WO2016082360A1

WO2016082360A1 - Method and device for implementing cpu service migration

Info

Publication number: WO2016082360A1
Application number: PCT/CN2015/073414
Authority: WO
Inventors: 严军; 肖鹏; 宋亚东
Original assignee: 中兴通讯股份有限公司
Priority date: 2014-11-25
Filing date: 2015-02-28
Publication date: 2016-06-02
Also published as: CN105700951A; CN105700951B

Abstract

A method and device for implementing CPU service migration. The method comprises: setting that a system CPU uses a user mode. The method further comprises: when an average value of CPU load weights on all the data plane cores reaches a load threshold, adjusting a preset service migration factor according to a corresponding preset step length and a corresponding preset period; and when the adjusted service migration factor satisfies a migration threshold, migrating services on the CPU according to a preset migration policy. In this technical solution, a CPU load threshold is set to trigger adjustment of service migration, and the adjustment is determined according to a migration threshold, so that services on the CPU are adjusted when the set migration threshold is satisfied and the power consumption of the CPU is reduced; besides, a low load migration factor and a low load migration threshold are set, and adjustment and service migration are performed according to a preset CPU descending policy, so the power consumption of the CPU is reduced and the adjustment efficiency of the CPU power consumption is improved.

Description

Method and device for realizing CPU service migration

Technical field

The present application relates to computer technology, and more particularly to a method and apparatus for realizing CPU migration of a central processing unit.

Background technique

The CPU's energy saving is usually implemented on the basis of technologies such as frequency reduction and shutdown supported by the chip itself. Different CPU energy saving strategies are different in energy saving strategies. There are different energy-saving strategies for each operating system (OS), such as the Linux operating system, which provides multiple frequency regulators for the X86 CPU; among them, the Ondemand regulator, in the supported variable frequency range, automatically according to the load. Adjust the operating frequency of the CPU core; the Performance regulator sets the CPU to the highest frequency supported to provide optimal performance; the Userspace regulator can set the CPU to operate at a specific frequency without CPU load as needed. . In addition, there are corresponding regulators for CPU energy saving adjustment for other operating systems and different types of CPUs. For different operating systems and different models of CPUs, there are the same regulators as the Userspace regulator on the X86 CPU. In general, the Userspace regulator can be referred to as a user mode regulator, and setting a user mode regulator on the CPU is also called setting the system CPU usage user mode.

Related CPU power saving methods usually set the mode in which the system works, such as "energy saving mode", "high performance mode", etc., and trigger the system to enter the energy saving mode by means of a button or other manual operation. Or by detecting the CPU occupancy rate to calculate the CPU target frequency that needs to be adjusted, this way can adjust the CPU core to the lowest frequency operation, but in the case of low traffic, there is still a large waste of energy, which cannot be the most Excellent energy saving effect.

Summary of the invention

In order to solve the above problems, the embodiments of the present invention provide a method and an apparatus for implementing CPU service migration, which can reduce the power consumption of the CPU and better implement system energy saving.

In order to achieve the above object, an embodiment of the present invention provides a method for implementing CPU service migration. include:

When the average value of the CPU load proportion on all the data plane cores reaches the load threshold, the preset service mobility factor is adjusted according to the corresponding preset step size and the preset period;

When the adjusted service mobility factor meets the migration threshold, the services on the CPU are migrated according to a preset migration policy.

Optionally, the method further includes:

When the average value of the CPU load on all data planes reaches the load threshold, before the preset service migration factor is adjusted according to the corresponding preset step and preset period, the control plane is divided according to the CPU performance requirements of the system service thread. Core and data face kernel, and record the distribution of business application threads in the data plane kernel;

When the system starts, set the data plane core to the exclusive binding core; when the business application thread starts, set the affinity binding to each data plane core.

Optionally, the service mobility factor includes a low load mobility factor; the migration threshold includes a low load migration threshold; and the migrating the services on the CPU according to the preset migration policy includes:

When the low load migration factor satisfies the low load migration threshold, the service on the CPU is decremented according to a preset CPU decrement policy, and the service on the data plane of the decremented part is allocated according to a preset allocation policy, and the corresponding is set according to a preset After the allocation of the policy is assigned, after the corresponding service application thread is migrated to the corresponding unclosed data plane core, the data plane core of the migrated service is closed, and the CPU frequency is adjusted to meet the low load operation requirement.

