Resource scheduling method and device for data center

Abstract

The invention provides a resource scheduling method and device for a data center. The method comprises the following steps: calculating a good throughput according to the throughput of an application, a request processing rate and a service satisfaction degree; and distributing a processor core and an on-chip last-stage cache for the application according to the good flux so as to maximize the good flux of the data center. The invention further provides a storage medium and electronic equipment. Therefore, the invention can maximize the good flux in the data center.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a resource scheduling method, device, storage medium and electronic equipment for a data center. Background Art

[0002] Data centers are an important infrastructure for cloud computing, but they often have difficulty coping with the problems of low system resource utilization and difficulty in ensuring the service quality of hybrid applications. In order to fully utilize the rich resources of data centers (such as processor cores, last-level cache, memory bandwidth, etc.), data centers often run various types of applications at the same time. These applications share system resources. Resource sharing brings greater uncertainty to application performance, making it difficult to ensure the service quality of hybrid applications.

[0003] Data center loads vary, and can be roughly divided into two types: latency critical (LC) and best efforts (BE). LC applications (such as online service applications such as web queries and social networks) use tail latency as a performance indicator. BE applications (such as big data mining applications) use IPC (Instructions Per Cycle) as a performance indicator. In data centers, the tail latency of LC applications is related to user experience and has a higher priority, while BE applications are mostly batch processing applications with a relatively low priority. However, the load of LC applications usually changes over time, and in many cases, existing resources cannot be fully utilized. In order to improve resource utilization, data centers usually adopt a mixed application deployment method to place multiple applications on a real machine or virtual machine. In order to solve the interference between applications, resource scheduling strategies are needed to allocate resources. However, the resource management of LC applications in existing data centers cannot achieve quality-guaranteed throughput.

[0004] In summary, the prior art obviously has inconveniences and defects in practical use, so it is necessary to improve it. Summary of the invention

[0005] In view of the above-mentioned defects, an object of the present invention is to provide a resource scheduling method, device, storage medium and electronic device for a data center, which can maximize the good throughput in the data center.

[0006] In order to solve the above-mentioned technical problems, the present invention is achieved as follows:

[0007] In a first aspect, an embodiment of the present invention provides a resource scheduling method for a data center, the method comprising:

[0008] Calculate good throughput based on application throughput, request processing rate, and service satisfaction;

[0009] According to the good throughput, processor cores and on-chip last-level caches are allocated to the applications to maximize the good throughput of the data center.

[0010] According to the resource scheduling method of the data center of the present invention, the good throughput QoS is calculated according to the throughput, request processing rate and service satisfaction of the application. i The calculation formula is:

[0011]

[0012] where q i is the throughput of the application, T si The time it takes for the data center to process a request for the application;

[0013] Q i is the entropy of application i, Q i The calculation formula is:

[0014]

[0015] Among them, M i Indicates the maximum tolerable tail delay of application i, TL i1 represents the tail delay of application i after being disturbed, R i Indicates the interference to the application, A i Indicates interference tolerance.

[0016] According to the resource scheduling method of the data center of the present invention, the step of calculating the good throughput according to the throughput, request processing rate and service satisfaction degree of the application also includes:

[0017] The calculation formula for good throughput QoS of multiple co-located applications is:

[0018] .

[0019] According to the resource scheduling method of the data center of the present invention, before the step of calculating the good throughput according to the throughput, request processing rate and service satisfaction degree of the application, the following step is further included:

[0020] Starting each of the applications and performing initial resource allocation;

[0021] According to the ReT value of the current interference tolerance residual of each application calculated, the applications are sorted according to the ReT value;

[0022] Selecting the application with the largest ReT value as the resource victim application;

[0023] The application with the smallest ReT value is selected as the resource benefiting application.

[0024] According to the resource scheduling method for a data center of the present invention, the calculation formula of the ReT value is:

[0025]

[0026] Among them, M i represents the maximum tail delay that application i can tolerate, TL i1 represents the tail delay of application i after being disturbed;

[0027] The step of selecting the application with the largest ReT value as the resource victim application further includes:

[0028] According to the ReT value, the application is graded from large to small, and the application with the smaller good throughput QoS at the same level is given priority as the resource victim application;

[0029] The step of selecting the application with the smallest ReT value as the resource beneficiary further includes:

[0030] The ReT values ​​are classified from small to large, and the application with the largest good throughput QoS at the same level is prioritized as the resource benefiting application.

