An Online Matching Method and System for Reusable Resources in Cloud Service Platforms

Abstract

The present invention discloses an online matching method and system for reusable resources in a cloud service platform, belonging to the field of cloud service resource matching, including: the cloud service platform generates a resource set I; the client submits the user's current resource request Q to the cloud service platform; the cloud service platform receives the current resource request Q, calculates and informs the user of the resource pricing V; if the user accepts the resource pricing V, the cloud service platform queries the resource set I to obtain a resource set S that meets the current resource request Q and is not occupied; performs a priority evaluation on each resource i in the resource set S, and selects the top n resources with high priority as the resource object I* for request matching. The method of the present invention can take into account the requests of users who have not arrived while meeting the current user's request, avoid subsequent users being unable to obtain suitable resources, improve the utilization rate and efficiency of the cloud service platform, and can realize the reusable use of resources with different attributes.

Description

Technical Field

[0001] The present invention belongs to the field of cloud service resource matching, and more specifically, relates to an online matching method and system for reusable resources in a cloud service platform. Background Art

[0002] Resource scheduling in a cloud platform is very important, and the quality of resource scheduling directly affects the performance of the cloud platform and user experience. Resource scheduling involves how to manage and allocate resources such as computing and storage to meet the needs of different users. Therefore, it is crucial to optimize the resource scheduling of the cloud platform and achieve intelligent resource management.

[0003] The resource scheduling problem in a cloud platform is usually abstracted into an online bipartite graph matching problem. In the prior art, a greedy algorithm or a heuristic strategy is generally used to solve this problem. However, the greedy algorithm or the heuristic strategy only considers the arrived user requests and the current resource configuration, and ignores the unarrived user requests, which may cause subsequent users to be unable to obtain suitable resources, resulting in low resource utilization and platform efficiency.

[0004] At the same time, cloud service providers offer resources such as storage resources, computing resources, and network support resources, all of which are reusable resources. Most of the existing cloud platform resource scheduling methods are only applicable to the case where resources are not reusable and can be allocated at most once. Summary of the Invention

[0005] In view of the deficiencies and improvement requirements of the prior art, the present invention provides an online matching method and system for reusable resources in a cloud service platform, aiming to improve the resource utilization and platform efficiency of the cloud service platform.

[0006] To achieve the above object, according to one aspect of the present invention, an online matching method for reusable resources in a cloud service platform is provided, including:

[0007] The cloud service platform generates a resource set I;

[0008] The client submits the current resource request Q of the user to the cloud service platform;

[0009] The cloud service platform receives the current resource request Q, calculates and informs the user of the resource price V;

[0010] If the user accepts the resource price V, the cloud service platform queries the resource set I to obtain a resource set S that meets the current resource request Q and is not occupied; performs a priority evaluation on each resource i in the resource set S, and selects the top n resources with high priority as the resource objects I* for request matching, where n represents the number of resources required in the current resource request Q;

[0011] The priority evaluation includes:

[0012] Generate a random value y for each resource i in the resource set S i ~U[0,1];

[0013] Calculate the value of the priority evaluation function G(y i ), and select the n resources corresponding to the n values with the highest priority evaluation function values as the top n resources with high priority. Among them, G(y i ) = r i *(1 - g(y i ))), r i represents the pricing of resource i, and g(y i ) represents the valuation function of resource i.

[0014] Furthermore, it also includes: The cloud service platform re - evaluates the priority of each resource i in the resource set S at regular time intervals D; where the time interval D is the initialized re - evaluation time interval.

[0015] Furthermore, it also includes:

[0016] The cloud service platform rejects the current resource request Q with a probability of P = e -d / D ; where d is the time occupied by the resources in the current resource request Q, and D is the initialized re - evaluation time interval.

[0017] Furthermore, the valuation function g(y i ) is:

[0018]

[0019] where β is the estimation coefficient.

[0020] Furthermore, the random values y of each resource i i are independent and identically distributed.