Optionally, the service mobility factor includes a high load mobility factor; the migration threshold includes a high load migration threshold; and the migrating the services on the CPU according to the preset migration policy includes:

When the high load migration factor satisfies the high load migration threshold, the service on the CPU restores the service to the corresponding CPU according to the distribution of the business application thread on each data plane core recorded, and adjusts the frequency of the migrated CPU to meet the work. Claim.

Optionally, the average value of the CPU load proportions on all the data plane cores includes: an accumulation of time occupied by data other than the management overhead of each service running on each data plane core per unit time;

The load threshold includes a high load threshold and a low load threshold.

Optionally, the traffic mobility factor includes a high load mobility factor and a low load mobility factor; Adjusting the business mobility factor according to the corresponding preset step size and preset period includes:

Adjusting the traffic migration factor by using the first preset step size according to the preset period for the high load migration factor;

Adjusting the traffic migration factor by using the second preset step size according to the preset period for the low load mobility factor;

The first preset step size is greater than the second preset step size.

Optionally, migrating the services on the CPU includes:

Close the service-related timers of all service application threads that perform service migration on the data plane core to perform service migration.

Conducting business migration includes:

When the high load migration factor satisfies the high load migration threshold, the service on the CPU is restored to the corresponding CPU according to the business application thread service on each data plane core of the record, and the frequency of the migrated CPU is adjusted to meet the work. When the low load migration factor satisfies the low load migration threshold, the service on the CPU is decremented according to the preset CPU decrement policy, and the service on the data plane of the decremented part is allocated according to the preset allocation policy, and the corresponding service is allocated. After the application process migrates to the corresponding unclosed data plane core, the data plane core of the migrated service is closed, and the CPU frequency is adjusted to meet the low load working requirement;

When the service is migrated to the corresponding data plane core, the service-related timer of the migration service is started, so that the service application thread continues to work.

On the other hand, the present application further provides an apparatus for implementing a CPU migration of a central processing unit, including: an adjustment unit and a migration unit;

Adjusting the unit, when the average value of the CPU load proportion on all the data plane cores reaches the load threshold, adjusting the preset service mobility factor according to the corresponding preset step size and the preset period;

The migration unit is configured to migrate the services on the CPU according to a preset migration policy when the adjusted service migration factor satisfies the migration threshold.

Optionally, the device further includes a core binding unit and a core setting unit; wherein

The core binding unit is configured to divide the control plane core and the data plane core according to the CPU performance requirements of the system service thread, and record the distribution of the service application thread in the data plane kernel;

The core setting unit is set to set the data plane core as the exclusive binding core when the system is started; when the business application thread starts, set the affinity binding to each data plane core.

Optionally, the traffic mobility factor includes a low load mobility factor; the migration threshold includes a low load migration threshold; the migration unit is set to,

When the low load migration factor satisfies the low load migration threshold, the service on the CPU is decremented according to a preset CPU decrement policy, and the service on the data plane of the decremented part is allocated according to a preset allocation policy, and the corresponding service application thread is allocated. After migrating to the corresponding unclosed data plane core, close the data plane core of the migrated service and adjust the CPU frequency to meet the low load operation requirements.

Optionally, the traffic mobility factor includes a high load mobility factor; the migration threshold includes a high load migration threshold; and the migration unit is configured to record the traffic on the CPU according to the high load migration factor when the high load migration threshold is met. The distribution of business application threads on each data plane core restores the service to the corresponding CPU and adjusts the frequency of the CPU that is migrated to meet the work requirements.

Optionally, the adjusting unit is configured to calculate, as the average value of the load proportion of the data face kernel, of the time occupied by the data other than the management overhead of each service running on each data face core per unit time, when all When the average value of the CPU load proportion on the data plane core reaches the load threshold, the preset service mobility factor is adjusted according to the corresponding preset step size and preset period;

The load threshold includes a high load threshold and a low load threshold.

