CN106815076A - Bilateral cloud resources of virtual machine Optimal Distributing System and method based on compound mechanism - Google Patents

Abstract

A bilateral cloud virtual machine resource optimization allocation system and method based on a double-entry auction, the system includes: a bidding document collection module, a winner matching module, a virtual machine type and number confirmation module, a resource target confirmation module, a bilateral request module, and a bilateral allocation module , the present invention can solve the problem that the resource allocation target cannot reflect the resource supply and demand situation, and maximize the benefits of both parties in the allocation.

Description

Bilateral Cloud Virtual Machine Resource Optimal Allocation System and Method Based on Duplex Mechanism

technical field

The present invention relates to a technology in the field of information security, in particular to a system and method for optimal allocation of bilateral cloud virtual machine resources based on a duplex mechanism.

Background technique

The main role of the distribution mechanism based on double auction is to eliminate the monopoly of buyers or sellers in the distribution process, so that the distribution process can fully reflect the actual supply and demand relationship, ensure the benefits of buyers and sellers in the distribution, and then encourage both parties to participate in the distribution to go. The allocation mechanism based on the double-entry auction is gradually becoming a hotspot in the allocation of modern resource demand situations and in the field of Internet finance. If this mechanism can be used to encourage both parties to fully participate in emerging resource demand situations, the benefits will be huge. Due to the difficulty in the design of the double auction mechanism, an effective and reasonable double auction allocation mechanism is eager for many companies. With the continuous development of academic and industrial research, more and more double auction mechanisms will be applied to different fields.

Due to the different bilateral demands, the bids proposed by each party are also different, which makes the winner decision process in the existing double auction mechanism unusable and difficult to make direct changes to adapt. At the same time, it is impossible to obtain a credible double auction mechanism if the mechanism design of both sides is directly aimed at the content of their respective bidding documents. Finally, since the user's load is constantly fluctuating, it is difficult to estimate the actual demand of the user in advance. Formally for these three reasons, there is a need for a new mechanism to address such a bilateral allocation of resource requirements.

Contents of the invention

The present invention overcomes the deficiencies of the existing virtual machine allocation mechanism, and provides a two-way cloud virtual machine resource optimization allocation system and method based on a double-entry auction, and solves the problem that the resource allocation target cannot reflect the resource supply and demand situation in the resource demand situation through the double-entry auction allocation mechanism problem, and maximize the benefits of both parties in the distribution.

The present invention is achieved through the following technical solutions:

The present invention relates to a bilateral cloud virtual machine resource optimization allocation system based on a double auction, including: a bid collection module, a winner matching module, a virtual machine type and number confirmation module, a resource target confirmation module, a bilateral request module and a bilateral allocation module, Among them: the bid collection module is connected with the service provider, cloud service user and winner matching module. The type and number confirmation module and the resource target confirmation module are connected and connected to transmit provider and user winner information, the virtual machine type and number confirmation module is connected to the resource target confirmation module and transmit virtual machine type and number configuration information, virtual machine type and number The confirmation module and resource target confirmation module are connected to the bilateral request module and transmit the resources that need to be allocated by the provider and the resource target information that needs to be paid by the user. The bilateral request module is connected to the cloud service provider, cloud service user and bilateral allocation module. This module first Get the corresponding resources and prices from the provider and the user, and then transmit the information to the bilateral allocation module, which is connected to the cloud service provider and the cloud service user and transmits the final allocation information.

The present invention relates to a bilateral virtual machine resource optimization allocation method of the above system, comprising the following steps:

Step 1. The auctioneer collects their respective bids from cloud service providers and cloud service users;

Step 2. According to the collected bids, match and obtain the respective winners of both sides;

Step 3. According to the winning information, confirm the type and number of virtual machines that the provider needs to allocate;

Step 4. According to the winning information and the confirmed virtual machine allocation type and number, confirm the user's transaction resource target;

Step 5. According to the confirmed virtual machine type and number and the user's transaction resource target, send an allocation request to both sides;

Step 6. According to the result of the request feedback, the bilateral allocation is completed and the allocation result is notified.

technical effect

Compared with the prior art, the present invention solves the problem that in the allocation mechanism of the virtual machine, the object of allocated resources cannot reflect the supply and demand relationship in the situation of resource demand. By allowing both parties to the allocation to submit their real needs in the bidding documents, it reflects the actual situation in the allocation of resource requirements, so as to ensure that both cloud service providers and users can maximize the benefits of virtual machine allocation.

