WO2024051424A1 - 分配网络资源的方法、装置及存储介质 - Google Patents

分配网络资源的方法、装置及存储介质 Download PDF

Info

Publication number
WO2024051424A1
WO2024051424A1 PCT/CN2023/111572 CN2023111572W WO2024051424A1 WO 2024051424 A1 WO2024051424 A1 WO 2024051424A1 CN 2023111572 W CN2023111572 W CN 2023111572W WO 2024051424 A1 WO2024051424 A1 WO 2024051424A1
Authority
WO
WIPO (PCT)
Prior art keywords
return
live broadcast
flow
broadcast service
traffic
Prior art date
Application number
PCT/CN2023/111572
Other languages
English (en)
French (fr)
Inventor
林思洁
彭文
杨俊彦
杨昌鹏
Original Assignee
华为云计算技术有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 华为云计算技术有限公司 filed Critical 华为云计算技术有限公司
Publication of WO2024051424A1 publication Critical patent/WO2024051424A1/zh

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/20Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
    • H04N21/21Server components or server architectures
    • H04N21/218Source of audio or video content, e.g. local disk arrays
    • H04N21/2187Live feed
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/20Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
    • H04N21/23Processing of content or additional data; Elementary server operations; Server middleware
    • H04N21/238Interfacing the downstream path of the transmission network, e.g. adapting the transmission rate of a video stream to network bandwidth; Processing of multiplex streams
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/20Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
    • H04N21/25Management operations performed by the server for facilitating the content distribution or administrating data related to end-users or client devices, e.g. end-user or client device authentication, learning user preferences for recommending movies
    • H04N21/262Content or additional data distribution scheduling, e.g. sending additional data at off-peak times, updating software modules, calculating the carousel transmission frequency, delaying a video stream transmission, generating play-lists
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/40Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
    • H04N21/47End-user applications
    • H04N21/472End-user interface for requesting content, additional data or services; End-user interface for interacting with content, e.g. for content reservation or setting reminders, for requesting event notification, for manipulating displayed content

Definitions

  • the present application relates to the field of communications, and in particular to a method, device and storage medium for allocating network resources.
  • the operator of the live broadcast service uses the content distribution network built by the cloud vendor to transmit the live broadcast service.
  • the anchor user of the live broadcast service sends a data stream to the content distribution network, and the audience user of the live broadcast service pulls the data stream from the content distribution network. In this way, the content distribution network is used to transmit the live broadcast service.
  • Operators of live broadcast services can apply to cloud vendors to use content distribution networks to transmit live broadcast services within a period of time.
  • Cloud vendors need to allocate network resources to the applied live broadcast services in the content distribution network.
  • the content distribution network uses the allocated network resources. network resources to transmit the live broadcast service.
  • the current way of allocating network resources is fixed and not flexible enough, resulting in low flexibility in allocating network resources.
  • This application provides a method, device and storage medium for allocating network resources to improve the flexibility of allocating network resources.
  • the technical solutions are as follows:
  • this application provides a method for allocating network resources.
  • the management device obtains the business demand information of the live broadcast service and the data stream attribute information of the live broadcast service in the first time period.
  • the business demand information is used To describe the situation in which the content distribution network transmits the data stream of the live broadcast service in the first time period.
  • the first time period is after the current time.
  • the content distribution network includes a central node and an edge node. The central node is used to send data from the live broadcast service to the edge node. The data stream of the anchor user, and the edge node is used to send the data stream to the audience user who requested the data stream.
  • the management device Based on the business demand information and data flow attribute information, the management device obtains the back-to-source traffic and outgoing traffic generated by the live broadcast service in the first time period.
  • the back-to-source traffic is the traffic generated by the data flow sent by the central node, and the outgoing traffic is the edge traffic.
  • the management device allocates network resources in the content distribution network based on the return-to-source traffic and the outflow traffic.
  • the traffic generated by the live broadcast service in the first time period is obtained based on the business attribute information and business demand information of the live broadcast service.
  • the data flow attributes of the live broadcast service are referenced when obtaining the traffic.
  • the data flow attributes of the segment live broadcast service may be different, and the return-to-source traffic obtained based on different data flow attribute information is different.
  • Network resources are allocated based on the obtained return-to-source traffic and outbound traffic, thereby improving the flexibility of allocating network resources.
  • the fees required for the content distribution network to transmit the live broadcast service are obtained based on the return-to-source traffic and outbound traffic.
  • the return-to-source traffic obtained based on different data flow attribute information is different. Obtaining fees based on the obtained return-to-source traffic and outbound traffic can improve the flexibility of billing.
  • the service demand information includes the outflow traffic
  • the data flow attribute information includes a first ratio between cold flow, warm flow and hot flow of the live broadcast service.
  • the return rate corresponding to the live broadcast service in the first time period is obtained based on the first ratio, and the return rate is the ratio between the return traffic and the outbound traffic.
  • the return rate corresponding to the live broadcast service can be accurately obtained based on the first ratio. Based on the return rate, the accuracy of obtaining the return traffic can be improved.
  • the cold flow is a data flow for which the number of requested audience users does not exceed the cold flow threshold
  • the warm flow is a data flow for which the number of requested audience users exceeds the cold flow threshold and does not exceed the hot flow threshold
  • the hot flow is A data stream whose number of requested audience users exceeds the hot stream threshold
  • the hot stream threshold is greater than the cold stream threshold
  • the business demand information also includes the total number of data streams sent to the central node by anchor users who need live broadcast services in the first time period. Based on the total number of data streams and the first ratio, the return rate corresponding to the live broadcast service in the first time period is obtained. Since the total number of data streams sent by anchor users of live broadcast services to the central node will also affect the return-to-origin rate of live broadcast services, based on the total number of data streams and the first ratio, the accuracy of obtaining the return-to-origin rate can be improved.
  • the number of edge nodes corresponding to each flow type in the first time period is obtained, and each Streaming types have different number of viewers.
  • the number of standard data flows corresponding to each flow type and the first ratio the number of data flows corresponding to each flow type in the first time period is obtained.
  • the return rate corresponding to the live broadcast service in the first time period is obtained. In this way, the return rate corresponding to the live broadcast service in the first time period is obtained based on the first ratio.
  • the number of edge nodes corresponding to each flow type in the first time period is obtained. Based on network information, the accuracy of obtaining the number of edge nodes can be improved, thereby improving the accuracy of obtaining the return rate.
  • the historically acquired business requirement information does not include the business requirement information of the live broadcast service and/or the historically acquired data stream attribute information does not include the data stream attribute information of the live broadcast service, based on the first The ratio is used to obtain the return rate corresponding to the live broadcast service in the first time period.
  • the service demand information includes the outflow traffic
  • the data flow attribute information includes a first ratio between cold flow, warm flow and hot flow included in the live broadcast service.
  • the return rate is the ratio between the return traffic and the outbound traffic.
  • the return rate prediction model Based on the return-to-source rate and outbound traffic, obtain the return-to-source traffic. Since the data flow attribute information of the live broadcast service will affect the return rate of the live broadcast service, the return rate corresponding to the live broadcast service can be accurately obtained based on the return flow, first ratio and return rate prediction model.
  • the return rate prediction model The efficiency of obtaining the return-to-origin rate is high, and it also improves the efficiency of allocating network resources.
  • the business demand information also includes the total number of data streams sent to the central node by anchor users who need live broadcast services in the first time period. Based on the total number of data streams, return traffic, first ratio and return rate prediction model, obtain the return rate corresponding to the live broadcast service in the first time period. Since the total number of data streams sent by anchor users of live broadcast services to the central node will also affect the return-to-origin rate of live broadcast services, introducing the total number of data streams can improve the accuracy of obtaining the return-to-origin rate.
  • the total number of edge nodes included in the content distribution network in the first time period is obtained. Based on the total number of edge nodes, the total number of data flows, return flow, the first ratio and the return rate prediction model, the return rate corresponding to the live broadcast service in the first time period is obtained. Among them, the total number of edge nodes included in the content distribution network will affect the return rate of the live broadcast service. Introducing the total number of edge nodes can improve the accuracy of obtaining the return rate.
  • At least one derived characteristic corresponding to the live broadcast service is obtained based on the outflow traffic, the total number of data flows, and/or the total number of edge nodes. Based on at least one derived feature, the total number of edge nodes, the total number of data streams, return traffic, the first ratio and the return rate prediction model, obtain the return rate corresponding to the live broadcast service in the first time period. Among them, introducing at least one derived feature can further improve the accuracy of obtaining the return rate.
  • At least one derived feature includes one or more of the following: the total number of audience users of the live broadcast service in the first time period, the average audience corresponding to the data stream sent by the anchor user of the live broadcast service The number of users, the average outflow traffic corresponding to the data streams sent by the anchor users of the live broadcast service, the average number of data streams sent by the edge nodes in the first time period, and the average number of audience users requesting data streams from the edge nodes in the first time period , or, the average outflow traffic sent by the edge node during the first time period.
  • the total number of audience users is obtained. and/or, based on the total number of audience users and the total number of data streams to obtain the average number of audience users. And/or, based on the outflow traffic and the total number of data flows, obtain the average outflow traffic. And/or, based on the total number of data flows and the total number of edge nodes, obtain the average number of data flows. And/or, based on the total number of audience users and the total number of edge nodes, obtain the average number of audience users. And/or, obtain the average outflow traffic based on the outflow traffic and the total number of edge nodes. Thus, obtaining the at least one derived feature is achieved.