Optionally, the service mobility factor includes a high load mobility factor and a low load mobility factor; and the adjusting unit is configured to adjust the traffic mobility factor by using the first preset step size according to the preset period for the high load mobility factor; The low load migration factor is adjusted according to a preset period using a second preset step size;

The first preset step size is greater than the second preset step size.

Optionally, the migration unit is configured to: when the adjusted service mobility factor meets the migration threshold, close a service-related timer of all service application threads on the data plane core for performing service migration, and perform service migration;

The business migration includes:

The application also provides a computer storage medium having stored therein computer executable instructions for performing the above method.

Compared with the related art, the technical solution provided by the embodiment of the present invention includes: setting a system CPU to use a user mode, and further comprising: when a mean value of a CPU load proportion on all data plane cores reaches a load threshold, a preset service mobility factor The adjustment is performed according to the preset preset step size and the preset period; when the adjusted service migration factor satisfies the migration threshold, the services on the CPU are migrated according to the preset migration policy. The embodiment of the present invention triggers the adjustment of the service migration by setting the CPU load threshold, and determines the adjustment according to the migration threshold to implement the adjustment of the service on the CPU when the set migration threshold is met, thereby reducing the power consumption of the CPU. In addition, through the setting of the low load migration factor and the low load migration threshold, the adjustment and service migration are performed according to the preset CPU decrementing strategy, which reduces the CPU power consumption and improves the efficiency of CPU power consumption adjustment.

BRIEF abstract

FIG. 1 is a flowchart of a method for implementing CPU service migration according to an embodiment of the present invention;

2 is a schematic diagram of a control relationship between a management thread and a service thread according to an embodiment of the present invention;

3 is a schematic flowchart of implementing a service migration and migration in an embodiment of the present invention;

FIG. 4 is a structural block diagram of an apparatus for implementing CPU service migration according to an embodiment of the present invention.

Preferred embodiment of the invention

Embodiments of the present application will be described in detail below with reference to the accompanying drawings. It should be noted that, in the case of no conflict, the features in the embodiments and the embodiments in the present application may be arbitrarily combined with each other.

FIG. 1 is a flowchart of a method for implementing CPU service migration according to an embodiment of the present invention. As shown in FIG. 1 , the method includes:

Step 100: When the average value of the CPU load proportion on all the data plane cores reaches the load threshold, adjust the preset service mobility factor according to the preset preset step size and the preset period.

Optionally, before the step, the method further includes: dividing the control plane core and the data plane core according to the performance requirement of the system service thread; and

When the system starts, set the data plane core to the exclusive binding core; when the business application thread starts, set the affinity binding to each data plane core, and record the distribution of the business application thread in the data plane kernel.

It should be noted that, according to the system service thread, the control plane core and the data plane core are divided into CPU performance requirements, including: binding a thread with lower CPU performance requirements to the control plane core; and having higher CPU performance requirements. The business application thread is bound to the data plane core. The kernel thread of the operating system (OS), the kernel interrupt of the OS, the application thread of the OS, and the thread for user configuration interaction in the service are threads that have lower CPU performance requirements according to experience of those skilled in the art. According to the experience of those skilled in the art, when other threads outside the thread that have lower CPU performance requirements are determined to be bound to the control plane core, the binding process can be adjusted according to actual conditions. The definition of the data face kernel and the control face kernel are common knowledge of those skilled in the art. Setting the data plane kernel as the exclusive binding core and setting the affinity binding to each data plane core is a setting made by those skilled in the art according to the definition of the data plane kernel and the control plane kernel, and is common knowledge of those skilled in the art.

Optionally, in step 100, the average value of the CPU load proportions on all the data plane cores includes: an accumulation of time occupied by data other than the management overhead of each service running on each data plane core per unit time;

The load threshold includes a high load threshold and a low load threshold.

It should be noted that the high load threshold and the low load threshold herein are based on those skilled in the art. The value obtained by actual experience; optionally, the high load threshold means that the CPU load is calculated to be 10% according to the traffic load, and the low load threshold is calculated to reduce the CPU load to 2% according to the traffic load. Here, the determination of the high load threshold and the low load threshold is related to the urgency of the system performance requirements. In a scenario with high performance requirements, for some services, when multi-core operation support is required, the high load threshold and low load The threshold can be set lower, otherwise it can be set higher. The urgency of the system is determined by the technician based on system requirements.