The present invention can only lose 7.01% of the total social benefit under the condition of guaranteeing the credibility; because the optimization technology is adopted in the winner matching part, the computational complexity is reduced from O(M ^K ) to O(MK), where M is the number of providers participating in the auction, and K is the number of virtual machine types.

Description of drawings

FIG. 1 is a schematic diagram of a virtual machine allocation process in a double auction;

Figure 2 is a flowchart of the auction mechanism;

Fig. 3 is a schematic diagram of experimental results obtained by combining different algorithms with the present invention and optimal auction mechanism (without considering credibility);

Fig. 4 is a schematic diagram of the comparative experiment results of the present invention.

detailed description

As shown in Figure 1, this embodiment specifically includes: a bid collection module, a winner matching module, a virtual machine type and number confirmation module, a resource target confirmation module, a bilateral request module, and a bilateral allocation module, wherein: the bid collection module and the service The provider, cloud service user and winner matching module are connected. This module first collects the bidding information from the provider and the user, and then transmits the bidding information with the winner matching module. The winner matching module is connected with the virtual machine type and number confirmation module, resource bid The confirmation module is connected and transmits the provider and user winner information, the virtual machine type and number confirmation module is connected with the resource target confirmation module and transmits the virtual machine type and number configuration information, the virtual machine type and number confirmation module, the resource target confirmation module and The bilateral request module is connected and transmits the resources that need to be allocated by the provider and the resource target information that needs to be paid by the user. The bilateral request module is connected with the cloud service provider, the cloud service user and the bilateral allocation module. The module first obtains the corresponding information from the provider and the user resources and prices, and then transmit this information to the bilateral allocation module, which is connected to the cloud service provider and cloud service user and transmits the final allocation information.

This embodiment first carries out the following definition

a. Define the form and effect of the bidding documents of buyers and sellers in the double-entry auction, specifically:

Buyer's Bid: Where: K is the number of virtual machines, x ⁽ⁱ⁾ is the number of virtual machines of each type (also called a virtual machine configuration decision), and c ⁽ⁱ⁾ is the cost paid for this virtual machine decision.

Seller's bid: Among them: m is a virtual machine type, is the number of type-m virtual machines, is the unit price of type-m virtual machines. It should be noted that the seller's bidding document actually includes K sub-bidding documents, respectively representing K types of virtual machines.

b. Define the utility of buyers and sellers:

Buyer's utility: Users rent virtual machines to complete their tasks, the revenue is R ⁽ⁱ⁾ , so its utility is: in: is the utility of user-i after winning, R ⁽ⁱ⁾ is the income of user-i for completing the task, and c ⁽ⁱ⁾ is the cost of purchasing virtual machines for user-i.

Seller's utility: Cloud service providers will have a certain cost to start a virtual machine So the utility of type-m virtual machine is: in: is the utility produced by virtual machine type-m after provider-j wins, Number of virtual machines of type-m that won the auction for provider-j The unit price of virtual machine type-m provided by provider-j, Unit cost to sell virtual machine type-m for provider-j.

Therefore, the total seller utility is in: is the total utility of winning the auction for provider-j, K is the total number of virtual machine types offered by the provider, is the utility produced by VM type-m after provider-j wins.

c. Define the utility of the auctioneer: Among them: w _a is the utility generated by the auctioneer in one allocation, N is the number of users participating in the auction, M is the number of providers participating in the auction, is the utility of user-i after winning, The total utility of winning the auction for provider-j.

As shown in Figure 1 and Figure 2, the above system is realized in the following ways:

Step 1 assigns a match for both winners;

1.1) Since the resource target of the buyer's bidding document is for the combination of virtual machines, and the resource target of the seller is unit price, direct comparison cannot be made. The auctioneer first calculates the bidding density of the buyer's bidding documents as: Wherein: μ=(μ ₁ , . . . , μ _K ) ^T is the maximum service rate of each type of virtual machine.

1.2) After calculating the bid density for each buyer’s bid, we arrange all the buyer’s bids in descending order according to the bid density, and record the sorted bid density as φ ⁽¹⁾ ≥…≥φ ^(N) . For the seller's bids, they are arranged in ascending order according to the unit price of each type of virtual machine. For type-m seller’s bids, the sorted unit price is

1.3) After the sorting is over, the auctioneer judges whether the formula is valid, so as to judge whether someone wins the auction: Among them: i' is the index of the user with the largest bidding density among the buyers. If there is a j=j* that satisfies the above inequality, then the buyer with the largest density of bids wins the buyer’s bid, and all the buyers who satisfy of sellers have won the seller's bid. If there are multiple js satisfying the above inequality, the auctioneer will choose the largest j as j*.