  • the first ratio and The return rate prediction model obtains the return rate corresponding to the live broadcast service in the first time period to improve the accuracy of obtaining the return rate.
  • a return-to-origin rate prediction model is obtained based on at least one training sample, because each training sample includes the return-to-origin traffic obtained historically, the first allocation ratio obtained historically, and the return-to-origin rate based on historical acquisition.
  • the return rate obtained from the source flow rate and the first ratio can improve the accuracy of the return rate prediction model.
  • this application provides a device for allocating network resources, for performing the method in the first aspect or any possible implementation of the first aspect.
  • the apparatus includes a unit for performing the method in the first aspect or any possible implementation of the first aspect.
  • the present application provides a computer device, including at least one processor and a memory, the at least one processor being coupled to the memory, reading and executing instructions in the memory, to implement the first aspect or the third aspect. method in any possible implementation of one aspect.
  • the present application provides a computer program product.
  • the computer program product includes a computer program stored in a computer-readable storage medium, and the computing program is loaded by a processor to implement the above first aspect or the third aspect.
  • any possible implementation method any possible implementation method.
  • the present application provides a computer-readable storage medium for storing a computer program, which is loaded by a processor to execute the method of the above-mentioned first aspect or any possible implementation of the first aspect.
  • this application provides a chip, including a memory and a processor.
  • the memory is used to store computer instructions
  • the processor is used to call and run the computer instructions from the memory to execute the above first aspect or any possibility of the first aspect. method of implementation.
  • Figure 1 is a schematic structural diagram of a content distribution network provided by an embodiment of the present application.
  • Figure 2 is a schematic structural diagram of a network architecture provided by an embodiment of the present application.
  • FIG. 3 is a schematic structural diagram of another content distribution network provided by an embodiment of the present application.
  • Figure 4 is a flow chart of a method for allocating network resources provided by an embodiment of the present application.
  • Figure 5 is a flow chart of another method for allocating network resources provided by an embodiment of the present application.
  • Figure 6 is a flow chart of another method for allocating network resources provided by an embodiment of the present application.
  • Figure 7 is a flow chart of another method for allocating network resources provided by an embodiment of the present application.
  • Figure 8 is a schematic structural diagram of a black box distribution component provided by an embodiment of the present application.
  • Figure 9 is a schematic structural diagram of a device for allocating network resources provided by an embodiment of the present application.
  • Figure 10 is a schematic structural diagram of a computer device provided by an embodiment of the present application.
  • the live broadcast platform uses basic cloud services provided by cloud vendors, mainly using the content distribution network built by cloud vendors to support the live broadcast business.
  • Cloud vendors will build content distribution network nodes around the world.
  • the nodes of the content distribution network include origin nodes, central nodes, edge nodes, etc.
  • the content distribution network is an end-to-end full-link network.
  • the anchor users of the live broadcast service access from the edge nodes of the content distribution network and send the data stream of the live broadcast service to the content distribution network. Audience users of the live broadcast service will also access from the edge nodes of the content distribution network, and obtain the data stream of the live broadcast service from the content distribution network for viewing.
  • the live broadcast platform pays the cloud vendor based on the traffic generated by the live broadcast service budgeted by the cloud vendor.
  • the content distribution network 100 includes an origin node 101 , a central node 102 and an edge node 103 .
  • the origin node 101 communicates with at least one central node 102, and for each central node 102, the central node 102 communicates with at least one edge node 103.
  • the edge nodes 103 in the content distribution network 100 are distributed in various places.
  • the anchor users of the live broadcast service access the edge nodes 103 of the content distribution network 100, and the anchor users send the data stream of the live broadcast service to the edge nodes 103.
  • the edge node 103 is configured to receive the data flow and send the data flow to the central node 102 communicating with the edge node 103 .
  • the central node 102 is used to receive the data stream and send the data stream to the source station node 101.
  • the source station node 101 is used to receive the data stream and distribute the data stream to each central node 102 in the content distribution network 100 except the central node 102.
  • Each central node 102 in the content distribution network 100 can obtain the data stream and cache the data stream.
  • Audience users of the live broadcast service access the edge node 103 of the content distribution network 100, and the audience users request data streams of the live broadcast service from the edge node 103.
  • the edge node 103 If the edge node 103 caches the data stream, the edge node 103 is used to send the cached data stream to the viewer user. If the edge node 103 does not cache the data stream, the edge node 103 is used to request the data stream from the central node 102 communicating with the edge node 103, receive the data stream sent by the central node 102, and send the data stream to the audience user. data stream and cache that data stream.
  • the data flow of the live broadcast service is classified into cold flow, warm flow or hot flow.
  • Cold flow is the data flow for which the number of requested audience users does not exceed the cold flow threshold.
  • Warm flow is the data flow for which the number of requested audience users exceeds the cold flow threshold and does not exceed the hot flow threshold.
  • Hot flow is the data flow for which the number of requested audience users exceeds the hot flow threshold. flow, the hot flow threshold is greater than the cold flow threshold.
  • the hot flow threshold and/or the cold flow threshold are thresholds defined by the content distribution network 100 .
  • the hot flow threshold and/or the cold flow threshold defined by the content distribution network 100 may be different in different time periods. That is to say, the hot flow threshold and/or the cold flow threshold are not fixed.
  • the total number of edge nodes included in the content distribution network 100 may be different. That is to say, in different time periods, the number of edge nodes 103 included in the content distribution network 100 may be increased, or the total number of edge nodes included in the content distribution network 100 may be reduced. 100 edge nodes.
  • the anchor user and the audience user of the live broadcast service may be two clients with different functions for the live broadcast service.
  • the anchor user can be the anchor client, and the audience user can be the audience client.
  • the network architecture 200 includes a live broadcast platform 201, a scheduling platform 202, and a management device 203.
  • the management device 203 communicates with the live broadcast platform 201 and the scheduling platform 202 respectively.
  • the scheduling platform 202 and the management device 203 belong to the operator of the above-mentioned content distribution network 100.
  • the scheduling platform 202 is used to obtain network information of the content distribution network 100 in the first time period.
  • the network information includes, for example, one or more, at the first time Information such as the total number of edge nodes included in the content distribution network 100 in the segment, the cold flow threshold of the content distribution network 100 in the first time period, or the hot flow threshold of the content distribution network 100 in the first time period.
  • the first time period may be a time period after the current time.
  • the live broadcast platform 201 is used to apply to the cloud vendor to use the content distribution network 100 to transmit the live broadcast service within the first time period.
  • it sends the live broadcast service within the first time period to the management device 203.
  • the business requirement information of the service and the data flow attribute information of the live broadcast service are used to apply to the cloud vendor to use the content distribution network 100 to transmit the live broadcast service within the first time period.
  • the management device 203 is configured to receive the business demand information and the data flow attribute information, and obtain the traffic generated by the live broadcast service in the first time period based on the business demand information and the data flow attribute information. Allocate network resources in the content distribution network based on the traffic generated by the live broadcast service, so that the content distribution network transmits the data stream of the live broadcast service based on the network resources in the first time period.
  • the management device 203 may also obtain the fees required by the content distribution network 100 to transmit the live broadcast service based on the traffic. In this way, the operator of the live broadcast service pays the cloud vendor a fee based on the fee obtained by the management device 200 .
  • the first time period may be one week, one month, two months, or half a year, etc.
  • the live broadcast platform 201 may apply to the cloud vendor at the end of each month to use the content distribution network 100 to transmit live broadcasts in the next month.
  • the first time period is the next month.
  • the management device 203 may be used to obtain the network information of the content distribution network 100 in the first time period from the scheduling platform 202, and obtain the network information of the content distribution network 100 in the first time period based on the network information, the business demand information and the data flow attribute information.
  • the traffic generated by the live broadcast service in the first time period includes the back-to-source traffic and outbound traffic generated by the live broadcast service in the first time period.
  • the back-to-origin traffic is the traffic generated by each data flow sent by the central node included in the content distribution network 100 to the edge nodes included in the content distribution network 100. That is, the back-to-source traffic is equal to the sum of the data volume of each data flow, and the outflow traffic is the content
  • the traffic generated by each data stream sent by the edge nodes included in the distribution network 100 to the audience users of the live broadcast service, that is, the outflow traffic is equal to the sum of the data amounts of each data stream.
  • the ratio between return flow and outflow flow is called return rate.
  • the content distribution network 100 includes a central node A, a central node B, edge nodes 1 and 2 communicating with the central node A, and an edge node 3 communicating with the central node B.
  • Audience user 1 and audience user 2 access edge node 1 and request data stream 1 from edge node 1.
  • Data stream 1 is the data stream sent by edge node 1 and received by central node A.
  • Audience user 3 and audience user 4 access edge node 2 and request data flow 2 from edge node 2.
  • Data flow 2 is the data flow sent by center node A received by edge node 2.
  • Audience user 5 accesses edge node 3 and requests data stream 3 from edge node 3.
  • Data stream 3 is the data stream sent by edge node 3 and received by central node B.
  • the return-to-source traffic is equal to the data volume of data stream 1 received by edge node 1, and the data volume of data stream 2 received by edge node 2.
  • the sum of the data volume and the data volume of data flow 3 received by edge node 3, that is, the return-to-source traffic is equal to 6M.
  • the outflow traffic is equal to the data volume of data stream 1 sent by edge node 1 to viewer user 1, the data volume of data stream 1 sent by edge node 1 to viewer user 2, and the data volume of data stream 2 sent by edge node 2 to viewer user 3.