Optionally, the service mobility factor includes a high load mobility factor and a low load mobility factor; adjusting the traffic mobility factor according to the corresponding preset step size and the preset period includes:

Adjusting the service mobility factor by using the first preset step size according to the preset period for the high load mobility factor; adjusting the service mobility factor by using the second preset step size for the low load mobility factor according to the preset period; The preset step size is greater than the second preset step size.

It should be noted that the high load migration factor and the low load migration factor are values obtained according to the experience of those skilled in the art; optionally, the high load migration factor is 300; the low load migration factor is -300; when the service migration is required The damping time of the migration factor becomes small, and the migration factor can be adjusted as required. The preset period is a value obtained by a person skilled in the art according to experience; optionally, the preset period may be 1S, and when the preset period is 1S, the first preset step size is 300, and the second preset is The step size is 1; that is, when the load is high, the CPU performance is improved with higher efficiency; when the load is low, the service migration is moderately adjusted. By adjusting the service migration, the service performance requirement can be met first; Secondly, it can also save CPU power consumption.

Step 101: When the adjusted service mobility factor meets the migration threshold, the services on the CPU are migrated according to a preset migration policy.

Optionally, when the traffic mobility factor includes a low-load mobility factor and the migration threshold includes a low-load migration threshold, in this step, the services on the CPU are migrated according to the preset migration policy, including:

It should be noted that, when performing service migration, the CPU core is reduced according to the set decrement policy for each service migration; for example, when the system service is processed by 8 CPUs, when the low load mobility factor satisfies the low load migration threshold , reduce the number of CPU work to 4 CPUs; once again meet the low load migration threshold, reduce to 2; of course, can be carried out according to other preset decrement values, optionally, according to business changes and CPU performance. The allocation of the data on the data plane of the decremented part according to the pre-set allocation policy refers to the allocation method according to the actual situation of the person skilled in the art, and optionally, the method of equal allocation may be adopted; of course, the service classification may also be adopted. After the method of distribution, the specific situation is determined according to the actual situation.

Optionally, when the service mobility factor includes a high load mobility factor, and the migration threshold includes a high load migration threshold, the performing the migration on the CPU according to the preset migration policy includes:

When the high load migration factor satisfies the high load migration threshold, the service on the CPU is restored to the corresponding CPU according to the recorded business application thread on each data plane core, and the frequency of the migrated CPU is adjusted to meet the working requirement.

Optionally, the migrating the service on the CPU includes: closing a service-related timer of the service application thread that performs the service migration on the data plane core, and performing service migration; where the service migration includes:

When the high load migration factor satisfies the high load migration threshold, the service on the CPU is restored to the corresponding CPU according to the recorded business application thread service on each data plane core, and the frequency of the migrated CPU is adjusted to meet the working requirement;

Or, when the low load migration factor satisfies the low load migration threshold, the service on the CPU is decremented according to a preset CPU decrement policy, and the service on the data plane of the decremented part is allocated according to a preset allocation policy, and the corresponding service is allocated. After the application process migrates to the corresponding unclosed data plane core, the data plane core of the migrated service is closed, and the CPU frequency is adjusted to meet the low load working requirement;

It should be noted that, when performing service migration, adjusting the CPU frequency to meet the low load working requirement or the high load working requirement belongs to the technical means used by those skilled in the art, and optional, low negative The frequency required for the load operation is adjusted to use the lowest frequency supported by the CPU, and the frequency required for high load operation is adjusted to use the highest frequency.

2 is a schematic diagram of a control relationship between a management thread and a service thread according to an embodiment of the present invention; as shown in FIG. 2, in the implementation process, a management thread is set on a control plane core to perform service migration, thereby avoiding a service migration caused by a service migration. Work, the management thread realizes the record of each data plane core business application thread, the CPU load proportion calculation of the data plane core, the adjustment of the service migration factor and the judgment of whether the migration threshold is satisfied, and whether the service migration is performed according to the migration threshold; According to the actual situation, the business migration process can be realized by other hardware or chips, and the implementation of the management thread is only one of the implementation methods.