[0021] Furthermore, it also includes:

[0022] If there are no resources in the resource set I that meet the current resource request Q or the number of unoccupied resources is less than n, the cloud service platform rejects the current resource request Q;

[0023] Or / and it also includes:

[0024] Remove the resource object I* from the resource set I;

[0025] The cloud service platform deploys the current resource request Q on the resource object I*.

[0026] According to a second aspect of the present invention, there is provided an online matching method for reusable resources for a cloud service platform, which is used for a cloud service platform and includes:

[0027] Generating a resource set I, receiving a current resource request Q, calculating and informing the user of the resource price V; wherein, the current resource request Q is submitted to the cloud service platform through a client;

[0028] If the user accepts the resource price V, query the resource set I to obtain a resource set S that meets the current resource request Q and is not occupied; perform a priority evaluation on each resource i in the resource set S, and select the top n resources with high priority as the resource objects I* for request matching, where n represents the number of resources required in the current resource request Q;

[0029] The priority evaluation includes:

[0030] Generating a random value y for each resource i in the resource set S i ~U[0,1];

[0031] Calculating the value of the priority evaluation function G(y i ) and selecting the n resources corresponding to the top n values of the priority evaluation function value as the top n resources with high priority, where G(y i ) = r i *(1 - g(y i ))), r i represents the price of resource i, and g(y i ) represents the valuation function of resource i.

[0032] According to a third aspect of the present invention, there is provided a cloud service platform, including a computer-readable storage medium and a processor;

[0033] The computer-readable storage medium is used to store executable instructions;

[0034] The processor is used to read the executable instructions stored in the computer-readable storage medium and execute the online matching method described in the second aspect.

[0035] According to a fourth aspect of the present invention, there is provided an online matching method for reusable resources for a cloud service platform, which is used for a client and includes:

[0036] Submitting the user's current resource request Q to the cloud service platform; the cloud service platform is the cloud service platform described in the third aspect.

[0037] According to a fifth aspect of the present invention, there is provided an online matching system for reusable resources for a cloud service platform, including a cloud service platform and a client;

[0038] The cloud service platform is the cloud service platform described by the third party;

[0039] The client is used to submit the current resource request Q of the user to the cloud service platform.

[0040] Generally speaking, through the above technical solutions conceived by the present invention, the following beneficial effects can be achieved:

[0041] (1) In the method of the present invention, by using the random value corresponding to each resource i in the resource set S that meets the user request as the random variable of the valuation function, and taking the resource pricing and the valuation function value as the basis for priority evaluation, while introducing a certain degree of randomness, the required number of resources with high evaluation values are selected as the matching resource objects, so that the most suitable resource can be allocated to the current request each time, rather than blindly allocating the best resource in the resource set S to the current request. While meeting the current resource request, the requests of subsequent users are also taken into account, increasing the probability that the cloud service platform can allocate suitable resources to subsequent user requests, thereby improving the resource utilization rate of the cloud platform and the efficiency of the platform. At the same time, the flexibility and adaptability of the matching method are also improved, and the limitations of greedy or heuristic strategies are alleviated.

[0042] (2) The priority evaluation function of the present invention takes both resource pricing and resource valuation into account, taking into account its own benefits while improving the resource utilization rate of the cloud service platform; in addition, the method of the present invention is relatively simple to implement and can realize the reusable use of resources with different attributes.

[0043] (3) Preferably, the cloud service platform re-evaluates the priority of each resource i in the current resource set S at time interval D, further increasing the randomness of the evaluation, avoiding the resources with high initial priorities from always becoming the optimal resources for each match, so that the evaluation method falls into local optimality, and further improving the resource utilization rate of the cloud platform and the efficiency of the platform.

[0044] (4) Preferably, the cloud service platform rejects the user's request with a certain probability, which can prevent resource requests from always occupying the current resources and making it impossible for subsequent users to use the resources normally, further improving the resource utilization rate of the cloud platform and the efficiency of the platform.