  • the sum of the data volume of data stream 2 sent by edge node 2 to viewer user 4 and the data volume of data stream 3 sent by edge node 3 to viewer user 5, that is, the outflow traffic is equal to 10M.
  • the return-to-source rate is equal to the ratio between the return-to-source flow and the outflow flow, that is, the return-to-source rate is equal to 0.6.
  • the management device 203 includes a first price corresponding to outbound traffic and a second price corresponding to return-to-source traffic.
  • the management device 203 obtains the fee required for the content distribution network 100 to transmit the live broadcast service based on the outflow traffic, the return flow, the first price, and the second price.
  • the management device 203 includes two network resource allocation methods, including a white box allocation method and a black box allocation method.
  • the management device 203 selects a distribution method based on the business demand information and data flow attribute information of the live broadcast service, and uses the selected distribution method to obtain the content distribution network 100 to transmit the live broadcast service. the traffic generated.
  • the white-box allocation method is a method of obtaining the traffic based on the first mapping relationship included in the management device 203.
  • the white-box allocation method please refer to the method 400 shown in FIG. 4, which will not be described in detail here.
  • the first mapping relationship is used to save the mapping relationship between the range of the number of audience users, the amount of standard data flow data, and the number of standard edge nodes.
  • the first mapping relationship defines multiple stream types, and each stream type corresponds to a different number of audience users.
  • the first mapping relationship includes records corresponding to each stream type. Multiple stream types are in one-to-one correspondence with multiple audience user number ranges. Each stream type corresponds to a different audience user number range.
  • the multiple stream types include a first stream type, and the number of audience users requesting the data stream of the first stream type is within the range of the number of audience users corresponding to the first stream type.
  • the record corresponding to the first stream type includes the range of the number of audience users, the number of standard data streams, and the number of standard edge nodes corresponding to the first stream type.
  • the number of standard data flows corresponding to the first flow type is used to indicate the number of data flows belonging to the first flow type that the edge node of the content distribution network obtains from the central node of the content distribution network.
  • the standard number of edge nodes corresponding to the first flow type is used to indicate the average number of edge nodes for each data flow requesting the first flow type.
  • Each value included in the record corresponding to the first stream type is a baseline value.
  • the record corresponding to flow type 1 includes the range of the number of audience users "[1,3)", the number of standard data flows "103246” and the number of standard edge nodes "2.1".
  • the range of the number of audience users "[1,3)” indicates that the number of audience users for each data stream of request flow type 1 is greater than or equal to 1 and less than 3.
  • the standard number of data flows "103246” indicates that the edge node of the content distribution network requests 103246 data flows belonging to flow type 1 from the central node of the content distribution network.
  • the standard number of edge nodes "2.1” means that the average number of edge nodes per data flow for request flow type 1 is 2.1. The meanings of other records in the first mapping relationship shown in Table 1 will not be listed one by one.
  • the data flow of the live broadcast service is divided into three main types: cold flow, warm flow and hot flow.
  • the first mapping relationship divides the data stream of the live broadcast service into multiple stream types based on the number of requested audience users, and the multiple stream types are sub-types of the main type.
  • the plurality of flow types include a first flow type. If the upper limit of the range of the number of audience users corresponding to the first flow type does not exceed the cold flow threshold, the first flow type is a subtype of the cold flow.
  • the first flow type is a subtype of warm flow. If the lower limit value of the range of the number of audience users corresponding to the first flow type exceeds the hot flow threshold, the first flow type is a subtype of the hot flow.
  • the black box allocation method is a method of obtaining the traffic based on the return rate prediction model included in the management device 203.
  • the return rate prediction module is an intelligent model trained in advance by the management device 203.
  • this black box allocation method please refer to the subsequent method 600 shown in Figure 6, which will not be described in detail here.
  • the management device 203 includes a decision-making component 2031 , a white box allocation component 2032 and a black box allocation component 2033 .
  • the management device 203 may be a cluster of computing devices, and the decision-making component 2031, the white-box allocation component 2032, and the black-box allocation component 2033 may be different computing devices.
  • the management device 203 may be a computing device, and the decision-making component 2031, the white-box allocation component 2032, and the black-box allocation component 2033 may be different modules of the computing device.
  • the decision-making component 2031 is used to select an allocation method based on the business demand information and data flow attribute information of the live broadcast service.
  • the white-box allocation component 2032 is triggered, and when the black-box allocation method is selected, the black-box allocation component 2033 is triggered. .
  • the white-box allocation component 2032 is used to obtain the traffic generated by the content distribution network transmitting the live broadcast service in a white-box allocation manner under the trigger of the decision-making component 2031.
  • the black box allocation component 2033 is used to obtain the traffic generated by the content distribution network transmitting the live broadcast service in a black box allocation manner under the trigger of the decision-making component 2031.
  • an embodiment of the present application provides a method 400 for allocating network resources.
  • the method 400 is applied to the network architecture 200 shown in Figure 2.
  • the method 400 is executed by the management device in the network architecture 200.
  • the method 400 adopts a white box allocation method, which includes the following processes of steps 401-404.
  • Step 401 The management device receives the business demand information of the live broadcast service and the data flow attribute information of the live broadcast service in the first time period.
  • the business demand information includes the outflow traffic that needs to be generated by the live broadcast service in the first time period.
  • the data flow attribute information The information includes a first ratio between multiple data streams of the live broadcast service, the multiple data streams include at least two of a cold stream, a warm stream, and a hot stream, and the first time period is located after the current time.
  • the business demand information may also include other information.
  • the business demand information may also include the total number of data streams sent by anchor users who need live broadcast services to the central node of the content distribution network in the first time period. Therefore, the service requirement information is used to describe the situation in which the content distribution network transmits the data stream of the live broadcast service in the first time period.
  • the management device receives an application request sent by the live broadcast platform.
  • the application request includes the business requirement information of the live broadcast service and the data flow attribute information of the live broadcast service in the first time period.
  • the application request is used to request the management device to allocate the first time period to the live broadcast service.
  • the management device may also obtain network information of the content distribution network from the scheduling platform.
  • the network information includes one or more of the following: the total number of edge nodes included in the content distribution network in the first time period, the content distribution network in the first time period The cold flow threshold, or the hot flow threshold of the content distribution network in the first time period, etc., the cold flow threshold is smaller than the hot flow threshold.
  • the management device receives the network information of the content distribution network sent by the scheduling platform.
  • schedule The platform may not send the network information of the content distribution network to the management device, or may continue to send the network information of the content distribution network to the management device.
  • the second time period is the time period before the first time period. If the network information of the content distribution network in the first time period is different from the network information of the content distribution network in the second time period, the scheduling platform sends the network information of the content distribution network to the management device.
  • Step 402 When the business requirement information historically received by the management device does not include the business requirement information of the live broadcast service and/or the data stream attribute information historically received by the management device does not include the data stream attribute information of the live broadcast service, the management device based on the first ratio Obtain the return rate corresponding to the live broadcast service in the first time period.
  • step 402 the business requirement information historically received by the management device does not include the business requirement information of the live broadcast service, the data stream attribute information historically received by the management device does not include the data stream attribute information of the live broadcast service, and/or the management device has historically received
  • the management device obtains the return rate corresponding to the live broadcast service in the first time period based on the first ratio.
  • step 402 when the service demand information also includes the total number of data streams, the management device obtains the return-to-origin rate corresponding to the live broadcast service in the first time period based on the total number of data streams and the first ratio.
  • the management device obtains the return-to-origin rate corresponding to the live broadcast service in the first time period through the following process.
  • the process includes the following operations 4021-4023.
  • the decision-making component included in the management device receives the business requirement information and data flow attribute information of the live broadcast service.
  • the business requirement information received in the management device history does not include the business requirement information of the live broadcast service and/or the management device history.
  • the white box allocation component of the management device is triggered to perform the following operations 4021-4023 to obtain the return-to-origin rate.
  • the management device obtains the target data stream corresponding to each stream type in the first time period based on the number of standard data streams corresponding to each stream type in the multiple stream types, the standard ratio of the live broadcast service, and the first ratio.
  • the number of flow types includes the first flow type.
  • the number of target data flows corresponding to the first flow type is the number of data flows of the first flow type that the edge node of the content distribution network requests to send from the central node of the content distribution network.
  • the standard ratio It is the ratio between at least two of cold flow, warm flow and hot flow.
  • the management device includes a first mapping relationship and the standard ratio.
  • the first mapping relationship includes the number of standard data flows corresponding to each flow type.
  • the standard ratio between the cold flow, warm flow and hot flow of the live broadcast service is expressed as Rc0:Rw0:Rh0; the first ratio between the cold flow, warm flow and hot flow of the live broadcast service is expressed as Rc:Rw:Rh.
  • the main type to which the first flow type belongs determine the data flow of the first flow type. Is it cold flow, warm flow or hot flow. If the main type of the first flow type is cold flow, the number of target data flows corresponding to the first flow type is equal to S*Rc/Rc0, where S is the number of standard data flows corresponding to the first flow type. If the main type of the first flow type is warm flow, the number of target data flows corresponding to the first flow type is equal to S*Rw/Rw0. If the main type of the first stream type is hot stream, the number of target data streams corresponding to the first stream type is equal to S*Rh/Rh0.
  • the management device includes the first mapping relationship as shown in Table 2 below, the cold flow threshold and hot flow threshold based on the content distribution network, and the audience corresponding to each flow type in Table 2.
  • the user number range determines that the main type of flow type 1 and the main type of flow type 2 are cold flow, the main type of flow type 3 and the main type of flow type 4 are warm flow, the main type of flow type 5 and the main type of flow type 6 are The main type is heat flow.