Through the generated control command, it is sent to the corresponding data plane core to complete the opening and closing of the service-related timer in the service migration process, the moving in and out of the service application thread, and the opening and closing of the data plane core.

The business migration is implemented by the interaction between the added management thread and the data plane core. When the management thread closes the service-related timer, the management thread actually terminates the work of the service application thread on the data plane core. At this time, according to the preset The migration strategy is to move in and out of the data plane core. After completing the move in or out, open the service related timer and restart the business application thread of the service. The stopping, migrating and opening of the business application thread is realized by generating corresponding control instructions in the management thread, and the generation of data of the control instruction is a conventional technical means by those skilled in the art.

The elliptical dashed line in Figure 2 represents the loop.

The time taken by the "execution of business callback processing" and "processing of business processes" in the service 1 in FIG. 2 is the basis for the management thread to calculate the proportion of the CPU load of the data face kernel.

FIG. 3 is a schematic diagram of a process for implementing a service migration and migration according to an embodiment of the present invention; as shown in FIG. 3, it is assumed that a service application thread of service B is a thread 2 on a CPU 2, and a service application thread of service A is a CPU 1 Thread 1. Here, the specific process is still explained by the management thread.

Assume that the migration threshold of the service migration is met, and the service B needs to be migrated to the CPU 1 to continue the service according to the preset migration policy. At this time, the management thread sends a request command (by entering the migration mode). Equation), notifying CPU 1 and CPU 2 that service migration is required, and CPU1 thread 1 and CPU2 thread 2 respectively stop the service-related timers on CPU 1 and CPU 2 according to the notification instruction;

The management thread requests the thread 2 to migrate the service B to the CPU 1 by sending a request message (or a control instruction or a migration message), and continues to execute the service B through the thread 1 on the CPU 1, and the thread 1 receives the CPU 2 moving into the service B. Request to complete the move of business B;

The management thread closes the data plane core of the CPU 2 according to the migration of the service B from the CPU 2 and the migration to the CPU 1, adjusts the frequency of the data plane core of the CPU 1, starts the service-related timer, and continues the service A and the service. B's work.

4 is a structural block diagram of an apparatus for implementing CPU service migration according to an embodiment of the present invention, including: an adjustment unit and a migration unit;

The device of the embodiment of the present invention further includes a core binding unit and a core setting unit; wherein

The core binding unit is configured to divide the control plane core and the data plane core according to the CPU performance requirement of the system service thread, and record the distribution of the service application thread in the data plane kernel;

When the business migration factor includes a low load migration factor and the migration threshold contains a low load migration threshold, the migration unit is set to

When the business migration factor includes a high load migration factor and the migration threshold contains a high load migration threshold, the migration unit is set to

The adjustment unit is set to accumulate the time occupied by the data other than the management overhead for each business operation on each data plane core in a unit time as the load weight of the data plane core, and the CPU load proportion on all the data plane cores. When the mean value reaches the load threshold, the preset service migration factor is adjusted according to the corresponding preset step size and preset period;

The load threshold includes a high load threshold and a low load threshold.

The adjustment unit is configured to: when the average value of the CPU load proportion on all the data plane cores reaches the load threshold, perform the service migration according to the preset service migration factor according to the preset load period according to the preset period. Adjustment of the factor; adjusting the service mobility factor by using the second preset step size according to the preset period for the low load mobility factor;

The first preset step size is greater than the second preset step size.

The migration unit is configured to: when the adjusted service mobility factor meets the migration threshold, close the service-related timers of all service application threads on the data plane core for service migration, and perform service migration;

Conducting business migration includes:

When the high load migration factor satisfies the high load migration threshold, the service on the CPU is restored to the corresponding CPU according to the recorded business application thread service on each data plane core, and the frequency of the migrated CPU is adjusted to meet the working requirement; When the low load migration factor satisfies the low load migration threshold, the service on the CPU is decremented according to the preset CPU decrement policy, and the service on the data plane of the decremented part is allocated according to the preset allocation policy, and the corresponding service is allocated. After the application process migrates to the corresponding unclosed data plane core, the data plane core of the migrated service is closed, and the CPU frequency is adjusted to meet the low load working requirement;

One of ordinary skill in the art will appreciate that all or a portion of the steps described above can be accomplished by a program that instructs the associated hardware, such as a read-only memory, a magnetic or optical disk, and the like. Optionally, all or part of the steps of the foregoing embodiments may also be implemented by using one or more integrated circuits. Accordingly, each module/unit in the foregoing embodiment may be implemented in the form of hardware, or may be implemented by using a software function module. Formal realization. The invention is not limited to any specific form of combination of hardware and software.

