METHOD FOR CONTROLLING THE ALLOCATION REQUESTS OF AN IT RESOURCE

Description

[0001] The present invention relates to a method for managing allocation requests, in a cloud computing system, for a computing resource originating from a user's terminal.

[0002] It also relates to a system for managing allocation requests.

[0003] A "cloud" computing system, also known as a "cloud computing" system, comprises a set of remote computing devices linked together by a communication network, such as the Internet, providing users with on-demand access to computing resources, such as storage space, software, computing power, etc.

[0004] In general, cloud computing systems are organized, according to a centralized model, into computer centers or data processing centers (known as "data centers" in Anglo-Saxon terminology), each data processing center hosting computer equipment or devices to provide services to users.

[0005] Data centers are managed by cloud service providers (“Cloud Service Provider” CSP in Anglo-Saxon terminology).

[0006] Cloud service providers primarily offer three categories of services: Infrastructure as a Service (IaaS), Platform as a Service (PaaS), and Software as a Service (SaaS). Infrastructure as a Service (LaaS) provides access to virtualized computing resources, such as virtual machines installed on servers in data centers. Platform as a Service (PaaS) offers virtual resources such as virtual machines with an operating system installed, and Software as a Service (SaaS) allows users to access software installed on servers in data centers.

[0007] Infrastructure as a Service (IaaS) is primarily intended for companies wishing to outsource services to a cloud service provider instead of purchasing IT infrastructure. Using the services offered by cloud service providers allows a company to, for example, adjust its IT resources according to peak business activity periods.

[0008] Cloud service providers may own multiple data centers, spread across different geographical areas that are sometimes far apart, with the data centers being accessible via a communication network, such as the Internet.

[0009] Generally, when a customer device subscribes to a service provider's offering, one of the service provider's data centers manages the customer account and the device's access to the services. Furthermore, in traditional cloud systems, the services requested by that customer device are provided by that data center. However, it is possible for the customer to select a different data center to host a service.

[0010] The client device (or service user), such as a business customer of the cloud service provider, accesses the services by connecting to the communication network via an access network. This access network is, for example, a mobile or fixed telephone network of an operator or a private network of a company, for example of the LAN (local area network) type.

[0011] Very often, the client device and the data center providing the service are far apart. Given this distance between the location of the user requesting the service and the location of the data center providing the service, the quality of service may not be satisfactory, particularly for computing resources requiring significant bandwidth, sometimes making access to services difficult or even impossible when latency is high and / or available bandwidth is low.

[0012] In this context, there is a need to improve the quality of services provided by cloud service providers.

[0013] One existing solution for reducing latency when using services is to increase the number of data centers in the cloud computing system in order to distribute computing resources so that users have access to data centers that are close to them. These types of cloud computing systems are known as distributed cloud computing systems and reduce the distance between the user and the data center providing services.

[0014] In particular, in this type of solution, the cloud service provider adds data centers, in order to reduce the distance between at least one of the additional data centers and the user.

[0015] Of course, the distance should be understood as "network distance," meaning it is equivalent to the time required to transmit data between two points. Thus, if the distance between the client device and a first data center is less than the distance between the client device and a second data center, the data is transmitted faster to the first data center than to the second.

[0016] For example, a client device might be located in one geographic area (such as a country) and outsource services to a service provider with a data center located in a second geographic area. According to the decentralization solution described above, the service provider can add data centers in the first geographic area where the client device is located, so that the distance between the client device and at least one of these data centers in the first geographic area is less than the distance between the client device and the data center located in the second geographic area.

[0017] In this type of cloud computing system, when a client device or user terminal requests the allocation of a computing resource, a management device determines which computing center will provide the requested service, for example, which one will allocate the requested resource. This management device selects, based on various criteria, the most suitable computing center from among all the computing centers in the cloud computing system to provide the requested service. The management device can be a management server, a device within a management server, or within a computing center. This decision-making process can vary depending on the cloud computing system. In some systems, the management device assesses the availability of resources in each computing center and, based on this information, selects the most appropriate computing center according to the client's request requirements.In other systems, the management device does not send allocation requests to the different computer centers but selects itself, according to predefined parameters, the most suitable computer center for providing the requested service.