[0045] (5) Preferably, by using the valuation function of the present invention, both the valuation of the resource itself can be captured and the generated randomness can be utilized. When the random value y of the resource i is small, the value corresponding to the valuation function of the present invention does not change much. Therefore, in the priority evaluation, it depends more on the self-valuation of the resource; and when the random value gap is relatively large, especially when the random value y of the resource iWhen it is relatively large, the evaluation of the priority depends more on the magnitude of the random value, making the priority evaluation mechanism of the present invention more reasonable and improving the competition ratio of the matching method of the present invention.

[0046] (6) Preferably, the random values corresponding to each generated resource are independently and identically distributed, which can make the generated random values independent of the attributes of the resources themselves, separate the introduced randomness from the valuation, and reduce the complexity of the priority evaluation.

[0047] In summary, the method of the present invention can take into account the requests of users who have not arrived while satisfying the current user request, avoid subsequent users from being unable to obtain suitable resources, improve the utilization rate and efficiency of the cloud service platform, and enable the reusable use of resources with different attributes. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a flowchart of the online matching method for reusable resources oriented to the cloud platform of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0049] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0050] As Figure 1 shown, the online matching method for reusable resources oriented to the cloud service platform of the present invention mainly includes:

[0051] The cloud service platform generates a resource set I of the platform;

[0052] The client submits the current resource request Q of the user to the cloud service platform, and the current resource request Q includes information such as the user budget, the category and quantity of the required resources, and the time d occupied by the resources.

[0053] The cloud service platform receives the current resource request Q, calculates and informs the user of the resource price V;

[0054] If the user accepts the resource price V, the cloud service platform queries the resource set I according to the current resource request Q of the user to obtain a resource set S that meets the current resource request Q and is not occupied; perform a priority evaluation on each resource i in the resource set S, and select the top n resources with high priorities as the resource objects I* for request matching, where n represents the quantity of the required resources in the current resource request Q.

[0055] Specifically, in the step of generating the resource set I of the cloud service platform, the information structure of each resource in the resource set I includes information such as the pricing of the resource, the category of the resource, and whether the resource is occupied; among them, the category of the resource, that is, the attribute of the resource, includes attributes such as the memory of the resource, the number of cores, and CPU / GPU. In practical applications, the attributes of the resource are determined according to specific applications. In the embodiments of the present invention, the pricing of the resource represents the pricing per hour for occupying the resource.

[0056] In the embodiments of the present invention, the initialization function Setup() is used to count the information of the resources in the current cloud service platform, and the generated information structure of the resources is:

[0057] (res.r, res.att1, res.att2,...., res.att m , res.occ) = Setup(resource)

[0058] Among them, resource represents each resource in the resource set I of the cloud service platform; res.r represents the pricing of the resource; res.att i represents each attribute of the resource, where i takes 1, 2,..., m; res.occ represents the occupancy of the resource. If the resource is occupied, then res.occ = 1, otherwise, res.occ = 0.

[0059] Specifically, the current resource request Q of the user is generated on the local client.

[0060] Specifically, the resource pricing V is obtained by the cloud service platform evaluating the task volume of the user's current resource request Q and quantifying the value corresponding to the task volume, that is, the resource pricing V is:

[0061] V = f(cat, n, d)

[0062] Among them, cat is the category of the resource, n is the number of resources, d is the time of resource occupancy, and f(*) is a function for the cloud service platform to quantify the value corresponding to the task volume. Specifically, the priority evaluation of each resource i in the resource set S includes:

[0063] For each resource i in the resource set S, a random value y i ~U[0, 1] is generated; preferably, the generated random values corresponding to each resource are independently and identically distributed, which can make the generated random values not depend on the attributes of the resources themselves, separate the introduced randomness from the valuation, and reduce the complexity of the priority evaluation.

[0064] Calculate the value of the priority evaluation function G(y i ), and select the priority evaluation function G(y i) The n resources corresponding to the n values with the earliest values are used as the top n resources with high priority. Among them, G(y i ) = r i *(1 - g(y i ))), where r i represents the pricing of resource i, and g(y i ) represents the valuation function of resource i.