  • Table 3 Based on Rc0, Rc and the number of standard data flows S1 corresponding to flow type 1, the number of target data flows corresponding to flow type 1 is obtained as S1*Rc/Rc0.
  • the number of target data flows corresponding to flow type 2 is obtained as S2*Rc/Rc0.
  • the number of target data flows corresponding to flow type 3 is obtained as S3*Rw/Rw0.
  • the number of target data flows corresponding to flow type 4 is obtained as S4*Rw/ Rw0.
  • the number of target data streams corresponding to stream type 5 is obtained as S5*Rh/Rh0.
  • the number of target data streams corresponding to stream type 6 is obtained as S6*Rh/ Rh0.
  • the management device includes the total number of standard data streams of the live broadcast service. If the business requirement information also includes the total number of data flows, the management device obtains the data based on the number of standard data flows corresponding to each flow type, the standard ratio, the first ratio, the total data of the data flow and the total number of standard data flows. The number of target data flows corresponding to each flow type in the first time period.
  • the multiple flow types include the first flow type. If the main type of the first flow type is cold flow, the number of target data flows corresponding to the first flow type is equal to S*Q*Rc/(Q0*Rc0), and S is the first flow The number of standard data streams corresponding to the type, Q is the total number of data streams, and Q0 is the total number of standard data streams. If the main type of the first flow type is warm flow, the number of target data flows corresponding to the first flow type is equal to S*Q*Rw/(Q0*Rw0). If the main type of the first stream type is hot stream, the number of target data streams corresponding to the first stream type is equal to S*Q*Rh/(Q0*Rh0).
  • the management device includes the first mapping relationship shown in Table 2 above, the cold flow threshold and hot flow threshold based on the content distribution network, and the audience corresponding to each flow type in Table 2.
  • the user number range determines that the main type of flow type 1 and the main type of flow type 2 are cold flow, the main type of flow type 3 and the main type of flow type 4 are warm flow, the main type of flow type 5 and the main type of flow type 6 are The main type is heat flow.
  • Rc0, Rc, Q, Q0 and the number of standard data flows S1 corresponding to flow type 1 is obtained as S1*Q*Rc/(Q0*Rc0).
  • the number of target data streams corresponding to stream type 2 is S2*Q*Rc/(Q0*Rc0). Based on Rw0, Rw, Q, Q0 and the standard data flow number S3 corresponding to flow type 3, the number of target data flows corresponding to flow type 3 is obtained as S3*Q*Rw/(Q0*Rw0). In the same way, the corresponding number of flow type 4 is obtained The number of target data streams is S4*Q*Rw/(Q0*Rw0). Based on the number of standard data streams S5 corresponding to Rh0, Q, Q0, Rh and stream type 5, the number of target data streams corresponding to stream type 5 is obtained as S5*Q*Rh/(Q0*Rh0). In the same way, the number of target data streams corresponding to stream type 5 is obtained. The number of target data streams is S6*Q*Rh/(Q0*Rh0).
  • the management device obtains the number of target edge nodes corresponding to each flow type in the first time period.
  • the number of target edge nodes corresponding to the first flow type is to obtain the average number of edge nodes for the data flow of the first flow type.
  • the first mapping defines the primary type for each stream type.
  • the management device determines whether the main type of the first flow type is cold flow, warm flow or hot flow based on the cold flow threshold and the hot flow threshold. If the main type of the first flow type changes from the cold flow defined by the first mapping relationship to a warm flow, or the warm flow defined by the first mapping relationship changes to a hot flow, then the number of standard edge nodes corresponding to the first flow type is reduced. If the data flow of the first flow type changes from the warm flow defined by the first mapping relationship to a cold flow, or the hot flow defined by the first mapping relationship changes to a warm flow, then the number of standard edge nodes corresponding to the first flow type is increased. If the main type of the first flow type is the same as the main type defined by the first mapping relationship, the number of standard edge nodes corresponding to the first flow type does not need to be changed.
  • the network information of the content distribution network in the first time period includes the total number of edge nodes of the content distribution network in the first time period, and the management device includes the total number of standard edge nodes.
  • the management device calculates the ratio between the total number of edge nodes and the total number of standard edge nodes, and converts the ratio Multiply the number of standard edge nodes corresponding to each flow type to obtain the number of target edge nodes corresponding to each flow type.
  • the management device includes a first coefficient greater than one and a second coefficient less than one.
  • the operation of the management device to increase the number of standard edge nodes corresponding to the first flow type is: calculating the product between the number of standard edge nodes corresponding to the first flow type and the first coefficient to obtain the first value, and converting the first mapping relationship The number of standard edge nodes corresponding to the saved first flow type is replaced with the first numerical value.
  • the operation of the management device to increase the number of standard edge nodes corresponding to the first flow type is: calculating the product between the number of standard edge nodes corresponding to the first flow type and the second coefficient to obtain the second value, and converting the first mapping relationship The number of standard edge nodes corresponding to the saved first flow type is replaced with the second value.
  • the main type of flow type 2 is the temperature defined by the first mapping relationship.
  • the flow becomes cold.
  • the management device replaces the number of standard edge nodes corresponding to flow type 2 in the above table 2 with K2*x to increase the number of standard edge nodes corresponding to flow type 2.
  • the main type of flow type 5 is changed from the warm flow defined by the first mapping relationship to the heat flow.
  • the number of standard edge nodes corresponding to flow type 5 in the above Table 2 is replaced with K5*y to reduce the number of standard edges corresponding to flow type 5. Number of nodes.
  • the ratio between the total number of edge nodes calculated by the management device and the total number of standard edge nodes is z.
  • the ratio z is multiplied by the number of standard edge nodes corresponding to each flow type in Table 3 to obtain the number of target edge nodes corresponding to each flow type.
  • the results obtained are shown in the following table 5 shown.
  • the ratio z is multiplied by the number of standard edge nodes corresponding to each flow type in Table 4 to obtain the number of target edge nodes corresponding to each flow type.
  • the results obtained are shown in the following table 6 shown.
  • the number of target data flows and the number of target edge nodes corresponding to each flow type, as well as the number of standard data flows and the number of standard edge nodes corresponding to each flow type, are obtained through the following first formula.
  • M is the total number of flow types
  • stream i is the number of target data flows corresponding to the i-th flow type
  • node i is the number of target edge nodes corresponding to the i-th flow type
  • Stream i is the i-th flow type.
  • the corresponding number of standard data flows Node i is the number of standard edge nodes corresponding to the i-th flow type
  • C is the change multiple of the return-to-origin rate
  • E 0 is the standard return-to-origin rate defined by the first mapping relationship
  • E is the first time The return rate corresponding to the live broadcast service in the segment.
  • Step 403 The management device obtains the return-to-source traffic based on the return-to-source rate and the outbound traffic.
  • step 403 the management device multiplies the return-to-origin rate and the outflow traffic to obtain the return-to-source traffic.
  • Step 404 The management device allocates network resources in the content distribution network based on the return-to-source traffic and the outbound traffic.
  • the management device allocates network resources between the central node of the content distribution network and the edge nodes of the content distribution network based on the return-to-source traffic.
  • the network resources include bandwidth resources, etc., so that the content distribution network reserves the allocated Network resources, which are used to transmit data streams that edge nodes request to send to the central node.
  • the management device allocates network resources of edge nodes of the content distribution network based on the outflow traffic.
  • the network resources include bandwidth resources, etc., so that the content distribution network reserves the allocated network resources.
  • the network resources are used to transmit live broadcast services from the edge nodes to The viewer user sends the data stream requested by the viewer user.
  • the management device may also obtain the fee required by the content distribution network to transmit the live broadcast service within the first time period.
  • the management device includes a first price corresponding to outbound traffic and a second price corresponding to return-to-source traffic.
  • the management device obtains the fee required for the content distribution network to transmit the live broadcast service based on the outflow traffic, return flow, first price, and second price.
  • At least one basic feature when obtaining the return rate corresponding to the live broadcast service, at least one basic feature is obtained, and the at least one basic feature includes the return traffic and the first ratio.
  • the at least one basic feature and the return rate are combined into a training sample.
  • the at least one basic characteristic further includes one or more of the following: the total number of edge nodes or the total number of data flows.
  • At least one derived feature is obtained based on the at least one basic feature, and the at least one basic feature, the at least one derived feature and the return-to-origin rate are combined into a training sample.
  • the detailed implementation process of obtaining the at least one derived feature will be described in detail in subsequent embodiments and will not be introduced here.
  • the management device receives the business demand information of the live broadcast service and the data stream attribute information of the live broadcast service within the first time period.
  • the management device obtains the return-to-origin rate corresponding to the live broadcast service in the first time period based on the first ratio, and obtains the return-to-origin traffic generated by the live broadcast service in the first time period based on the return-to-origin rate and the outflow traffic included in the business demand information.
  • Network resources are allocated in the content distribution network based on the return-to-source traffic and outbound traffic.
  • the return-to-source traffic and outbound traffic generated by the live broadcast service in the first time period are obtained based on the business attribute information and business demand information of the live broadcast service. In this way, the flow of the live broadcast service is referenced when obtaining the traffic.
  • Data flow attributes The data flow attributes of live broadcast services in different time periods may be different. Different data flow attributes have different impacts on return-to-source traffic. When allocating network resources based on return-to-source traffic and outbound traffic in different time periods, allocation can be improved. Flexibility of network resources. In addition, the fees required for the content distribution network to transmit live broadcast services are obtained based on the return-to-source traffic and outbound traffic in different time periods, thereby improving billing flexibility.