[0031] According to the resource scheduling method of the data center of the present invention, after the step of allocating processor cores and on-chip last-level cache to the application according to the good throughput to maximize the good throughput of the data center, the method further includes:

[0032] The number of scheduling rounds increases by one;

[0033] Determine whether the current scheduling round number reaches a predetermined scheduling round number threshold;

[0034] If so, the application is terminated; otherwise, the process returns to the step of calculating the ReT value of the current interference tolerance remainder of each application and sorting the applications according to the ReT value.

[0035] According to the resource scheduling method for a data center of the present invention, the application is a delay-sensitive application.

[0036] In a second aspect, an embodiment of the present invention provides a resource scheduling device for a data center constructed based on any one of the methods described above, the device comprising:

[0037] A good throughput calculation module is used to calculate the good throughput according to the application throughput, request processing rate and service satisfaction degree;

[0038] A resource allocation module is used to allocate processor cores and on-chip last-level cache to the application according to the good throughput, so as to maximize the good throughput of the data center.

[0039] In a third aspect, an embodiment of the present invention provides a storage medium for storing a computer program for executing any one of the data center resource scheduling methods.

[0040] In a fourth aspect, an embodiment of the present invention provides an electronic device, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor implements any one of the resource scheduling methods for a data center when executing the computer program.

[0041] Thus, the resource scheduling strategy of the data center of the present invention includes: calculating good throughput according to the throughput, request processing rate and service satisfaction of the application; and allocating processor cores and on-chip last-level cache to the application according to the good throughput to maximize the good throughput of the data center. Thus, the present invention can maximize the good throughput in the data center. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a flowchart of a resource scheduling method for a data center provided in Embodiment 1 of the present invention;

[0043] Figure 2 It is a flow chart of a resource scheduling strategy for a data center provided in Embodiment 2 of the present invention;

[0044] Figure 3 It is a structural diagram of a resource scheduling device for a data center provided in Embodiment 1 of the present invention;

[0045] Figure 4 It is a structural diagram of a resource scheduling device for a data center provided in Embodiment 2 of the present invention;

[0046] Figure 5 It is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0048] It should be noted that references to "one embodiment", "embodiment", "example embodiment", etc. in this specification refer to the embodiment described, which may include specific features, structures or characteristics, but not every embodiment must include these specific features, structures or characteristics. In addition, such expressions do not refer to the same embodiment. Furthermore, when describing specific features, structures or characteristics in conjunction with an embodiment, whether or not there is an explicit description, it has been shown that it is within the knowledge of those skilled in the art to combine such features, structures or characteristics into other embodiments.

[0049] In addition, certain words are used in the specification and subsequent claims to refer to specific components or parts. Those with ordinary knowledge in the relevant field should understand that manufacturers can use different nouns or terms to refer to the same component or part. This specification and subsequent claims do not use differences in names as a way to distinguish components or parts, but use differences in the functions of components or parts as the criteria for distinction. "Including" and "including" mentioned throughout the specification and subsequent claims are open-ended terms and should be interpreted as "including but not limited to". In addition, the word "connected" here includes any direct and indirect electrical connection means. Indirect electrical connection means include connection through other devices.

[0050] The resource scheduling method for a data center provided by the embodiment of the present invention is described in detail below through specific embodiments and application scenarios in conjunction with the accompanying drawings.

[0051] Chinese patent application CN202210303023.4 discloses a resource scheduling method, application identification method and related equipment for a cloud computing system, and proposes the concept of LC entropy. When there are n LC applications in a computer system, the entropy Q of LC application i is i The definition of is shown in Formula 1, which represents the interference that LC application i cannot tolerate. i represents the maximum tolerable tail delay of LC application i, TL i1 represents the tail delay of LC application i after being interfered. i Greater than the interference tolerance A i , that is, when the tail delay of LC application i after interference is greater than the maximum tolerable tail delay of the application, , otherwise Q i =0.