[0018] However, centralized management of resource allocation requests by a management system can be complex, especially when the number of data centers in the cloud computing system is large. Furthermore, the centralized selection of a data center for resource allocation (or service delivery) delays the allocation of the requested computing resource. The paper "A hierarchical structure for optimal resource allocation in geographically distributed clouds" (HASAN ZIAFAT ET AL., FUTURE GENERATION COMPUTER SYSTEMS, vol. 90, 2018-08-23, pages 539-568, DOI: 10.1016 / j.future.2018.08.027) discloses a hierarchical structure for resource allocation in a geographically distributed cloud computing system. In the process disclosed by this paper, intermediate servers ("brokers") implement the process and direct or respond to resource allocation requests.These intermediate servers are organized in a hierarchical structure comprising geographical brokers, middle brokers, and cloud brokers. The user sends their initial request to a geographical broker, and routing occurs between entities within the cloud computing system.

[0019] The document "Cost-based data center selection policy for large-scale networks" (REKHA PM ET AL., 2014 INTERNATIONAL CONFERENCE ON COMPUTATION OF POWER, ENERGY, INFORMATION AND COMMUNICATION (ICCPEIC), IEEE, 2014-04-16, pages 18-23, DOI: 10.1109 / ICCPEIC.2014.6915333) outlines a cost-based data center selection policy for large-scale networks. This selection policy is implemented by service brokers. Service brokers forward resource requests according to policies that may be based on proximity, cost, or dynamic reconfiguration of retransmission policies.

[0020] The invention aims to improve this situation.

[0021] The scope of the invention is defined by the attached independent claims. Other embodiments are defined in their respective dependent claims.

[0022] To this end, the invention aims, according to a first aspect, at a method of managing a request for the allocation of a computing resource in a cloud computing system, said request being issued by a user terminal, said cloud system comprising at least two computing centers connected to each other by a communication network, said management method being implemented by an access device allowing the user terminal to access said communication network via an access network and being configured to determine a routing path of the request to a predefined service address for the cloud computing system.

[0023] According to the invention, the management process comprises: the transmission of the allocation request, received from the user's terminal, to a first computing center; and if, according to an indicative response regarding the availability of the computing resource, the computing resource is not available in the first computing center, the retransmission of the allocation request to an adjacent computing center, this adjacent computing center being the next computing center in the routing path, the retransmission of the allocation request to an adjacent computing center being repeated until either a computing center responds that the resource is available or the request has been retransmitted to all computing centers in the routing path.

[0024] Thus, the request is first sent to the first computing center. If the computing resource is not available in this first computing center, the request is forwarded to the adjacent computing center or the next computing center in the routing path established by the access device to reach the requested service.

[0025] Therefore, the request for allocation of the computing resource is addressed progressively to the computing centers along the routing path, starting from the first computing center.

[0026] Note that the data centers in the routing path are contacted one by one until a positive response is received for resource allocation or until all data centers have been contacted. If none of the data centers responds favorably to the requested resource allocation, the user is notified.

[0027] In addition, the routing path is established by the access device to reach a so-called service address, this address being provided by the service provider for the cloud computing system.

[0028] As noted above, distance should be understood as "network distance," meaning that when a resource allocation request is routed between the user's terminal and the service address, using the routing path established by the access device, the computer center closest to the user's terminal is the computer center that first receives the routed request.

[0029] Thanks to these features, managing IT resource allocation is more efficient because the centralized determination of the data center that will provide the service in a distributed cloud computing system is avoided. This simplifies the process of determining the data center that will provide the service and allows for faster implementation without overloading the network.

[0030] Furthermore, in the event of network congestion, the impact of network congestion on allocation request management is minimized.

[0031] In summary, the management of the computing resource allocation request from the user's terminal is thus optimized. This allows the computing resource to be allocated more quickly, without overloading the communication network. It should be noted that in the prior art of a distributed cloud computing system where a management device queries data centers about resource availability, not only is the allocation request management slower, but additional data traffic is also generated, contributing to the overload of the communication network.

