine the current VNF load growth trend based on the above historical statistical information.

For the solution of the prediction based on the historical data, the embodiment of the present invention further provides a more specific implementation, as follows: As shown in FIG. 6, the foregoing resource scheduling apparatus further includes:

a statistical unit 601, configured to count resource consumption at each time point;

The trend calculation unit 602 is configured to determine, according to the resource consumption amount of each time point obtained by the statistical unit 601, a time period in which the VNF load growth trend is an upward trend, and a VNF load growth trend. A period of time that is a downward trend.

Based on the embodiment of the present invention, the trend obtaining unit 402 is configured to determine that the current VNF load growth trend is an upward trend if the current time belongs to a time period in which the VNF load growth trend is an upward trend; The time period in which the above-mentioned VNF load growth trend is a downward trend determines that the current VNF load growth trend is a downward trend.

The embodiment of the present invention comprehensively considers: The network activity is affected by the time of the crowd, and the network load will have periodic fluctuations every day, which is reflected in a certain network node or application, its processing load, the number of requests to be processed, The forwarding throughput also has a corresponding change in the period; the embodiment of the present invention divides the trend of the load into a number of "growth stages" and "reduction stages", and the time period based on the life rules of the crowd can be used to get rid of the load. The limit on the value used. In the embodiment of the present invention, "the resource consumption amount at each time point is counted, and the time period in which the VNF load growth trend is an upward trend is determined according to the statistically obtained resource consumption amount at each time point, and the VNF load growth trend is a downward trend. The time period "This part can be performed periodically. Generally speaking, it can be calculated periodically in days or in weekly cycles. For special dates, such as: statutory holidays, etc., statistics can be made separately. The details are as follows: Count the resource consumption at each time point in the legal holiday, and determine the time period in which the VNF load growth trend is an upward trend according to the statistical resource consumption at each time point, and the time period in which the VNF load growth trend is a downward trend. . This historical data will be used on the next legal holiday corresponding to it.

The embodiment of the present invention further provides another resource scheduling apparatus, as shown in FIG. 7, comprising: a receiver 701, a transmitter 702, a processor 703, and a memory 704;

The receiver 701 is configured to perform control: after receiving the resource allocation request, obtain a current VNF load growth trend; if the current VNF load growth trend is an upward trend, select at least one VM allocation in the optional VM. Resources that balance the above optional VM load.

The VNF load growth trend is a parameter used to indicate the change of the VM resource consumption of the entire system. The VNF load growth trend in the embodiment of the present invention can be represented by a graph. For example, the corresponding graph of time and load, this graph will show a rising or falling trend, and the current time can predict whether the current VNF load growth trend is rising or falling; the VNF load growth trend in the embodiment of the present invention is still It can be expressed by the control value, setting the load high value and the low load value. When the current load is higher than the load high value, the current VNF load growth trend is considered to be an upward trend. When the current load is lower than the load low value , that the current VNF load growth trend is a downward trend. The parameter used to indicate the change of the VM resource consumption of the entire system may also have other representations, which is not limited by the embodiment of the present invention.

In the embodiment of the present invention, when the network load is increasing, the VM is allocated resources, and the load balancing of each VM is performed, so that the load of each VM can be maintained at a relatively reasonable load level, thereby ensuring that the request can be processed in time. And in the case that the number of requests is increased, each VM can share the newly added resource requirements, avoid overloading of the VM, improve request processing efficiency and reduce delay, thereby improving service quality; improving network impact resistance, and System security.

The embodiment of the present invention further provides an optional implementation manner of selecting at least one VM allocation resource in the optional VM to balance the foregoing optional VM load, as follows: The receiver 701 is used to control execution: Selecting at least one VM to allocate resources in the VM, so that the optional VM load balancing includes: sorting according to the load of the VM, selecting the VM to allocate resources according to the load of the VM from low to high; or, sorting according to the VM load, selecting The lowest load VM in the optional VM allocates resources.