[0052]

[0053] The LC entropy of the entire computer system is defined as shown in Equation 2.

[0054]

[0055] Chinese patent application CN202210303023.4 discloses a strategy for allocating processor cores and on-chip last-level caches, and proposes system entropy for quantifying interference in the system. Different from this, the present invention defines the concept of good throughput for LC applications, considers the throughput of LC applications, and proposes a resource scheduling strategy for improving the good throughput of data centers.

[0056] Figure 1 1 is a flow chart of a resource scheduling method for a data center provided in Embodiment 1 of the present invention, the method comprising:

[0057] Step S101, calculating good throughput according to the throughput, request processing rate and service satisfaction of the application. Preferably, the application is a delay-sensitive application.

[0058] Preferably, the present invention designs a scheduling strategy based on good throughput theory, and allocates processor cores and on-chip last-level cache according to throughput differences, request processing rates, and service satisfaction levels of different LC applications to maximize good throughput in the data center.

[0059] Preferably, the good throughput is equal to the product of throughput, request processing rate and service satisfaction, and increases with increasing throughput, increases with increasing request processing rate and increases with increasing service satisfaction.

[0060] Preferably, under the premise of ensuring that the resource benefiting applications are determined based on the service satisfaction level, the applications are classified according to the service satisfaction level values, and the priority allocation order of applications in the same category is determined according to the good throughput.

[0061] Preferably, the good throughput QoS is calculated based on the application throughput, request processing rate and service satisfaction. i The calculation formula is:

[0062]

[0063] where q i is the throughput of the application, T si The time it takes for a data center to process a request for an application.

[0064] Q i is the entropy of application i, Q i The calculation formula is:

[0065]

[0066] Among them, M i Indicates the maximum tolerable tail delay of application i, TL i1 represents the tail delay of application i after being disturbed, R iIndicates the interference to the application, A i Indicates interference tolerance.

[0067] The calculation formula for good throughput QoS of multiple co-located applications is:

[0068] .

[0069] Step S102 , allocating processor cores and on-chip last-level caches to applications according to the good throughput, so as to maximize the good throughput of the data center.

[0070] The present invention belongs to the field of data centers and is used for resource management and throughput evaluation in computer systems. More specifically, it is used for resource management of LC applications in data centers to improve throughput with quality assurance. The present invention proposes a good throughput measurement for data centers and designs a resource scheduling strategy for good throughput of data centers. Good throughput is defined as the throughput portion that meets the quality of service requirements for LC applications, and quantifies the throughput with quality assurance in data centers.

[0071] Figure 2 : is a flow chart of a resource scheduling method for a data center provided in Embodiment 2 of the present invention, the method comprising:

[0072] Step S201, start each application and perform initial resource allocation. Preferably, the application is a delay-sensitive application.

[0073] Step S202: rank the applications according to the ReT values ​​calculated for the current residual interference tolerance of the applications.

[0074] Step S203: Select the application with the largest ReT value as the resource victim application.

[0075] Step S204: Select the application with the smallest ReT value as the resource benefiting application.

[0076] Preferably, the calculation formula of the ReT value is:

[0077]

[0078] Among them, M i represents the maximum tail delay that application i can tolerate, TL i1 It represents the tail delay of application i after being disturbed.

[0079] The step of selecting the application with the largest ReT value as the resource victim application further includes:

[0080] The applications are graded from large to small according to the ReT value, and those with smaller good throughput QoS at the same level are given priority as resource-affected applications.

[0081] The step of selecting the application with the smallest ReT value as the resource beneficiary further includes:

[0082] Applications are graded according to ReT values ​​from small to large, and applications with large throughput QoS at the same level are given priority as resource benefiting applications.

[0083] Step S205, calculating the good throughput according to the application throughput, request processing rate and service satisfaction degree.

[0084] Preferably, the good throughput QoS is calculated based on the application throughput, request processing rate and service satisfaction. i The calculation formula is:

[0085]

[0086] where q i is the throughput of the application, T si The time it takes for a data center to process a request for an application.