The embodiments disclosed in the present application are as described above, but the descriptions are only for the purpose of understanding the present application, and are not intended to limit the present application, such as the specific implementation method in the embodiments of the present invention. Any modifications and changes in the form and details of the embodiments may be made by those skilled in the art without departing from the spirit and scope of the disclosure. The scope defined by the appended claims shall prevail.

Industrial applicability

The above technical solution triggers the adjustment of the service migration by setting the CPU load threshold, and determines the adjustment according to the migration threshold to implement the adjustment of the service on the CPU when the set migration threshold is met, thereby reducing the power consumption of the CPU. In addition, through the setting of the low load migration factor and the low load migration threshold, the adjustment and service migration are performed according to the preset CPU decrementing strategy, which reduces the CPU power consumption and improves the efficiency of CPU power consumption adjustment.

Claims

A method for implementing CPU migration of a central processing unit includes:

When the average value of the CPU load proportion on all the data plane cores reaches the load threshold, the preset service mobility factor is adjusted according to the corresponding preset step size and the preset period;

When the adjusted service mobility factor meets the migration threshold, the services on the CPU are migrated according to a preset migration policy.
The method of claim 1 further comprising:

When the average value of the CPU load on all data planes reaches the load threshold, before the preset service mobility factor is adjusted according to the corresponding preset step size and preset period, the control plane core is divided according to the system service thread's CPU performance requirement. And data face kernel, and record the distribution of business application threads in the data plane kernel;

When the system starts, set the data plane core to the exclusive binding core; when the business application thread starts, set the affinity binding to each data plane core.
The method of claim 2, wherein

The traffic mobility factor includes a low load mobility factor; the migration threshold includes a low load migration threshold;

The migration of the services on the CPU according to the preset migration policy includes:

When the low load migration factor satisfies the low load migration threshold, the service on the CPU is decremented according to a preset CPU decrement policy, and the service on the data plane of the decremented part is allocated according to a preset allocation policy, and the corresponding is set according to a preset After the allocation of the policy is assigned, after the corresponding service application thread is migrated to the corresponding unclosed data plane core, the data plane core of the migrated service is closed, and the CPU frequency is adjusted to meet the low load operation requirement.
The method of claim 2, wherein

The traffic mobility factor includes a high load mobility factor; the migration threshold includes a high load migration threshold;

The migration of the services on the CPU according to the preset migration policy includes:

When the high load migration factor satisfies the high load migration threshold, the traffic on the CPU is based on the recorded The distribution of business application threads on each data plane core restores the service to the corresponding CPU and adjusts the frequency of the CPU that is migrated to meet the work requirements.
The method according to claim 1 or 2, wherein

The average value of the CPU load proportions on all the data plane cores includes: the accumulation of time occupied by data other than the management overhead of each service running on each data plane core per unit time;

The load threshold includes a high load threshold and a low load threshold.
The method according to claim 1 or 2, wherein

The service mobility factor includes a high load mobility factor and a low load mobility factor; the adjusting the traffic mobility factor according to the corresponding preset step size and the preset period includes:

Adjusting the traffic migration factor by using the first preset step size according to the preset period for the high load migration factor;

Adjusting the traffic migration factor by using the second preset step size according to the preset period for the low load mobility factor;

The first preset step size is greater than the second preset step size.
The method of claim 1 or 2, wherein the migrating the services on the CPU comprises:

Close the service-related timers of all service application threads that perform service migration on the data plane core to perform service migration.