Step 2. Decision-making on the target of the final transaction resource and the number of transaction virtual machines, specifically:

Transaction resource target: for the winning buyer,

For each winning seller,

The number of virtual machines in the transaction: the winner of the purchase,

For each winning seller,

At this point, the auctioneer has executed all the mechanisms.

This embodiment has carried out the simulation experiment under the following environment:

Virtual machine type: There are 5 types in total, and the operating capacity refers to the number of requests that the virtual machine can receive per hour.

Cloud service providers: 4 in total. The number of servers owned by each provider is randomly generated between [500, 750], and each server can run 40 m4.large, 20 m4.xlarge, 10 m4.2xlarge, 5 m4.4xlarge or 2 m4.10xlarge virtual machine.

Cloud service users: 10 in total, whose workload we use the TPC-W standard test to generate. The comparison of test results is shown in Figure 3 and Figure 4.

The above specific implementation can be partially adjusted in different ways by those skilled in the art without departing from the principle and purpose of the present invention. The scope of protection of the present invention is subject to the claims and is not limited by the above specific implementation. Each implementation within the scope is bound by the invention.

Claims (5)

1. A two-sided cloud virtual machine resource optimization allocation system, comprising: a bidding document collection module, a winner matching module, a virtual machine type and number confirmation module, a resource target confirmation module, a bilateral request module and a bilateral allocation module, wherein: the bidding document collection module Connected with service providers, cloud service users and winner matching module, the module first collects bidding information from providers and users, and then transmits bidding information with the winner matching module, and the winner matching module confirms the type and number of virtual machines The confirmation module of the resource target is connected and connected and transmits the provider and user winner information, the virtual machine type and number confirmation module is connected with the resource target confirmation module and transmits the configuration information of the virtual machine type and number, the virtual machine type and number confirmation module, and the resource target The confirmation module is connected with the bilateral request module and transmits the resources that need to be allocated by the provider and the resource target information that needs to be paid by the user. The bilateral request module is connected with the cloud service provider, the cloud service user and the bilateral allocation module. Get the corresponding resources and prices from everywhere, and then transmit this information to the bilateral allocation module, which is connected with the cloud service provider and cloud service user and transmits the final allocation information. 2. The bilateral virtual machine resource optimization allocation method of the system according to claim 1, comprising the following steps: Step 1. The auctioneer collects their respective bids from cloud service providers and cloud service users; Step 2. According to the collected bids, match and obtain the respective winners of both sides; Step 3. According to the winning information, confirm the type and number of virtual machines that the provider needs to allocate; Step 4. According to the winning information and the confirmed virtual machine allocation type and number, confirm the user's transaction resource target; Step 5. According to the confirmed virtual machine type and number and the user's transaction resource target, send an allocation request to both sides; Step 6. According to the result of the request feedback, the bilateral allocation is completed and the allocation result is notified. 3. method according to claim 2, is characterized in that, described matching specifically comprises: 1.1) Since the resource bid of the buyer's bid is for the combination of virtual machines, and the bid of the seller's resource is unit price, direct comparison cannot be made; the auctioneer first calculates the bid density of the buyer's bid as: Where: μ=(μ ₁ , . . . , μ _K ) ^T is the maximum service rate of each type of virtual machine; 1.2) After calculating the bidding density for each buyer’s bidding document, we arrange all the buyer’s bidding documents in descending order according to the bidding document density, and note that the sorted bidding document density is φ ⁽¹⁾ ≥…≥φ ^(N) ; while for the seller’s bidding documents , it is arranged in ascending order according to the unit price of each type of virtual machine; for type-m seller bids, the sorted unit price is 1.3) After the sorting is over, the auctioneer judges whether the formula is valid, so as to judge whether someone wins the auction: Among them: i' is the label of the user with the largest density of bids among the buyers; when there is a j=j* that satisfies the above inequality, then the buyer with the largest density of bids wins the buyer’s bidding, and all satisfying The sellers of all win the seller's bidding; when there are multiple j satisfying the above inequality, the auctioneer will choose the largest j as j*. 4. The method according to claim 2, characterized in that, the types and numbers of virtual machines to be allocated refer to: the winner of the purchase, For each winning seller, 5. The method according to claim 2, characterized in that, the transaction resource target refers to: for the winning buyer, c ^(win) = φ ⁽²⁾ x ^(win)T μ, for each winning seller ,