[0087] Q i is the entropy of application i, Q i The calculation formula is:

[0088]

[0089] The calculation formula for good throughput QoS of multiple co-located applications is:

[0090] .

[0091] Among them, M i Indicates the maximum tolerable tail delay of application i, TL i1 represents the tail delay of application i after being disturbed, R i Indicates the interference to the application, A i Indicates interference tolerance.

[0092] Step S206 , allocating processor cores and on-chip last-level caches to applications according to the good throughput, so as to maximize the good throughput of the data center.

[0093] Step S207, the scheduling round number is increased by one.

[0094] Step S208, determine whether the current scheduling round number reaches the predetermined scheduling round number threshold, if yes, execute step S209, otherwise return to step S202. That is, according to the ReT value of the current interference tolerance residue of each application, the applications are sorted according to the ReT value.

[0095] Step S209, terminate the application.

[0096] The present invention also proposes a corresponding scheduling strategy to maximize the good throughput in the data center. The present invention defines the concept of good throughput for LC applications, considers the throughput of LC applications, and proposes a resource scheduling strategy for improving the good throughput of the data center.

[0097] The present invention defines good throughput for LC applications to evaluate the throughput and service quality of LC applications, and designs a resource scheduling strategy oriented to good throughput to improve the overall good throughput of the data center.

[0098] In actual applications, the throughput of LC applications varies and should not be treated equally in resource scheduling. The system throughput is also an important evaluation indicator of the data center. For example, imagine two scenarios (A and B). LC The resource contention in scenario A is slight (the request arrival rate QPS of all LC applications is at a low level), while the resource contention in scenario B is severe (the request arrival rate QPS of all or most LC applications is at a high level). LC Similarly, it is obvious that the QoS effect of scenario B is more difficult to achieve than that of scenario A. Therefore, it is necessary to propose a throughput-oriented LC entropy definition.

[0099] The present invention defines the quality of service of LC application i as QoS i As shown in formula 3, where q i is the throughput of LC application i (i.e., the request arrival rate or request processing rate of application i when the system is in a stable state, both of which are equal). si The time it takes for the server to process a request from application i. Since the difficulty of implementing requests from different applications may vary, it is meaningless to directly compare the throughput. Therefore, the raw throughput q i Considering the unit request service time T si , to homogenize the throughput.

[0100]

[0101] The present invention is called Quality of Service QoS i Good Throughput, q i is the throughput (i.e., flux), T si To unify the coefficient of difficulty of request processing between applications, 1-Q i is the quality factor. In short, q i •T si and 1-Q i , respectively reflecting the throughput and quality of the service, and their product represents the quality of service, or good throughput.

[0102] QoS (Quality of Service or Good Throughput) of Co-located Multiple LC Applications i As shown in Formula 4.

[0103]

[0104] The following is a simplified example, assuming that there are three scenarios A, B, and C, each of which has multiple LC applications co-located.

[0105] In scenario A, LC applications A1 and A2 are co-located, q A1 =2, Q A1 =0.1, T s1 =5,q A2 =6, Q A2 =0.5, T s2 =2. Application A2 has higher entropy (5 times that of application A1), higher throughput (3 times that of application A1), shorter time taken for a single request of application A2 (40% of that of application A1), and better QoS. A1 =2•5•0.9=9, QoS A2 =6•2•0.5=6, the overall service quality QoS A =15.

[0106] In scenario B, LC applications B1 and B2 are co-located, q B1 =2, Q B1 =0.1, T s1 =5,q B2 =12, Q B2 =0.5, T s2 =2, the entropy of application B2 is higher (5 times that of application A1), the throughput of application B2 is higher (6 times that of application B1), the time consumed by a single request of application B2 is shorter (40% of that of application B1), and the QoS B1 =2•5•0.9=9, QoS B2 =12•2•0.5=12, the overall quality of service QoS B =21.

[0107] In scenario C, LC applications C1, C2, and C3 are co-located, q C1 =2, Q C1 =0.2, T s1 =5,q C2 =12, Q C2 =0.6, T s2 =2,q C3 =4, Q C2 =0.2, T s2 =1, easy to get QoS C1 =2•5•0.8=8, QoS C2=12•2•0.4=9.6, QoS C3 =4•1•0.8=4, the overall service quality QoS B =21.6.