Compared with the upward trend of the VNF load growth trend, the VNF load growth trend is a downward trend. The embodiment of the present invention further provides a resource scheduling scheme in a case where the VNF load growth trend is a downward trend, as follows: The receiver 701 is further configured to control execution: if the current VNF load growth trend is a downward trend, Then, at least one VM allocation resource is selected in the optional VM, so that the load tends to concentrate in the above-mentioned optional VM.

In the above embodiment, after the above implementations are used to make the load tend to be concentrated in the optional VM, it is convenient to select a VM that has idle or bears a small amount of traffic when the VM is released, thereby reducing the adverse impact on the service. therefore. The receiver 701 is further configured to control execution: when the VM is released, the VM that selects idle or bears less than a certain threshold is selected for release.

In the process of network load degradation, Scale in is needed and the VM resources are continuously released to ensure the most efficient use of resources. In the process of releasing the VM, if there are still users, calls, and sessions being processed on the VM to be released. Or the communication context, then directly releasing the VM will cause the communication of these users to be interrupted, and if all the users on the VM complete the communication process and then release the VM, they cannot be released to other VNF/APPs that need resources in time, resulting in a decrease in resource utilization. . In the embodiment of the present invention, when the VNF load growth trend is a downward trend, the at least one VM allocation resource is selected in the optional VM, so that the load tends to be concentrated in the optional VM. As far as possible, it is distributed to a part of the VM, and the other part of the VM is freed as much as possible, so that when Scale in, the free VM can be quickly released, and communication interruption can be avoided or reduced.

The embodiment of the present invention further provides an optional implementation manner of selecting at least one VM to allocate resources in the optional VM, so that the load tends to be concentrated in the optional VM, as follows: The receiver 701 is used to control execution. Selecting at least one VM allocation resource in the optional VM, so that the load tends to be concentrated in the above optional VM includes: sorting according to the load of the VM, and selecting the VM allocation resource according to the load of the VM from high to low; or , sort according to the load of the VM, and select the VM with the highest load among the optional VMs.

Regarding the implementation of selecting at least one VM to allocate resources in the optional VM to make the load tend to be concentrated in the above-mentioned optional VM, there may be other alternatives, for example: sorting according to the load of the VM, selecting the load A higher optional VM is used as an optional VM set. When there is a resource allocation request, the VM is randomly selected from the optional VM set; or, a load threshold is set, which is higher than the load threshold. VM as an optional VM collection, when there is resource allocation, please In order to obtain time, the VM allocation resource is arbitrarily selected from the optional VM set; therefore, the above implementation of selecting at least one VM allocation resource in the optional VM to balance the above-mentioned optional VM load is a preferred implementation. It should not be construed as a limitation of the invention.

In the embodiment of the present invention, the current VNF load growth trend may be determined based on the current load, or may be based on the historical data. The optional implementation manner is as follows: For controlling execution: Obtaining the current VNF load growth trend includes: obtaining historical statistical information of resource consumption, and determining a current VNF load growth trend according to the historical statistical information.

For the solution based on the prediction of the historical data, the embodiment of the present invention further provides a more specific optional implementation, as follows: The receiver 701 is configured to control execution: before receiving the resource allocation request, the resource consumption at each time point is counted. And a time period in which the VNF load growth trend is an upward trend and a time period in which the VNF load growth trend is a downward trend according to the statistically obtained resource consumption amount at each time point; the scheme for statistical historical data based on the embodiment of the present invention, The receiver 701 is configured to perform control: determining, according to the historical statistical information, that the current VNF load growth trend comprises: determining that the current VNF load growth trend is an upward trend if the current time belongs to a time period in which the VNF load growth trend is an upward trend; If the current time belongs to the time period in which the above-mentioned VNF load growth trend is a downward trend, it is determined that the current VNF load growth trend is a downward trend.