The business migration includes:

When the high load migration factor satisfies the high load migration threshold, the service on the CPU is restored to the corresponding CPU according to the business application thread service on each data plane core of the record, and the frequency of the migrated CPU is adjusted to meet the work. When the low load migration factor satisfies the low load migration threshold, the service on the CPU is decremented according to the preset CPU decrement policy, and the service on the data plane of the decremented part is allocated according to the preset allocation policy, and the corresponding service is allocated. After the application process migrates to the corresponding unclosed data plane core, the data plane core of the migrated service is closed, and the CPU frequency is adjusted to meet the low load working requirement;

Start the service-related timing of the migration service when the service is migrated to the corresponding data plane core. The business application thread continues to work.
An apparatus for implementing a CPU migration of a central processing unit, comprising: an adjustment unit and a migration unit; wherein

The adjusting unit is configured to adjust the preset service mobility factor according to the preset preset step size and the preset period when the average value of the CPU load proportion on all the data plane cores reaches the load threshold;

The migration unit is configured to migrate the services on the CPU according to a preset migration policy when the adjusted service migration factor satisfies the migration threshold.
The apparatus according to claim 8, further comprising a core binding unit and a core setting unit; wherein

The core binding unit is configured to divide the control plane core and the data plane core according to the CPU performance requirements of the system service thread, and record the distribution of the service application thread in the data plane kernel;

The core setting unit is set to set the data plane core as the exclusive binding core when the system is started; when the business application thread starts, set the affinity binding to each data plane core.
The apparatus of claim 8, wherein the traffic mobility factor comprises a low load mobility factor; the migration threshold comprises a low load migration threshold;

The migration unit is configured to

When the low load migration factor satisfies the low load migration threshold, the service on the CPU is decremented according to a preset CPU decrement policy, and the service on the data plane of the decremented part is allocated according to a preset allocation policy, and the corresponding service application thread is allocated. After migrating to the corresponding unclosed data plane core, close the data plane core of the migrated service and adjust the CPU frequency to meet the low load operation requirements.
The apparatus of claim 8 wherein said traffic mobility factor comprises a high load mobility factor; said migration threshold comprises a high load migration threshold;

The migration unit is configured to

When the high load migration factor satisfies the high load migration threshold, the service on the CPU restores the service to the corresponding CPU according to the distribution of the business application thread on each data plane core recorded, and adjusts the frequency of the migrated CPU to meet the work. Claim.
The device according to claim 8 or 9, wherein

The adjusting unit is configured to calculate, as the average value of the load proportion of the data plane core, of the time occupied by the data except for the management overhead for each service running on each data plane core in a unit time, when all the data face kernels When the average value of the upper CPU load reaches the load threshold, the preset service migration factor is adjusted according to the preset preset step size and the preset period;

The load threshold includes a high load threshold and a low load threshold.
The apparatus of claim 8 or 9, wherein the traffic mobility factor comprises a high load mobility factor and a low load mobility factor;

The adjustment unit is configured to

Adjusting the service mobility factor by using the first preset step size according to the preset period for the high load mobility factor; and adjusting the service mobility factor by using the second preset step size according to the preset period for the low load mobility factor;

The first preset step size is greater than the second preset step size.
The device according to claim 8 or 9, wherein

The migration unit is configured to: when the adjusted service mobility factor meets the migration threshold, shut down the service-related timers of all the service application threads that perform service migration on the data plane core, and perform service migration;

The business migration includes:

When the high load migration factor satisfies the high load migration threshold, the service on the CPU is restored to the corresponding CPU according to the business application thread service on each data plane core of the record, and the frequency of the migrated CPU is adjusted to meet the work. When the low load migration factor satisfies the low load migration threshold, the service on the CPU is decremented according to the preset CPU decrement policy, and the service on the data plane of the decremented part is allocated according to the preset allocation policy, and the corresponding service is allocated. After the application process migrates to the corresponding unclosed data plane core, the data plane core of the migrated service is closed, and the CPU frequency is adjusted to meet the low load working requirement;

When the service is migrated to the corresponding data plane core, the service-related timer of the migration service is started, so that the service application thread continues to work.
A computer storage medium having stored therein computer executable instructions for performing the method of any one of claims 1-7.