[0108] Application C2 has a higher entropy (5 times that of application A1), application B2 has a higher throughput (6 times that of application B1), and a single request of application B2 takes a shorter time (40% of that of application B1). B1 =2•5•0.9=9, QoS B2 =12•2•0.5=12, the overall quality of service QoS B =21.

[0109] By defining QoS, it is possible to compare the individual service qualities of different applications in the same scenario (e.g. QoS A1 >QoS A2 , QoS B1 <QoS B2 ), and can also compare the overall service quality of different scenarios (such as QoS A＜ QoS B ), and the number of co-located applications in the compared scenarios does not need to be the same (for example, there are 3 applications in scenario C, and 2 applications in scenarios A and B). The service quality of individual applications in different scenarios can also be compared (for example, QoS B2 >QoS C2 ).

[0110] The QoS defined above takes into account both the degree of satisfaction of tail latency and throughput, as well as the difference in difficulty for the server to complete the same throughput of different applications. Therefore, it is essentially the throughput that meets user requirements after homogenization (Good throughput after homogenization), referred to as good throughput.

[0111] The present invention first starts the application and distributes resources evenly, then runs a specified number of rounds at a fixed time interval (i.e., scheduling interval). In each round, the scheduling strategy performs resource adjustment, and the application is terminated after the number of rounds is reached. Before each adjustment, the current interference tolerance residual (ReT) value of each application is calculated, and the calculation method is as shown in Formula 5, where M i represents the maximum tail delay that users of application i can tolerate, TL i1 It represents the tail delay of application i after being disturbed. During scheduling, the applications are sorted according to the size of the interference tolerance residual ReT, and the application with the largest ReT value is selected as the resource victim, and the application with the smallest ReT value is selected as the resource beneficiary.

[0112]

[0113] The scheduling strategy algorithm ARQ-QoS strategy proposed in the present invention modifies the ARQ scheduling strategy pseudo code proposed in the patent "Resource Scheduling Method, Application Identification Method and Related Devices of Cloud Computing System" as shown in Table 1.

[0114]

[0115] The findVictimRegion and findBeneficiaryRegion functions have been modified. The modification of the findVictimRegion function is shown in lines 25-40 of the pseudo code. The way to select the victim application has changed from being determined solely by the ReT value to being selected from ReT>0.5 and two service satisfaction levels, and in each corresponding level, the resource victim application is returned in ascending order of QoS value, that is, the resources of the application with a small current QoS value are preferentially deprived. The modification of the findBeneficiaryRegion function is shown in lines 43-48 of the pseudo code. The way to select the beneficiary application has changed from being determined solely by the ReT value to being selected from ReT<-0.2 and -0.2≤ReT<0.05, two service satisfaction levels, and in each corresponding level, the resource beneficiary application is returned in descending order of QoS value, that is, resources are preferentially allocated to the application with a large current QoS value.

[0116] Figure 3 1 is a schematic diagram of the structure of a resource scheduling device for a data center provided in Embodiment 1 of the present invention. The resource scheduling device 100 for a data center includes a good throughput calculation module 10 and a resource allocation module 20, wherein:

[0117] The good throughput calculation module 10 is used to calculate the good throughput according to the throughput, request processing rate and service satisfaction of the application. Preferably, the application is a delay-sensitive application.

[0118] The resource allocation module 20 is used to allocate processor cores and on-chip last-level caches to applications according to the good throughput, so as to maximize the good throughput of the data center.

[0119] Figure 4 1 is a schematic diagram of the structure of a resource scheduling device for a data center provided in Embodiment 2 of the present invention. The resource scheduling device 100 for a data center includes a good flux calculation module 10, a resource allocation module 20, an initialization module 30, an application selection module 40 and a round determination module 50, wherein:

[0120] The initialization module 30 is used to start each application and perform initial resource allocation. Preferably, the application is a delay-sensitive application.