The embodiment of the present invention comprehensively considers: The network activity is affected by the time of the crowd, and the network load will have periodic fluctuations every day, which is reflected in a certain network node or application, its processing load, the number of requests to be processed, The forwarding throughput also has a corresponding change in the period; the embodiment of the present invention divides the trend of the load into a number of "growth stages" and "reduction stages", and the time period based on the life rules of the crowd can be used to get rid of the load. The limit on the value used. In the embodiment of the present invention, "the resource consumption amount at each time point is counted, and the time period in which the VNF load growth trend is an upward trend is determined according to the statistically obtained resource consumption amount at each time point, and the VNF load growth trend is a downward trend. The time period "This part can be performed periodically. Generally speaking, it can be calculated periodically in days or in weekly cycles. For special dates, such as: statutory holidays, etc., statistics can be made separately. The details are as follows: Count the resource consumption at each time point within 4 days of the legal section, and determine the time period during which the VNF load growth trend is an upward trend based on the statistically obtained resource consumption at each time point, and the VNF load growth trend is a downward trend. period. This historical data It will be used on the next legal holiday corresponding to it.

FIG. 8 is a schematic structural diagram of a server according to an embodiment of the present invention, which is used as a resource scheduling device; the server 800 may generate a large difference due to different configurations or performances, and may include one or more central processing units (central processing units). , CPU 822 (eg, one or more processors) and memory 832, one or more storage media 830 storing application 842 or data 844 (eg, one or one storage device in Shanghai). Among them, the memory 832 and the storage medium 830 may be short-term storage or persistent storage. The program stored on storage medium 830 may include one or more modules (not shown), each of which may include a series of instruction operations on the server. Still further, central processor 822 can be configured to communicate with storage medium 830, which performs a series of instruction operations on storage medium 830.

Server 800 may also include one or more power supplies 826, one or more wired or wireless network interfaces 850, one or more input and output interfaces 858, and/or one or more operating systems 841, such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM and more.

The method flow steps involved in the above embodiments may be based on the server structure shown in FIG. It should be noted that the above devices are only divided according to functional logic, but are not limited to the above divisions, as long as the corresponding functions can be implemented; in addition, the specific names of the functional units are only for the purpose of distinguishing from each other, and are not used. To limit the scope of protection of the present invention.

In addition, those skilled in the art can understand that all or part of the steps in implementing the foregoing method embodiments may be performed by a program to instruct related hardware, and the corresponding program may be stored in a computer readable storage medium. The storage medium may be a read only memory, a magnetic disk or an optical disk or the like.

The above is only a preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the embodiments of the present invention. All should be covered by the scope of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

Rights request

A resource scheduling method, comprising:

2. The method according to claim 1, wherein the selecting at least one VM allocation resource in the optional VM, and the balancing the load of the optional VM comprises:

3. The method according to claim 1, further comprising:

If the current VNF load growth trend is a downward trend, at least one VM allocation resource is selected in the optional VM, so that the load tends to be concentrated in the optional VM.

The method according to claim 3, wherein the selecting at least one VM allocation resource in the optional VM, causing the load to concentrate in the optional VM includes:

The method according to any one of claims 1 to 4, wherein the obtaining the current VNF load growth trend comprises:

Obtain resources; historical statistics of % consumption, and determine the current VNF load growth trend based on the historical statistics.

The method according to claim 5, further comprising: before receiving the resource allocation request, further comprising:

7. A resource scheduling device, comprising: a request receiving unit, configured to receive a resource allocation request;

8. The resource scheduling apparatus according to claim 7, wherein:

The scheduling control unit is configured to perform the sorting according to the load of the VM, and select the VM allocation resources according to the load of the VM from low to high; or, according to the VM load, select the VM with the lowest load in the optional VM.

9. The resource scheduling apparatus according to claim 7, wherein:

The scheduling control unit is further configured to: if the current VNF load growth trend acquired by the trend obtaining unit is a downward trend, select at least one VM allocation resource in the optional VM, so that the load is in the optional The VM tends to concentrate.

10. The resource scheduling apparatus according to claim 9, wherein:

The scheduling control unit is configured to perform sorting according to the load of the VM, and select VM allocation resources according to the load of the VM from high to low; or, according to the load of the VM, select the VM with the highest load among the optional VMs. .

The resource scheduling apparatus according to any one of claims 7 to 10, wherein the trend obtaining unit is configured to obtain historical statistical information of the resource consumption amount, and determine a current VNF load growth trend according to the historical statistical information.

The resource scheduling apparatus according to claim 11, further comprising: a statistical unit, configured to collect resource consumption amount at each time point;