  • the management device obtains the data based on the first ratio.
  • the return rate corresponding to the live broadcast service in the first time period thus improving the accuracy of obtaining the return rate.
  • an embodiment of the present application provides a method 600 for allocating network resources.
  • the method 600 is applied to the network architecture 200 shown in Figure 2.
  • the method 600 is executed by the management device in the network architecture 200.
  • the method 600 adopts a black box allocation method, which includes the following processes of steps 601-604.
  • Step 601 This is the same as step 401 of the method 400 shown in Figure 4 and will not be described in detail here.
  • Step 602 When the business demand information historically received by the management device includes the business demand information of the live broadcast service and the data flow attribute information historically received by the management device includes the data flow attribute information of the live broadcast service, the management device based on the return flow and the first ratio and the return rate prediction model to obtain the return rate corresponding to the live broadcast business in the first time period.
  • the decision-making component included in the management device receives the business demand information and data flow attribute information of the live broadcast service.
  • the business demand information historically received by the management device includes the business demand information of the live broadcast service and the data flow attribute information historically received by the management device includes
  • the black box allocation component of the trigger management device obtains the return to origin rate corresponding to the live broadcast service in the first time period based on the return flow, the first ratio and the return to origin rate prediction model.
  • the business requirement information historically received by the management device includes the business requirement information of the live broadcast service
  • the data stream attribute information historically received by the management device includes the data stream attribute information of the live broadcast service
  • the network information historically received by the management device includes content distribution.
  • the management device obtains the return rate corresponding to the live broadcast service in the first time period based on the return traffic, the first ratio and the return rate prediction model.
  • the management device when the business demand information also includes the total number of data flows, the management device obtains the live broadcast within the first time period based on the total number of data flows, return flow, first ratio, and return rate prediction model. The return rate corresponding to the business.
  • the management device when the management device receives the network information of the content distribution network, the network information includes the total number of edge nodes included in the content distribution network within the first time period, the management device may be based on the total number of edge nodes, the The total number of data streams, return traffic, first ratio and return rate prediction model are used to obtain the return rate corresponding to the live broadcast service in the first time period.
  • step 602 the management device obtains the return rate corresponding to the live broadcast service in the first time period through the following method 1 and method 2. Next, Method 1 and Method 2 will be explained in detail.
  • the management device obtains at least one basic feature, and the at least one basic feature includes the return flow rate and the first ratio.
  • the at least one basic feature is input into the return rate prediction model, so that the return rate prediction model obtains the return rate corresponding to the live broadcast service in the first time period based on the at least one basic feature. Get the return-to-origin rate output by the return-to-origin prediction model.
  • the at least one basic characteristic further includes one or more of the following: the total number of edge nodes or the total number of data flows.
  • Method 2 The management device obtains at least one basic characteristic, the at least one basic characteristic includes the return flow rate and the first ratio, and obtains at least one derived characteristic based on the at least one basic characteristic.
  • the at least one basic feature and the at least one derived feature are input into the return rate prediction model, so that the return rate prediction model obtains the return rate prediction model corresponding to the live broadcast service in the first time period based on the at least one basic feature and the at least one derived feature.
  • Return rate Get the return-to-origin rate output by the return-to-origin prediction model.
  • the at least one basic characteristic further includes one or more of the following: the total number of edge nodes or the total number of data flows.
  • the at least one derived characteristic includes one or more of the following: the total number of audience users of the live broadcast service in the first time period, the average number of audience users corresponding to the data streams sent by the anchor users of the live broadcast service, the data sent by the anchor users of the live broadcast service.
  • the average outflow traffic corresponding to the stream, the average number of data streams sent by the edge node in the first time period, the average number of viewers requesting data streams from the edge node in the first time period, or, the average number of data streams sent by the edge node in the first time period The average outbound traffic sent.
  • the total number of audience users is obtained based on the outflow traffic and the standard data volume of the data stream of the live broadcast service.
  • the standard data amount is a fixed value.
  • the total data amount of the viewer users is equal to the ratio between the outflow traffic and the standard data amount.
  • the average number of audience users is obtained based on the total number of audience users and the total number of data streams.
  • the average number of audience users is equal to the ratio between the total number of audience users and the total number of data streams.
  • the average outflow flow is obtained based on the outflow flow and the total number of data flows.
  • the average outflow flow rate is equal to the ratio between the outflow flow rate and the total number of data flows.
  • the average number of data flows is obtained based on the total number of data flows and the total number of edge nodes.
  • the average number of data flows is equal to the ratio between the total number of data flows and the total number of edge nodes.
  • the average number of audience users is obtained based on the total number of audience users and the total number of edge nodes.
  • the average number of audience users is equal to the ratio between the total number of audience users and the total number of edge nodes.
  • the average outflow traffic is obtained based on the outflow traffic and the total number of edge nodes.
  • the average outflow traffic is equal to the ratio between the outflow traffic and the total number of edge nodes.
  • the return-to-origin rate prediction model is trained in advance.
  • the training process is: based on at least one training sample, the return-to-origin rate prediction model is obtained.
  • Each training sample includes the return-to-origin traffic of the live broadcast service acquired historically, the live broadcast service acquired historically The ratio between cold flow, warm flow and hot flow, and the return to source rate based on the historical return flow and the ratio.
  • each training sample includes at least one basic feature and a return rate corresponding to the at least one basic feature, or each training sample includes at least one basic feature, at least one derived feature, and the at least one basic feature.
  • the return rate corresponding to the at least one derived feature is not limited to the at least one basic feature.
  • the return-to-origin rate prediction model is obtained through the following operations 6021-6023.
  • 6021 Based on at least one training sample and the return-to-origin rate prediction model to be trained, identify the return-to-origin rate corresponding to each training sample.
  • At least one basic feature included in the training sample is input into the return-to-origin rate prediction model to be trained, so that the return-to-origin rate prediction model to be trained obtains the return-to-origin rate based on the at least one basic feature, and the return to origin rate to be trained is obtained.
  • the return-to-origin rate output by the source prediction model is used as the return-to-origin rate corresponding to the training sample.
  • 6022 Based on the return rate included in each training sample and the return rate corresponding to each training sample, calculate the loss value through the loss function, and adjust the parameters of the return rate prediction model to be trained based on the loss value.
  • 6023 Determine whether to continue training the return-to-origin rate prediction model to be trained. If it is determined to continue training the return-to-origin rate prediction model to be trained, return to step 6021. If it is determined not to continue training the return-to-origin rate prediction model to be trained, predict the return-to-origin rate to be trained.
  • the model serves as a return rate prediction model.
  • each verification sample includes at least one basic feature and a return rate corresponding to the at least one basic feature, or each verification sample includes at least one basic feature, at least one derived feature, and The return rate corresponding to the at least one basic feature and the at least one derived feature.
  • Obtain the return rate corresponding to each verification sample based on the return rate prediction model to be trained.
  • the accuracy of obtaining the return rate is calculated. When the accuracy does not exceed the specified threshold, it is determined to continue training the return-to-origin rate prediction model to be trained. When the accuracy exceeds the specified threshold, it is determined not to continue training the return-to-origin rate prediction model to be trained.
  • the black box allocation component includes a feature conversion module, a model training module and a model inference module.
  • the black box allocation component inputs the historically received business requirement information and data flow attribute information, as well as the historically acquired return-to-origin rate, into the feature conversion module.
  • the feature conversion module obtains training samples based on business demand information and data flow attribute information, and the model training module trains the return rate prediction module based on the training samples.
  • the model inference module obtains the return rate of the live broadcast service in the first time period based on the business demand information, data flow attribute information and return rate prediction model in the first time period.
  • Step 603 The management device obtains the return-to-source traffic based on the return-to-source rate and the outbound traffic.
  • step 603 the management device multiplies the return-to-origin rate and the outflow traffic to obtain the return-to-source traffic.
  • Step 604 The management device allocates network resources in the content distribution network based on the return-to-source traffic and the outbound traffic.
  • step 404 of the method 400 shown in Figure 4 For the detailed implementation process of the management device allocating network resources, please refer to the relevant content in step 404 of the method 400 shown in Figure 4, which will not be described in detail here.
  • the management device may also obtain the fee required by the content distribution network to transmit the live broadcast service within the first time period.
  • the management device includes a first price corresponding to outbound traffic and a second price corresponding to return-to-source traffic.
  • the management device obtains the fee required for the content distribution network to transmit the live broadcast service based on the outflow traffic, return flow, first price, and second price.
  • the management device receives the business demand information of the live broadcast service and the data stream attribute information of the live broadcast service within the first time period. Based on the return traffic, the first ratio and the return rate prediction model, the management device obtains the return rate corresponding to the live broadcast service in the first time period, and obtains the outflow traffic included in the first time period based on the return rate and the business demand information.
  • the return traffic generated by the live broadcast service during the time period.
  • Network resources are allocated in the content distribution network based on the return-to-source traffic and outbound traffic.
  • the return-to-source traffic and outbound traffic generated by the live broadcast service in the first time period are obtained based on the business attribute information and business demand information of the live broadcast service. In this way, the flow of the live broadcast service is referenced when obtaining the traffic.
  • Data flow attributes The data flow attributes of live broadcast services in different time periods may be different, thereby improving the flexibility of allocating network resources.