[0121] The application selection module 40 calculates the ReT value of the current interference tolerance surplus of each application and sorts the applications according to the ReT value; selects the application with the largest ReT value as the resource victim application; and selects the application with the smallest ReT value as the resource benefit application.

[0122] Preferably, the calculation formula of the ReT value is:

[0123]

[0124] Among them, M i represents the maximum tail delay that application i can tolerate, TL i1 It represents the tail delay of application i after being disturbed.

[0125] The step of selecting the application with the largest ReT value as the resource victim application further includes:

[0126] The applications are graded from large to small according to the ReT value, and those with smaller good throughput QoS at the same level are given priority as resource-affected applications.

[0127] The step of selecting the application with the smallest ReT value as the resource beneficiary further includes:

[0128] Applications are graded according to ReT values ​​from small to large, and applications with large throughput QoS at the same level are given priority as resource benefiting applications.

[0129] The good throughput calculation module 10 is used to calculate the good throughput according to the throughput, request processing rate and service satisfaction of the application. i The calculation formula is:

[0130]

[0131] where q i is the throughput of the application, T si The time it takes for a data center to process a request for an application.

[0132] Q i is the entropy of application i, Q i The calculation formula is:

[0133]

[0134] Among them, M i Indicates the maximum tolerable tail delay of application i, TL i1 represents the tail delay of application i after being disturbed, R i Indicates the interference to the application, A i Indicates interference tolerance.

[0135] The calculation formula for good throughput QoS of multiple co-located applications is:

[0136] .

[0137] The resource allocation module 20 is used to allocate processor cores and on-chip last-level caches to applications according to the good throughput, so as to maximize the good throughput of the data center.

[0138] The round judgment module 50 is used to increase the scheduling round number by one after the resource allocation module 20 performs a resource allocation; judge whether the current scheduling round number reaches a predetermined scheduling round number threshold; if so, terminate the application, otherwise return to the step of calculating the ReT value of the current interference tolerance remainder of each application, and sort each application according to the ReT value.

[0139] The present invention also provides a storage medium for storing Figure 1~Figure 3 A computer program of any of the resource scheduling methods of the data center. For example, a computer program instruction, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer, and can achieve the same technical effect. To avoid repetition, it will not be repeated here. The program instructions for calling the method of the present invention may be stored in a fixed or removable storage medium, and / or transmitted through a data stream in a broadcast or other signal-bearing medium and / or stored in a storage medium of a computer device that runs according to the program instructions.

[0140] According to one embodiment of the present invention, the present invention also provides a Figure 5 The electronic device 400 shown in the figure may optionally include a storage medium 200 for storing a computer program and a processor 300 for executing the computer program, wherein when the computer program is executed by the processor 300, any of the above-mentioned data center resource scheduling methods is implemented, triggering the electronic device 400 to execute the methods and / or technical solutions based on the aforementioned multiple embodiments, and the same technical effects can be achieved. To avoid repetition, it is not repeated here. It should be noted that the electronic devices in the embodiments of the present invention include mobile electronic devices and non-mobile electronic devices. Exemplarily, the mobile electronic device may be a mobile phone, a tablet computer, a laptop computer, a PDA, an in-vehicle electronic device, a wearable device, a super mobile personal computer, a netbook or a personal digital assistant, etc., and the non-mobile electronic device may be a server, a network attached storage (Network Attached Storage, NAS), a personal computer (personal computer, PC), a television (television, TV), a teller machine or a self-service machine, etc., which is not specifically limited in the embodiments of the present invention.

[0141] It should be noted that the present invention can be implemented in software and / or a combination of software and hardware, for example, it can be implemented using an application-specific integrated circuit (ASIC), a general-purpose computer or any other similar hardware device. In one embodiment, the software program of the present invention can be executed by a processor to implement the above steps or functions. Similarly, the software program of the present invention (including related data structures) can be stored in a computer-readable recording medium, for example, a RAM memory, a magnetic or optical drive or a floppy disk and the like. In addition, some steps or functions of the present invention can be implemented using hardware, for example, as a circuit that cooperates with a processor to perform various steps or functions.