[0032] The advantages of the method according to the invention are even greater when the cloud system comprises a large number of data centers. Indeed, the complexity of centrally managing resource allocation requests increases as the number of data centers in the cloud computing system grows. Thus, the method proposed by the invention is even more advantageous in these cloud computing systems with a high number of data centers.

[0033] According to one characteristic, the first computing center is the first computing center in the determined routing path.

[0034] Thus, the first computer center is the computer center closest to the user's terminal.

[0035] The allocation request is first sent to the data center closest to the user's terminal. If the computing resource is not available at this first data center or the data center closest to the user, the request is forwarded to the adjacent or next data center in the routing path established by the access device to reach the requested service.

[0036] Therefore, the request for allocation of the computing resource is addressed progressively to the computing centers along the routing path, starting from the computing center closest to the user.

[0037] Note that the data centers along the routing path are contacted one by one until a positive response is received for resource allocation, or until the data center furthest from the user's terminal has been contacted (or in other words, until all data centers have been contacted). If none of the data centers responds favorably to the requested resource allocation, the user is notified.

[0038] It should also be noted that, in this implementation, the data center that first receives the routed request is the one closest to the user's terminal. Thus, since the resource is allocated closest to the user's terminal, the quality of service is optimal.

[0039] According to one characteristic, if the IT resource is not available in a computer center, the management process further includes receiving the indicative response of the availability of the IT resource.

[0040] This response comes from the data center to which the allocation request was sent. Thus, initially, the access device receives the response from the first data center in the routing path.

[0041] Thus, if the computing resource is not available in the first data center, the first data center responds to the access request to inform the client. Conversely, if the computing resource is available, the resource allocation process ends and the first data center makes the computing resource available to the client.

[0042] The same applies to subsequent data centers in the routing path. Thus, if the data center to which the access device sends the request is unavailable, it responds to the access device to inform it of this.

[0043] According to one embodiment, when the response indicates that the resource is not available, the indicative response of the availability of the computing resource includes the allocation request transmitted.

[0044] When the resource cannot be allocated by a computer center and the computer center sends a negative response to the access device, the access device receiving the allocation request that it had previously transmitted understands that the allocation request could not be accepted by the computer center to which it had sent the request, and retransmits it to an adjacent computer center.

[0045] According to another embodiment, the indicative response regarding the availability of the IT resource includes a message indicating that the requested IT resource is not available.

[0046] According to one characteristic, when the request has been retransmitted to all computer centers in the routing path and the response received from all computer centers indicates that the computing resource is not available, the process involves issuing a failure message to the user terminal.

[0047] The invention relates, according to a second aspect, to a device for managing a request for the allocation of a computing resource in a cloud computing system, comprising at least two computing centers connected to each other by a communication network, the allocation request being issued by a user terminal to the communication network via an access device allowing the user terminal to access said communication network via an access network and being configured to determine a routing path of the request to a predefined service address for the cloud computing system.

[0048] According to the invention, the management device comprises a transmission module configured for: forward the allocation request, received from the user's terminal, to a first computer center; and; retransmit the allocation request to the adjacent computer center, this adjacent computer center being the next computer center in the routing path, if according to an indicative response of the availability of the computing resource, the computing resource is not available in the first computer center, and reiterate the transmission of the allocation request to an adjacent computer center until either a computer center responds that the resource is available or the request has been retransmitted to all computer centers in the routing path.

[0049] The invention relates, according to a third aspect, to an access device enabling a user's terminal to access a communication network comprising a management device conforming to the invention.

[0050] It should be noted that the communication network access device, through which the user's terminal accesses the cloud computing system, implements the system for managing a request to allocate a computing resource. In other words, the determination of the computing center that will allocate the computing resource is implemented by the communication network access infrastructure connecting the computing centers of the cloud computing system, that is to say, independently of the cloud computing system itself.

[0051] According to a fourth aspect, the invention relates to a computer program comprising a sequence of instructions for implementing the management process according to the invention, when loaded and executed by a processor.