  • the management device when the business demand information historically received by the management device includes the business demand information of the live broadcast service and the data stream attribute information historically received by the management device does not include the data stream attribute information of the live broadcast service, the management device based on the return flow and the first ratio and the return rate prediction model to obtain the return rate corresponding to the live broadcast service in the first time period. This improves the accuracy of obtaining the return rate and at the same time predicts the return rate. Use the test model to obtain the return-to-origin rate, improve the efficiency of obtaining the return-to-origin rate, and thereby improve the efficiency of allocating network resources.
  • an embodiment of the present application provides a device 900 for allocating network resources.
  • the device 900 is deployed on the management device of the network architecture 200 shown in Figure 2, the management device of the method 400 shown in Figure 4 or the management device shown in Figure 6.
  • Method 600 is shown on the management device.
  • the device 900 includes:
  • the information acquisition module 901 is used to obtain the business demand information of the live broadcast service and the data flow attribute information of the live broadcast service in the first time period.
  • the business demand information is used to describe the data flow of the live broadcast service transmitted by the content distribution network in the first time period.
  • the content distribution network includes a central node and an edge node.
  • the central node is used to send the data stream from the anchor user of the live broadcast service to the edge node. The viewer user sends this data stream;
  • the traffic acquisition module 902 is used to obtain the back-to-source traffic and out-flow traffic generated by the live broadcast service in the first time period based on the business demand information and the data flow attribute information.
  • the back-to-source traffic is generated by the data flow sent by the central node.
  • Traffic, the outflow traffic is the traffic generated by the data flow sent by the edge node;
  • the resource allocation module 903 is used to allocate network resources in the content distribution network based on the return-to-origin traffic and the outbound traffic.
  • step 401 of the method 400 shown in Figure 4 and step 601 of the method 600 shown in Figure 6, which will not be repeated here.
  • step 401 of the method 400 shown in Figure 4 and step 601 of the method 600 shown in Figure 6, which will not be repeated here.
  • the traffic acquisition module 902 obtains the back-to-source traffic and outbound traffic generated by the live broadcast service in the first time period.
  • steps 402-403 of the method 400 shown in Figure 4 and the steps of the method 600 shown in Figure 6 The relevant content in steps 602-603 will not be described in detail here.
  • step 404 of the method 400 shown in Figure 4 and step 604 of the method 600 shown in Figure 6, which will not be described in detail here.
  • the business demand information includes the outflow traffic
  • the data flow attribute information includes the first ratio between cold flow, warm flow and hot flow of the live broadcast service
  • Traffic acquisition module 902 used for:
  • the return rate is the ratio between the return traffic and the outbound traffic
  • the return-to-source traffic is obtained.
  • the traffic acquisition module 902 obtains the return-to-origin rate corresponding to the live broadcast service in the first time period based on the first ratio.
  • a cold flow is a data flow in which the number of requested audience users does not exceed the cold flow threshold.
  • a warm flow is a data flow in which the number of requested audience users exceeds the cold flow threshold and does not exceed the hot flow threshold.
  • a hot flow is a data flow in which the number of requested audience users exceeds the cold flow threshold. Data flow with hot flow threshold, the hot flow threshold is greater than the cold flow threshold.
  • the traffic acquisition module 902 obtains the back-to-source traffic based on the return-to-source rate and the outflow traffic.
  • the business demand information also includes the total number of data streams sent to the central node by anchor users who need live broadcast services in the first time period,
  • the traffic acquisition module 902 is configured to obtain the return rate corresponding to the live broadcast service in the first time period based on the total number of data streams and the first ratio.
  • the traffic acquisition module 902 is used for:
  • the number of edge nodes corresponding to each flow type in the first time period is obtained, and the number of audience users requesting the data flow of each flow type is different.
  • the return rate corresponding to the live broadcast service in the first time period is obtained.
  • the detailed implementation process of the traffic acquisition module 902 obtaining the number of edge nodes corresponding to each flow type in the first time period can be found in the relevant content in operation 4022 of the method 400 shown in Figure 4, which will not be described in detail here.
  • the detailed implementation process of the traffic acquisition module 902 obtaining the number of data flows corresponding to each flow type in the first time period can be found in the relevant content in operation 4021 of the method 400 shown in Figure 4, which will not be described in detail here.
  • the information acquisition module 901 is also used to acquire network information of the content distribution network.
  • the network information includes one or more of the following: the total number of edge nodes included in the content distribution network during the first time period, at the first time The cold flow threshold of the content distribution network in the segment, or the hot flow threshold of the content distribution network in the first time period;
  • the traffic acquisition module 902 is configured to obtain the number of edge nodes corresponding to each flow type in the first time period based on the network information and the standard number of edge nodes corresponding to each flow type.
  • the information acquisition module 901 obtaining network information of the content distribution network, please refer to the relevant content in step 401 of the method 400 shown in Figure 4 and step 601 of the method 600 shown in Figure 6, which will not be described in detail here. .
  • the traffic acquisition module 902 obtains the number of edge nodes corresponding to each flow type in the first time period based on the network information and the standard number of edge nodes corresponding to each flow type.
  • the relevant content in step 402 of 400 will not be described in detail here.
  • the traffic acquisition module 902 is configured to, when the historically acquired business requirement information does not include the business requirement information of the live broadcast service and/or the historically acquired data stream attribute information does not include the data stream attribute information of the live broadcast service, based on the third One ratio obtains the return rate corresponding to the live broadcast service in the first time period.
  • the business demand information includes the outflow traffic
  • the data flow attribute information includes a first ratio between cold flow, warm flow and hot flow included in the live broadcast service
  • Traffic acquisition module 902 used for:
  • the first ratio and the return rate prediction model Based on the return traffic, the first ratio and the return rate prediction model, obtain the return rate corresponding to the live broadcast business in the first time period.
  • the return rate is the ratio between the return traffic and the outbound traffic;
  • the return-to-source traffic is obtained.
  • the traffic acquisition module 902 obtains the return rate corresponding to the live broadcast service in the first time period based on the return flow, the first ratio and the return rate prediction model.
  • the traffic acquisition module 902 obtains the return rate corresponding to the live broadcast service in the first time period based on the return flow, the first ratio and the return rate prediction model.
  • the traffic acquisition module 902 obtains the back-to-source traffic based on the return-to-source rate and the outflow traffic.
  • the business demand information also includes the total number of data streams sent to the central node by anchor users who need live broadcast services in the first time period,
  • the traffic acquisition module 902 is configured to obtain the return rate corresponding to the live broadcast service in the first time period based on the total number of data streams, return traffic, first ratio, and return rate prediction model.
  • the traffic acquisition module 902 is used for:
  • the total number of edge nodes Based on the total number of edge nodes, the total number of data streams, return traffic, the first ratio and the return rate prediction model, obtain the return rate corresponding to the live broadcast service in the first time period.
  • the traffic acquisition module 902 is used for:
  • the total number of data flows and/or the total number of edge nodes obtain at least one derived characteristic corresponding to the live broadcast service
  • the total number of edge nodes, the total number of data streams, return traffic, the first ratio and the return rate prediction model obtain the return rate corresponding to the live broadcast service in the first time period.
  • the detailed implementation process for the traffic acquisition module 902 to obtain at least one derived feature corresponding to the live broadcast service refers to the relevant content in step 602 of the method 600 shown in Figure 6, which will not be described in detail here.
  • the traffic acquisition module 902 obtains the return traffic corresponding to the live broadcast service in the first time period based on at least one derived feature, the total number of edge nodes, the total number of data flows, return traffic, the first ratio and the return rate prediction model.
  • the source rate please refer to the relevant content in step 603 of the method 600 shown in Figure 6, which will not be described in detail here.
  • At least one derived characteristic includes one or more of the following: the total number of audience users of the live broadcast service in the first time period, the average number of audience users corresponding to the data streams sent by the anchor users of the live broadcast service, the anchor users of the live broadcast service The average outflow traffic corresponding to the data stream sent, the average number of data streams sent by the edge node in the first time period, the average number of audience users requesting data streams from the edge node in the first time period, or, in the first time period Average outflow traffic sent by inner edge nodes;
  • Traffic acquisition module 902 used for:
  • the average outflow traffic is obtained.
  • the traffic acquisition module 902 is configured to, when the historically acquired business demand information includes the demand information of the live broadcast service and the historically acquired data flow attribute information includes the data flow attribute information of the live broadcast service, based on the return flow and the first ratio and the return rate prediction model to obtain the return rate corresponding to the live broadcast business in the first time period.
  • the device 900 also includes:
  • the model acquisition module 904 is used to obtain a return-to-origin rate prediction model based on at least one training sample.
  • Each training sample includes the historically acquired return-to-origin traffic, the historically acquired first ratio, and the historically acquired return-to-origin traffic and the third ratio. The return rate obtained by one ratio.
  • the model acquisition module 904 acquires the return-to-origin rate prediction model based on at least one training sample.
  • the device 900 also includes:
  • the fee acquisition module 905 is used to obtain the fee required for the content distribution network to transmit the live broadcast service based on the return-to-source traffic and the outflow traffic.
  • the information acquisition module obtains the business demand information of the live broadcast service and the data flow attribute information of the live broadcast service in the first time period.
  • the business demand information is used to describe the transmission of the live broadcast service by the content distribution network in the first time period.
  • the first time period is after the current time.
  • the traffic acquisition module acquires the return-to-source traffic and outbound traffic generated by the live broadcast service in the first time period based on the business demand information and data flow attribute information.
  • the resource allocation module allocates network resources in the content distribution network based on the return-to-origin traffic and the outbound traffic.