[0142] The present invention can be implemented on a computer as a computer-implemented method, or implemented in dedicated hardware, or implemented in a combination of the two. The executable code or part thereof for the method according to the present invention can be stored on a computer program product. Examples of computer program products include memory devices, optical storage devices, integrated circuits, servers, online software, etc. Optionally, the computer program product includes a non-temporary program code component stored on a computer-readable medium so as to perform the method according to the present invention when the program product is executed on a computer.

[0143] In an alternative embodiment, the computer program comprises computer program code means adapted to perform all the steps of the method according to the invention when the computer program is run on a computer. Optionally, the computer program is embodied on a computer readable medium.

[0144] It should be noted that, in this article, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the method and device in the embodiment of the present invention is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0145] Of course, the present invention may have many other embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art may make various corresponding changes and modifications based on the present invention, but these corresponding changes and modifications should all fall within the scope of protection of the claims attached to the present invention.

Claims

1. A data center resource scheduling method, characterized in that: include: Calculate good throughput based on application throughput, request processing rate, and service satisfaction; According to the good throughput, processor cores and on-chip last-level caches are allocated to the applications to maximize the good throughput of the data center.

2. The resource scheduling method of a data center according to claim 1, characterized in that: The good throughput QoS is calculated according to the throughput, request processing rate and service satisfaction of the application i The calculation formula is: where q i is the throughput of the application, T si The time it takes for the data center to process a request for the application; Q i is the entropy of application i, Q i The calculation formula is: Among them, M i Indicates the maximum tolerable tail delay of application i, TL i1 represents the tail delay of application i after being disturbed, R i Indicates the interference to the application, A i Indicates interference tolerance.

3. The resource scheduling method of a data center according to claim 2, characterized in that: The step of calculating the good throughput according to the throughput, request processing rate and service satisfaction degree of the application also includes: The calculation formula for good throughput QoS of multiple co-located applications is: 。 4. The data center resource scheduling method according to claim 1, characterized in that: Before the step of calculating good throughput according to the throughput, request processing rate and service satisfaction degree of the application, the step further includes: Starting each of the applications and performing initial resource allocation; According to the ReT value of the current interference tolerance residual of each application calculated, the applications are sorted according to the ReT value; Selecting the application with the largest ReT value as the resource victim application; The application with the smallest ReT value is selected as the resource benefiting application.

5. The data center resource scheduling method according to claim 4, characterized in that: The calculation formula of the ReT value is: Among them, M i represents the maximum tail delay that application i can tolerate, TL i1 represents the tail delay of application i after being disturbed; The step of selecting the application with the largest ReT value as the resource victim application further includes: According to the ReT value, the application is graded from large to small, and the application with the smaller good throughput QoS at the same level is given priority as the resource victim application; The step of selecting the application with the smallest ReT value as the resource beneficiary further includes: The ReT values ​​are classified from small to large, and the application with the largest good throughput QoS at the same level is prioritized as the resource benefiting application.

6. The data center resource scheduling method according to claim 5, characterized in that: After the step of allocating processor cores and on-chip last-level cache to the application according to the good throughput to maximize the good throughput of the data center, the step further includes: The scheduling round number increases by one; Determine whether the current scheduling round number reaches a predetermined scheduling round number threshold; If so, the application is terminated; otherwise, the process returns to the step of calculating the ReT value of the current interference tolerance remainder of each application and sorting the applications according to the ReT value.

7. The data center resource scheduling method according to claim 1, characterized in that: The application is a delay-sensitive application.

8. A data center resource scheduling device constructed based on the method described in any one of claims 1 to 7, characterized in that: The device comprises: A good throughput calculation module is used to calculate the good throughput according to the application throughput, request processing rate and service satisfaction degree; A resource allocation module is used to allocate processor cores and on-chip last-level cache to the application according to the good throughput, so as to maximize the good throughput of the data center.

9. A storage medium, characterized in that: Used to store a computer program for executing the resource scheduling method for a data center as described in any one of claims 1 to 7.

10. An electronic device comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that: When the processor executes the computer program, the resource scheduling method for a data center according to any one of claims 1 to 7 is implemented.

Images