[0052] Finally, the invention relates, according to a fifth aspect, to a computer-readable information carrier on which is recorded a computer program comprising a sequence of instructions for implementing the management process according to the invention, when loaded into and executed by a processor.

[0053] The management device, the access device, the computer program and the information medium have characteristics and advantages similar to those described previously in relation to the process of managing an allocation request.

[0054] Other features and advantages of the invention will become apparent in the description below.

[0055] The attached drawings are given as non-exhaustive examples: [ Fig 1 ] there figure 1 represents a cloud-based computer system capable of implementing the processes according to the invention, [ Fig 2 ] there figure 2 illustrates steps in a process for managing an allocation request according to an embodiment of the invention, [ Fig 3 ] there figure 3 illustrates a hardware architecture capable of implementing the management process according to the invention, and [ Fig 4 ] there figure 4 illustrates exchanges between entities of the system according to an example of implementation of the process of managing an allocation request.

[0056] The invention finds its application in cloud computing systems. It is particularly applicable in the context of requests for the use, for example on demand, of infrastructure offered by a cloud service provider. This service is known as " 'infrastructure en tant que service » (IaaS, for “Infrastructure as a Service).

[0057] However, the invention also applies to the demand for other types of services, such as " la plateforme en tant que service » (PaaS, for “Platform as a Service”) or « le logiciel en tant que Service » (SaaS, for “Software as a Service”).

[0058] There figure 1 This represents a cloud computing system 1 capable of implementing the management method according to the invention. The cloud computing system 1 shown comprises several data centers (DCs). In the example shown, the cloud computing system 1 comprises three data centers, DC1, DC2, and DC3. These data centers, DC1, DC2, and DC3, are interconnected by a communication network or belong to different interconnected communication networks. In other words, the data centers DC1, DC2, and DC3 communicate with each other via one or more communication networks.

[0059] A user's terminal 10 uses the services provided by a cloud computing system 1. To use these services, user's terminal 10 must generate and send requests to cloud computing system 1, requesting the provision of services. A request for the provision of a service is a request for the allocation of a computing resource. By a request for the allocation of a computing resource, we mean a request by which user's terminal 10 requests that a computing resource be allocated to it in cloud computing system 1.

[0060] When a user subscribes to a service provider's services, the provider supplies the user with the necessary configuration data so that the user's terminal can access the services offered by the service provider. Specifically, the user's terminal is configured to send requests for a service or for the allocation of a computing resource in the cloud computing system to a service address. Thus, the terminal sends requests for the allocation of computing resources to this service address.

[0061] One of the data centers DC1, DC2, or DC3 can handle the management of the user's customer account. Sometimes, the user's customer account is managed by an infrastructure independent of the data centers.

[0062] User terminal 10 accesses communication network 200 via an access device 20. The access device 20 enables user terminal 10 to access communication network 200 through an access network 201. In a known manner, this access device 20 is configured to implement the routing of data sent by user terminal 10 to communication network 200, in particular to infrastructure addresses in the communication network.

[0063] When user terminal 10 issues a request to allocate a computing resource, the access device 20, using standard methods known to those skilled in the art, establishes the routing plan or path 300 between user terminal 10 and the service address predefined by the service provider. The routing path 300 contains the addresses of the data centers DC1, DC2, and DC3 of the cloud computing system 1 used to route the request to the service address.

[0064] According to the routing path 300 established by the access device 20, the allocation request issued by the user's terminal is transmitted to a first data center DC1, then to a second DC2, and finally to a third DC3. In the illustrated embodiment, the service address corresponds to the address of the third and final data center DC3.

[0065] Of course, this example is purely illustrative; the number of computer centers may be different.

[0066] It should be noted that, as is known to those skilled in the art, the routing path configured at access device 20 is the optimal path for forwarding the request to the service address. Thus, the first data center to which the request is destined is the data center located closest to the user's terminal. In other words, the data centers in routing path 300 are ordered from the closest to the user's terminal 10 to the furthest.