  • the traffic acquisition module acquires the return-to-origin traffic and outflow traffic generated by the live broadcast service in the first time period based on the business attribute information and business demand information of the live broadcast service.
  • the data flow attributes of the live broadcast service are referred to.
  • the data flow attributes of the live broadcast service may be different in different time periods.
  • the return-to-source traffic obtained based on the data flow attributes in different time periods is also different, thereby improving the flexibility of allocating network resources.
  • an embodiment of the present application provides a schematic diagram of a computer device 1000.
  • the computer device 1000 may be the management device in any of the above embodiments.
  • the computer device 1000 may be a management device in the network architecture 200 shown in FIG. 2 , or a management device in the method 400 shown in FIG. 4 or the method 600 shown in FIG. 6 .
  • the computer device 1000 includes at least one processor 1001, internal connections 1002, memory 1003 and at least one transceiver 1004.
  • the computer device 1000 is a device with a hardware structure and can be used to implement the functional modules in the device 900 described in Figure 9 .
  • the information acquisition module 901, traffic acquisition module 902, resource allocation module 903, model acquisition module 904 and cost acquisition module 905 in the device 900 shown in Figure 9 can be used through the at least one processor 1001 This is accomplished by calling the code in memory 1003.
  • the computer device 1000 can also be used to implement the functions of the management device in any of the above embodiments.
  • the above-mentioned processor 1001 can be a general central processing unit (CPU), a network processor (network processor, NP), a microprocessor, or an application-specific integrated circuit (ASIC). , or one or more integrated circuits used to control the execution of the program of this application.
  • CPU central processing unit
  • NP network processor
  • ASIC application-specific integrated circuit
  • the internal connection 1002 may include a path for transmitting information between the components.
  • the internal connection 1002 is a single board or a bus, etc.
  • the above-mentioned transceiver 1004 is used to communicate with other devices or communication networks.
  • the above-mentioned memory 1003 can be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (random access memory, RAM) or other types that can store information and instructions.
  • type of dynamic storage device which can also be electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disk storage, optical discs Storage (including compressed optical discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or can be used to carry or store the desired program code in the form of instructions or data structures and can be used by Any other media accessible by a computer, but not limited to this.
  • the memory can exist independently and be connected to the processor through a bus. Memory can also be integrated with the processor.
  • the memory 1003 is used to store the application code for executing the solution of the present application, and the processor 1001 controls the execution.
  • the processor 1001 is used to execute the application code stored in the memory 1003, and cooperate with at least one transceiver 1004, so that the computer device 1000 implements the functions in the patent method.
  • the processor 1001 may include one or more CPUs, such as CPU0 and CPU1 in FIG. 10 .
  • the computer device 1000 may include multiple processors, such as the processor 1001 and the processor 1007 in FIG. 10 .
  • processors may be a single-CPU processor or a multi-CPU processor.
  • a processor here may refer to one or more devices, circuits, and/or processing cores for processing data (eg, computer program instructions).

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Databases & Information Systems (AREA)
  • Human Computer Interaction (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)
  • Information Transfer Between Computers (AREA)

Abstract

本申请公开了一种分配网络资源的方法、装置及存储介质,属于通信领域。所述方法包括:获取在第一时间段内直播业务的业务需求信息和所述直播业务的数据流属性信息,所述业务需求信息用于描述在所述第一时间段内内容分发网络传输所述直播业务的数据流的情况,所述第一时间段位于当前时间之后;基于所述业务需求信息和所述数据流属性信息,获取在所述第一时间段内所述直播业务产生的回源流量和出流流量;基于所述回源流量和所述出流流量在所述内容分发网络中分配网络资源。本申请能够提高分配网络资源的灵活性。

Description

分配网络资源的方法、装置及存储介质
本申请要求于2022年9月6日提交的申请号为202211085847.5、发明名称为“分配网络资源的方法、装置及存储介质”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
技术领域
本申请涉及通信领域,特别涉及一种分配网络资源的方法、装置及存储介质。
背景技术
在视频直播场景中,直播业务的运营商使用云厂商搭建的内容分发网络来传输直播业务。直播业务的主播用户向内容分发网络发送数据流,直播业务的观众用户从内容分发网络拉取该数据流,如此实现使用内容分发网络来传输直播业务。
直播业务的运营商可以向云厂商申请在一段时间内使用内容分发网络来传输直播业务,云厂商在内容分发网络中需要对申请的直播业务分配网络资源,在该段时间内内容分发网络使用分配的网络资源来传输该直播业务。目前分配网络资源的方式固化,不够灵活,导致分配网络资源的灵活性低。
发明内容
本申请提供了一种分配网络资源的方法、装置及存储介质,以提高分配网络资源的灵活性。所述技术方案如下:
第一方面,本申请提供了一种分配网络资源的方法,在所述方法中,管理设备获取在第一时间段内直播业务的业务需求信息和直播业务的数据流属性信息,业务需求信息用于描述在第一时间段内内容分发网络传输直播业务的数据流的情况,第一时间段位于当前时间之后,内容分发网络包括中心节点和边缘节点,中心节点用于向边缘节点发送来自直播业务的主播用户的数据流,边缘节点用于向请求该数据流的观众用户发送该数据流。管理设备基于业务需求信息和数据流属性信息,获取在第一时间段内直播业务产生的回源流量和出流流量,回源流量是中心节点发送的数据流产生的流量,出流流量是边缘节点发送的数据流产生的流量。管理设备基于该回源流量和该出流流量在内容分发网络中分配网络资源。
由于获取直播业务的数据流属性信息,基于直播业务的业务属性信息和业务需求信息获取在第一时间段内直播业务产生的流量,这样在获取流量时参考了直播业务的数据流属性,不同时间段直播业务的数据流属性可能不同,基于不同数据流属性信息得到的回源流量不同,基于得到的回源流量和出流流量分配网络资源,从而提高了分配网络资源的灵活性。