[0065] Each resource i in the resource object I* that matches the obtained request * can be expressed as:

[0066] i * = argmax i∈S r i (1 - g(y i ))

[0067] Preferably, in the embodiments of the present invention, the constructed valuation function g(y i ) is:

[0068]

[0069] Among them, β is an estimation coefficient, which is taken according to experience. In the embodiments of the present invention, β = 0.89. In other embodiments, other forms of valuation functions can also be used.

[0070] The method of the present invention further includes: The cloud service platform re - evaluates the priority of each resource i in the current resource set S at time interval D, further increasing the randomness of the evaluation, avoiding that the resources with high priority obtained initially always become the optimal resources for each match, so that the evaluation method falls into a local optimum, and further improving the resource utilization rate of the cloud platform and the efficiency of the platform; where the time interval D is the re - evaluation time interval initialized by the cloud service platform.

[0071] The method of the present invention further includes: If the resource set S is an empty set, that is, there are no resources in the resource set I that meet the current resource request Q or the number of unoccupied resources is less than n, the cloud service platform rejects the user's request.

[0072] Preferably, the method of the present invention further includes: For each user's resource request, the cloud service platform rejects the user's request with a probability of P = e -d / D , where d represents the time occupied by the resources in the current resource request, and D represents the initialized re - evaluation time interval. Rejecting the user's request with a certain probability can prevent the resource request from always occupying the current resources, making it impossible for subsequent users to use the resources normally.

[0073] The method of the present invention further includes: removing the resource object I* containing n required resources from the resource set I, and the cloud service platform deploying the current resource request on the resource object I* that matches the resources.

[0074] The method of the present invention further includes: after the resource object I* currently occupied by the user expires, adding the returned resources to the resource set I, updating the occupancy of the resources, and checking the attributes of the resources for subsequent reuse by users.

[0075] In the method of the present invention, by using the independent and identically distributed random values of each resource i in the resource set S that meets the user request as the random variables of the valuation function, and using the resource pricing and the valuation function value as the basis for priority evaluation, while introducing a certain degree of randomness, the required number of resources with high evaluation values are selected as the matching resource objects, so that the most suitable resources can be allocated to the current request each time, rather than simply allocating the best resources in the resource set S to the current request. While meeting the current resource request, it also takes into account the requests of subsequent users, increasing the probability that the cloud service platform can allocate suitable resources to subsequent user requests, thereby improving the resource utilization rate and the efficiency of the cloud platform. At the same time, it also improves the flexibility and adaptability of the matching method, and alleviates the limitations of greedy or heuristic strategies.

[0076] Moreover, the priority evaluation function of the present invention takes into account both resource pricing and resource valuation, while improving the resource utilization rate of the cloud service platform, taking into account its own benefits.

[0077] In addition, the method of the present invention is relatively simple to implement and can realize the reusable use of resources with different attributes.

[0078] According to the second aspect of the present invention, the present invention provides an online matching method for reusable resources for a cloud service platform, which is applied to a cloud service platform and includes:

[0079] Generating a resource set I, receiving the current resource request Q, calculating and informing the user of the resource pricing V; wherein, the current resource request Q is submitted to the cloud service platform through the client;

[0080] If the user accepts the resource pricing V, then query the resource set I to obtain the resource set S that meets the current resource request Q and is not occupied; perform priority evaluation on each resource i in the resource set S, and select the top n resources with high priority as the resource object I* for request matching, where n represents the number of resources required in the current resource request Q;

[0081] Among them, the priority evaluation includes:

[0082] Generating a random value y for each resource i in the resource set S i ~U[0,1];

[0083] Calculate the value of the priority evaluation function G(y i ), and select the n resources corresponding to the n values with the highest priority evaluation function values as the top n resources with high priority. Among them, G(y i ) = r i *(1 - g(y i ))), r i represents the pricing of resource i, and g(y i ) represents the valuation function of resource i.

[0084] According to the third aspect of the present invention, a cloud service platform is provided, including a computer-readable storage medium and a processor;

[0085] The computer-readable storage medium is used to store executable instructions;

[0086] The processor is used to read the executable instructions stored in the computer-readable storage medium and execute the online matching method of the second aspect.