[0067] The user terminal 10 can be any type of device configured to access a communication network such as the Internet via the access device 20. The user terminal 10 is typically a computer, laptop, desktop, or workstation. However, it can also be a smartphone or other mobile communication device, such as a tablet.

[0068] The access device 20 is typically a home or residential gateway or router that allows the user terminal 10 to access the communication network 200, such as the Internet. The access network 201 can be a mobile phone network, a fixed telephone network providing ADSL (Asymmetric Digital Subscriber Line) Internet access, or a private network such as a LAN (Local Area Network) or WLAN (Wireless Local Area Network).

[0069] There figure 2 illustrates steps in a process for managing an allocation request in accordance with an embodiment of the invention.

[0070] The process of managing an allocation request is implemented by the access device 20. It should be noted that the allocation of the IT resource is implemented by a computer center DC1, DC2, DC3.

[0071] When the access device 20 receives an allocation request (E01) Req from a computing resource originating from user terminal 10, it transmits E10 the allocation request Req to a first computer center DC1. The first computer center DC1 is the first computer center in the routing path 300 established by the access device 20. In one embodiment, the first computer center DC1 is the computer center closest to the user terminal 10.

[0072] When the first data center, DC1, receives the request to allocate the computing resource from access device 20, it performs a check to verify the availability of the computing resource. This type of check is known to those skilled in the art and does not need to be described here.

[0073] Depending on the result of the verification, the first IT center DC1 implements the allocation of the IT resource and / or sends an indicative response of the availability of said resource to the access device 20. In particular, when, according to the result of the verification, the IT resource is available in the first IT center DC1, the allocation of the IT resource is implemented.

[0074] When, according to the result of said verification, the computing resource is not available, the computing center sends a response to the access device 20 indicating that the computing resource is unavailable. In one embodiment, this response includes the allocation request previously transmitted by the access device 20.

[0075] If access device 20 receives E02 a response from the first computer center indicating that the computing resource is not available in the first computer center DC1, it forwards E11 the allocation request to the next computer center in routing path 300. In the embodiment shown, the next computer center is the second computer center DC2, that is, the next computer center receiving the request along routing path 300 established by access device 20 to reach the service address.

[0076] The second data center, DC2, performs the same operations as the first data center; that is, it checks if the computing resource requested by the user is available, and if so, it allocates the resource. If the computing resource is not available, it informs the access device, which then forwards the request to the next data center, namely the third data center, DC3.The third computer center implements the same operations as the first and second computer centers so that either the computing resource is allocated by the third computer center DC3 or the allocation request is forwarded by the access device to the next computer center and so on until a computer center has the availability to allocate the computing resource requested by the user terminal, or all computer centers in the routing path have been solicited to verify whether they can allocate the requested computing resource.

[0077] It should be noted that when all the data centers in the routing path have been queried, the last data center, DC3 in routing path 300, or the data center furthest from the user's terminal, was queried.

[0078] Thus, the management process implements the retransmission of the allocation request to the next computer center DC1, DC2, DC3 in said routing path 300, until either the computing resource is available in a computer center DC1, DC2, DC3 or the request has been retransmitted E11, E12, E13 to the computer center DC3 furthest from the user terminal, or to all of the computer centers DC1, DC2, DC3.

[0079] In another embodiment not shown, if according to the indicative availability response the computing resource is not available in a computing center, the allocation request is further forwarded by the computing center to the adjacent computing center.

[0080] In this embodiment, the allocation request is transmitted to the next computing center via the access device and the additional computing center. This provides redundant means for implementing the transmission of the allocation request.

[0081] If the computing resource is not available in any of the additional computing centers DC1, DC2, DC3, the access device 20 sends an error message to the user terminal 10. This message contains information representative of the failure of the request to allocate the computing resource.

[0082] The messages exchanged between the access device and the computer centers, as well as between the access device and the communication terminal, conform to communication protocols such as those used by the Internet network.

[0083] There figure 3 schematically illustrates a hardware architecture of a management device capable of implementing the management process according to the invention.

[0084] In one embodiment, the management device is embedded or integrated into the access device 20.