[0087] According to the fourth aspect of the present invention, an online matching method for reusable resources for a cloud service platform is provided, which is used for a client and includes:

[0088] Submit the user's current resource request Q to the cloud service platform; the cloud service platform is the cloud service platform of the third aspect.

[0089] According to the fifth aspect of the present invention, an online matching system for reusable resources for a cloud service platform is provided, including a cloud service platform and a client;

[0090] The cloud service platform is the cloud service platform of the third aspect;

[0091] The client is used to submit the user's current resource request Q to the cloud service platform.

[0092] Those skilled in the art can easily understand that the above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An online matching method for reusable resources in a cloud service platform, characterized in that, include: The cloud service platform generates resource set I; The client submits the user's current resource request Q to the cloud service platform; The cloud service platform receives the current resource request Q, calculates and informs the user of the resource price V; If the user accepts the resource pricing V, the cloud service platform queries the resource set I and obtains the resource set S that meets the current resource request Q and is not occupied; Perform a priority evaluation on each resource i in the resource set S, and select the top n resources with the highest priority as the resource objects I* that match the request, where n represents the number of resources required in the current resource request Q; The priority assessment includes: Generate a random value y for each resource i in the resource set S i ~ U[0,1]; Calculate the priority evaluation function G(y i ) value, select the resources corresponding to the top n values of the priority evaluation function as the top n resources with high priority, where G(y i )=r i *(1-g(y i )), r i represents the pricing of resource i, g(y i ) represents the valuation function of resource i.

2. The method according to claim 1, wherein Also includes: The cloud service platform re-evaluates the priority of each resource i in the resource set S at every time interval D; wherein the time interval D is the initial re-evaluation time interval.

3. The method according to claim 1 or 2, characterized in that, Also includes: The cloud service platform rejects the current resource request Q with a probability of P = e -d / D ; where d is the time for which resources are occupied in the current resource request Q, and D is the initialized re-evaluation time interval.

4. The method according to claim 1, wherein The evaluation function g(y i )for: Where β is the estimated coefficient.

5. The method according to claim 1 or 2, characterized in that, The random value y of each resource i i are independently and identically distributed.

6. The method according to claim 1 or 2, characterized in that, Also includes: If there is no resource in the resource set I that satisfies the current resource request Q or the number of unoccupied resources is less than n, the cloud service platform rejects the current resource request Q; or / and also include: Remove the resource object I* from the resource set I; The cloud service platform deploys the current resource request Q on the resource object I*.

7. An online matching method for reusable resources in a cloud service platform, characterized in that, For cloud service platforms, including: Generate a resource set I, receive a current resource request Q, and calculate and inform the user of the resource price V; wherein the current resource request Q is submitted to the cloud service platform through the client; If the user accepts the resource pricing V, the resource set I is queried to obtain a set of unoccupied resources S that satisfy the current resource request Q. A priority evaluation is performed on each resource i in the resource set S, and the top n resources with the highest priority are selected as the resource objects I* that match the request, where n represents the number of resources required in the current resource request Q. The priority assessment includes: Generate a random value y for each resource i in the resource set S i ~U[0,1]; Calculate the priority evaluation function G(y i ) value, select the resources corresponding to the top n values of the priority evaluation function as the top n resources with high priority, where G(y i )=r i *(1-g(y i )), r i represents the pricing of resource i, g(y i ) represents the valuation function of resource i.

8. A cloud service platform, characterized in that, comprising a computer-readable storage medium and a processor; The computer-readable storage medium is used to store executable instructions; The processor is configured to read the executable instructions stored in the computer-readable storage medium to execute the online matching method according to claim 7.

9. An online matching method for reusable resources in a cloud service platform, characterized in that, For clients, including: Submit the user's current resource request Q to the cloud service platform; the cloud service platform is the cloud service platform described in claim 8.

10. An online matching system for reusable resources in a cloud service platform, characterized in that, Including cloud service platform and client; The cloud service platform is the cloud service platform according to claim 8; The client is used to submit the user's current resource request Q to the cloud service platform.

Images