[0085] The management device 20 includes a communication bus 100 to which the following are connected: a processing unit 101, named in the figure CPU (for "Central Processing Unit") and which may include one or more processors; a non-volatile memory 102, for example ROM (for "Read Only Memory"), EEPROM (for "Electrically Erasable Read Only Memory") or a Flash memory; a random access memory 103 or RAM (for "Random Access Memory"); an input / output interface 104, named in the figure I / O (for "Input / Output"), for example keys or buttons, a screen, a keyboard, a mouse or another pointing device such as a touch screen or a remote control allowing a user to interact with access device 20 via a graphical interface or human-machine interface; and a communication interface 105, named COM in the figure, adapted to exchange data for example with a computer center DC1, DC2, DC3 via a communication network 200.

[0086] The random access memory 103 includes registers adapted for storing variables and parameters created and modified during the execution of a computer program comprising instructions for implementing the management method according to the invention. The instruction codes of the program stored in non-volatile memory 102 are loaded into RAM 103 for execution by the CPU 101.

[0087] The non-volatile memory 102 is, for example, a rewritable memory of the EEPROM or Flash type that can constitute a medium within the meaning of the invention, that is to say, that can include a computer program comprising instructions for implementing the management method according to the invention. The rewritable memory may include the routing path 300 of the allocation request to the service address or parameters specifying the service address.

[0088] This program defines, through its instructions, functional modules of the management system that are implemented and / or control the hardware components described above. These modules include, in particular, a transmission module configured to: transmit (E10) the allocation request (Req), received from the user's terminal (10), to a first data center (DC1), the first data center (DC1) being the first in the determined routing path (300); and; retransmit (E11) the allocation request (Req) to the adjacent data center (DC2), this adjacent data center being the next data center in the routing path (300), if according to a response ( Rp ) indicative of the availability of the computing resource, the computing resource is not available in the first computing center (DC1), and retransmission (E12) of the allocation request to an adjacent computing center (DC3) until either a computing center responds that the resource is available or the request (Req) was retransmitted to all computer centers (DC1, DC2, DC3) of the routing path (300).

[0089] The aforementioned modules and means are controlled by the processor of the processing unit 101. They can take the form of a program executable by a processor, or a hardware form, such as a specialized integrated circuit (known in Anglo-Saxon terminology as ASIC for "Application-Specific Integrated Circuit"), a system on chip (known in Anglo-Saxon terminology as SoC for "System On Chip"), or an electronic component of the type programmable logic circuit, such as a component of type FPGA (for "Field-Programmable Gate Array").

[0090] Each DC1, DC2, DC3 data center has a hardware architecture similar to that described for the management device and will not be described.

[0091] In particular, each computer center includes: a verification module configured to check, upon receipt of said allocation request, the availability of the computing resource, an allocation module configured to allocate said computing resource, when said computing resource is available, and a response module configured to send an indicative response of the availability of said resource to the access device.

[0092] The user terminal 10 also includes a communication bus to which are connected a processing unit or microprocessor, non-volatile memory, random access memory or RAM, and a communication interface adapted in particular to exchange data with the access device 20.

[0093] There figure 4 illustrates an example of exchanges between user terminal 10, access device 20 and data centers DC1, DC2, DC3.

[0094] In the example shown, the third data center DC3 has the availability to allocate the computing resource requested by the user.

[0095] Of course, this example is given for purely illustrative purposes, the exchanges being different in different examples.

[0096] In this example, user terminal 10 issues an allocation request Req of a resource, this request Req being received E01 by access device 20. Access device 20 forwards E10 to the first data center DC1. This receives the request Req and checks the availability of the computing resource. Since the computing resource is unavailable, the DC1 computing center responds to access device 20. This device 20 receives the response E03 and retransmits the request E11. Req to the second computer center DC2.

[0097] The second data center, DC2, receives the request. Req and checks the availability of the IT resource E20. Since the IT resource is unavailable, the DC2 IT center responds to the access device 20. DC2 receives the response E03 and retransmits the request E11. Req to the third computer center DC3.

[0098] The third data center, DC3, receives the request. Req and checks the availability of the E20 computing resource. With the computing resource available, the third computing center DC3 implements the E21 allocation of the computing resource.

[0099] The user can thus use the requested computing resource which is allocated in a computer center selected by the access device 20 of his access provider to the communication network 200, such as the Internet.

Claims

1. Method for managing an allocation request to allocate a computing resource in a cloud computing system (1), said request being sent by a terminal (10) of a user, the cloud computing system comprising at least two data centres (DC1, DC2, DC3) connected to one another by a communication network (200), said terminal (10) of the user accessing said communication network (200) via an access device (20) by way of an access network (201), said management method being implemented by the access device (20), said access device (20) being configured so as to determine a routing path (300) for routing the request to a predefined service address for the cloud computing system (1), which address is provided by the service provider, characterized in that said management method comprises: - transmitting (E10) said allocation request (Req), received from the terminal (10) of the user, to a first data centre (DC1), said first data centre (DC1) being the first one in the determined routing path (300); and - if, according to a response (Rp) indicating the availability of the computing resource, the computing resource is not available in the first data centre (DC1), retransmitting (E11) said allocation request (Req) to an adjacent data centre, this adjacent data centre (DC2) being the next data centre in the routing path (300), the retransmission (E12) of the allocation request (Req) to an adjacent data centre (DC3) being reiterated until either a data centre (DC1, DC2, DC3) responds that the computing resource is available or the request (Req) has been retransmitted to all of the data centres (DC1, DC2, DC3) on the routing path (300), the service address being that of the last data centre (DC3) in the routing path (300).

2. Management method according to Claim 1, characterized in that, if the computing resource is not available in a data centre, said management method furthermore comprises receiving (E02) the response indicating the availability of the computing resource.

3. Management method according to Claim 1, characterized in that, when the received response (E02) indicates that the resource is not available, the response contains said transmitted (E10) or retransmitted (E11, E12) allocation request (Req).

4. Management method according to Claim 1, characterized in that, when the received response (E02) indicates that the resource is not available, the response contains a message indicating that the requested computing resource is not available.

5. Management method according to Claim 1, characterized in that, when the request (Req) has been retransmitted to all of the data centres (DC1, DC2, DC3) on the routing path (300) and the response received (Rp) from all of the data centres (DC1, DC2, DC3) indicates that the computing resource is not available, the method comprises sending a failure message to the user terminal (10).

6. Device for managing an allocation request to allocate a computing resource in a cloud computing system (1), the cloud computing system comprising at least two data centres (DC1, DC2, DC3) connected to one another by a communication network (200), the allocation request (Req) being sent by a terminal (10) of a user to the communication network (200) via an access device (20) by way of an access network (201), thereby allowing the terminal (10) of the user to access said communication network (200), said management device forming part of said access device (20), said access device (20) being configured so as to determine a routing path (300) for routing the request (Req) to a predefined service address for the cloud computing system (1), which address is provided by the service provider, characterized in that said management device comprises a transmission module configured so as to: - transmit (E10) the allocation request (Req), received from the terminal (10) of the user, to a first data centre (DC1), said first data centre (DC1) being the first one in the determined routing path (300); and - retransmit (E11) the allocation request (Req) to the adjacent data centre (DC2), this adjacent data centre being the next data centre in the routing path (300), if, according to a response (Rp) indicating the availability of the computing resource, the computing resource is not available in the first data centre (DC1); and - reiterate the transmission (E12) of the allocation request to an adjacent data centre (DC3) until either a data centre responds that the resource is available or that the request (Req) has been retransmitted to all of the data centres (DC1, DC2, DC3) on the routing path (300), the service address being that of the last data centre (DC3) in the routing path (300).

7. Computer program comprising a sequence of instructions for implementing the management method according to one of Claims 1 to 5 when it is loaded and executed by a processor.

8. Computer-readable information medium on which there is recorded a computer program comprising a sequence of instructions for implementing the management method according to one of Claims 1 to 5 when it is loaded into and executed by a processor.