在一种可能的实现方式中,基于该回源流量和出流流量,获取内容分发网络传输直播业务所需要的费用。基于不同数据流属性信息得到的回源流量不同,基于得到的回源流量和出流流量获取费用,可以提高计费的灵活性。
在另一种可能的实现方式中,业务需求信息包括该出流流量,该数据流属性信息包括直播业务的冷流、温流和热流之间的第一配比。基于第一配比获取在第一时间段内直播业务对应的回源率,该回源率是回源流量与出流流量之间的比例。基于回源率和出流流量,获取回源流量。
由于直播业务的数据流属性信息会影响直播业务的回源率,基于第一配比能够准确地获取到直播业务对应的回源率,基于该回源率能够提高获取回源流量的准确性。
在另一种可能的实现方式中,冷流是请求的观众用户数量不超过冷流阈值的数据流,温流是请求的观众用户数量超过冷流阈值且未超过热流阈值的数据流,热流是请求的观众用户数量超过热流阈值的数据流,热流阈值大于冷流阈值。
在另一种可能的实现方式中,业务需求信息还包括在第一时间段内需要直播业务的主播用户向中心节点发送的数据流总数量。基于该数据流总数量和第一配比获取在第一时间段内直播业务对应的回源率。由于直播业务的主播用户向中心节点发送的数据流总数量也会影响直播业务的回源率,基于该数据流总数量和第一配比,提高获取回源率的准确性。
在另一种可能的实现方式中,基于多个流类型中的每个流类型对应的标准边缘节点数,获取在第一时间段内每个流类型对应的边缘节点数,请求所述每个流类型的数据流的观众用户数量不同。基于每个流类型对应的标准数据流数量和第一配比,获取在第一时间段内每个流类型对应的数据流数量。基于每个流类型对应的数据流数量和边缘节点数,获取在第一时间段内直播业务对应的回源率。如此实现基于第一配比获取在第一时间段内直播业务对应的回源率。
在另一种可能的实现方式中,基于内容分发网络的网络信息和每个流类型对应的标准边缘节点数,获取在第一时间段内每个流类型对应的边缘节点数。基于网络信息可以提高获取边缘节点数的精度,进而提高获取回源率的精度。
在另一种可能的实现方式中,在历史获取的业务需求信息中不包括直播业务的业务需求信息和/或历史获取的数据流属性信息不包括直播业务的数据流属性信息时,基于第一配比获取在第一时间段内所述直播业务对应的回源率。
在另一种可能的实现方式中,业务需求信息包括该出流流量,数据流属性信息包括直播业务包括的冷流、温流和热流之间的第一配比。基于回流流量、第一配比和回源率预测模型,获取在第一时间段内直播业务对应的回源率,回源率是回源流量与出流流量之间的比例。基于回源率和出流流量,获取回源流量。由于直播业务的数据流属性信息会影响直播业务的回源率,基于回流流量、第一配比和回源率预测模型能够准确地获取到直播业务对应的回源率,另外回源率预测模型获取回源率的效率高,也提高了分配网络资源的效率。
在另一种可能的实现方式中,业务需求信息还包括在第一时间段内需要直播业务的主播用户向中心节点发送的数据流总数量。基于数据流总数量、回流流量、第一配比和回源率预测模型,获取在第一时间段内直播业务对应的回源率。由于直播业务的主播用户向中心节点发送的数据流总数量也会影响直播业务的回源率,引入该数据流总数量,可以提高获取回源率的准确性。
在另一种可能的实现方式中,获取在第一时间段内内容分发网络包括的边缘节点总数量。基于边缘节点总数量、数据流总数量、回流流量、第一配比和回源率预测模型,获取在第一时间段内所述直播业务对应的回源率。其中,内容分发网络包括的边缘节点总数量会影响直播业务的回源率,引入该边缘节点总数量,可以提高获取回源率的精度。
在另一种可能的实现方式中,基于所述出流流量、数据流总数量和/或边缘节点总数量,获取直播业务对应的至少一个衍生特征。基于至少一个衍生特征、边缘节点总数量、数据流总数量、回流流量、第一配比和回源率预测模型,获取在第一时间段内直播业务对应的回源率。其中,引入至少一个衍生特征,可以进一步提高获取回源率的精度。
在另一种可能的实现方式中,至少一个衍生特征包括如下一个或多个,在第一时间段内所述直播业务的观众用户总数量,直播业务的主播用户发送的数据流对应的平均观众用户数量,直播业务的主播用户发送的数据流对应的平均出流流量,在第一时间段内边缘节点发送的平均数据流数量,在第一时间段内向边缘节点请求数据流的平均观众用户数量,或者,在第一时间段内边缘节点发送的平均出流流量。
基于出流流量和直播业务的数据流的标准数据量,获取观众用户总数量。和/或,基于观众用户总数量 和数据流总数量,获取平均观众用户数量。和/或,基于出流流量和数据流总数量,获取平均出流流量。和/或,基于数据流总数量和边缘节点总数量,获取平均数据流数量。和/或,基于观众用户总数量和边缘节点总数量,获取平均观众用户数量。和/或,基于出流流量和边缘节点总数量,获取平均出流流量。从而实现了获取该至少一个衍生特征。
在另一种可能的实现方式中,在历史获取的业务需求信息包括直播业务的需求信息以及历史获取的数据流属性信息包括直播业务的数据流属性信息时,基于回流流量、第一配比和回源率预测模型,获取在第一时间段内直播业务对应的回源率,以提高获取回源率的精度。
在另一种可能的实现方式中,基于至少一个训练样本,获取回源率预测模型,由于每个训练样本包括历史获取的回源流量、历史获取的第一配比,以及基于历史获取的回源流量和第一配比得到的回源率,从而能够提高回源率预测模型的精度。
第二方面,本申请提供了一种分配网络资源的装置,用于执行第一方面或第一方面的任意一种可能的实现方式中的方法。具体地,所述装置包括用于执行第一方面或第一方面的任意一种可能的实现方式中的方法的单元。
第三方面,本申请提供了一种计算机设备,包括至少一个处理器和存储器,所述至少一个处理器用于与存储器耦合,读取并执行所述存储器中的指令,以实现第一方面或第一方面的任意一种可能的实现方式中的方法。
第四方面,本申请提供了一种计算机程序产品,所述计算机程序产品包括在计算机可读存储介质中存储的计算机程序,并且所述计算程序通过处理器进行加载来实现上述第一方面或第一方面任意可能的实现方式的方法。
第五方面,本申请提供了一种计算机可读存储介质,用于存储计算机程序,所述计算机程序通过处理器进行加载来执行上述第一方面或第一方面任意可能的实现方式的方法。
第六方面,本申请提供了一种芯片,包括存储器和处理器,存储器用于存储计算机指令,处理器用于从存储器中调用并运行该计算机指令,以执行上述第一方面或第一方面任意可能的实现方式的方法。
附图说明
图1是本申请实施例提供的一种内容分发网络的结构示意图;
图2是本申请实施例提供的一种网络架构的结构示意图;
图3是本申请实施例提供的另一种内容分发网络的结构示意图;
图4是本申请实施例提供的一种分配网络资源的方法流程图;
图5是本申请实施例提供的另一种分配网络资源的方法流程图;
图6是本申请实施例提供的另一种分配网络资源的方法流程图;
图7是本申请实施例提供的另一种分配网络资源的方法流程图;
图8是本申请实施例提供的一种黑盒分配部件的结构示意图;
图9是本申请实施例提供的一种分配网络资源的装置结构示意图;
图10是本申请实施例提供的一种计算机设备结构示意图。
具体实施方式
下面将结合附图对本申请实施方式作进一步地详细描述。
在直播业务中,直播平台使用云厂商提供的基础云服务,主要是使用云厂商搭建的内容分发网络来支撑直播业务。云厂商会在全球各地搭建内容分发网络的节点,内容分发网络的节点包括源站节点,中心节点和边缘节点等。内容分发网络是端到端全链路网络,直播业务的主播用户从内容分发网络的边缘节点接入,并向内容分发网络发送直播业务的数据流。直播业务的观众用户也会从内容分发网络的边缘节点接入,从内容分发网络中获取直播业务的数据流进行观看。在使用内容分发网络传输直播业务的数据流前,直播平台根据云厂商预算的直播业务产生的流量付费给云厂商。
参见图1所示的内容分发网络100,内容分发网络100包括源站节点101、中心节点102和边缘节点103。源站节点101与至少一个中心节点102通信,对于每个中心节点102,该中心节点102与至少一个边缘节点103通信。
内容分发网络100中的边缘节点103分布在各地,直播业务的主播用户接入内容分发网络100的边缘节点103,主播用户向边缘节点103发送直播业务的数据流。
该边缘节点103用于接收该数据流,向与该边缘节点103通信的中心节点102发送该数据流。
该中心节点102用于接收该数据流,向源站节点101发送该数据流。
源站节点101用于接收该数据流,将该数据流分发到内容分发网络100中除该中心节点102之外的各中心节点102上。内容分发网络100中的各中心节点102可以得到该数据流,并缓存该数据流。
直播业务的观众用户接入内容分发网络100的边缘节点103,且观众用户向该边缘节点103请求直播业务的数据流。
如果该边缘节点103缓存有该数据流,该边缘节点103用于向该观众用户发送缓存的该数据流。如果该边缘节点103没有缓存该数据流,该边缘节点103用于向与该边缘节点103通信的中心节点102请求该数据流,接收该中心节点102发送的该数据流,向该观众用户发送该数据流以及缓存该数据流。
在内容分发网络100中,直播业务的数据流分为冷流、温流或热流。冷流是请求的观众用户数量不超过冷流阈值的数据流,温流是请求的观众用户数量超过冷流阈值且未超过热流阈值的数据流,热流是请求的观众用户数量超过热流阈值的数据流,热流阈值大于冷流阈值。
该热流阈值和/或该冷流阈值是内容分发网络100定义的阈值。在不同时间段内容分发网络100定义的热流阈值和/或冷流阈值可能不同,也就是说,热流阈值和/或冷流阈值不是固定不变的。
在不同时间段内,内容分发网络100包括的边缘节点总数量可能不同,也就是说,在不同时间段内,可能增加内容分发网络100包括的边缘节点103,也可能减少内容分发网络100包括的边缘节点100。
直播业务的主播用户和观众用户可能是直播业务的两个功能不同的客户端。主播用户可以是主播客户端,观众用户可以是观众客户端。
参见图2,本申请实施例提供了一种网络架构200,该网络架构200包括直播平台201、调度平台202和管理设备203,管理设备203分别与直播平台201和调度平台202通信。
调度平台202和管理设备203属于上述内容分发网络100的运营商,调度平台202用于获取在第一时间段内容分发网络100的网络信息,该网络信息包括如一个或多个,在第一时间段内内容分发网络100包括的边缘节点总数量,在第一时间段内内容分发网络100的冷流阈值,或者,在第一时间段内内容分发网络100的热流阈值等信息。
第一时间段可能是在当前时间之后时间段。
直播平台201用于向云厂商申请在第一时间段内使用内容分发网络100来传输直播业务,在申请使用内容分发网络100来传输直播业务时,向管理设备203发送在第一时间段内直播业务的业务需求信息和直播业务的数据流属性信息。
管理设备203用于接收该业务需求信息和该数据流属性信息,基于该业务需求信息和该数据流属性信息,获取在第一时间段内直播业务产生的流量。基于直播业务产生的流量在内容分发网络中分配网络资源,以使内容分发网络在第一时间段内基于该网络资源传输直播业务的数据流。
管理设备203还可能基于该流量获取内容分发网络100传输直播业务所需要的费用。这样,直播业务的运营商基于管理设备200获取的费用向云厂商支付费用。
第一时间段可能是一周时间、一个月时间、两个月时间或半年时间等。例如,假设第一时间段为一个月时间,直播平台201可能在每个月的月末向云厂商申请在下一个月内使用内容分发网络100来传输直播 业务,第一时间段为下一个月的时间。
在一些实施例中,管理设备203可能用于从调度平台202中获取在第一时间段内内容分发网络100的网络信息,基于该网络信息、该业务需求信息和该数据流属性信息,获取在第一时间段内直播业务产生的流量。
在第一时间段内直播业务产生的流量包括在第一时间段内的直播业务产生的回源流量和出流流量。回源流量是内容分发网络100包括的中心节点向内容分发网络100包括的边缘节点发送的各数据流产生的流量,即回源流量等于该各数据流的数据量之和,出流流量是内容分发网络100包括的边缘节点向直播业务的观众用户发送的各数据流产生的流量,即出流流量等于该各数据流的数据量之和。回源流量与出流流量之间的比值称为回源率。
例如,参见图3,内容分发网络100包括中心节点A、中心节点B,与中心节点A通信的边缘节点1和边缘节点2,以及与中心节点B通信边缘节点3。观众用户1和观众用户2接入边缘节点1,并从边缘节点1中请求获取数据流1,数据流1是边缘节点1接收中心节点A发送的数据流。观众用户3和观众用户4接入边缘节点2,并从边缘节点2中请求获取数据流2,数据流2是边缘节点2接收中心节点A发送的数据流。观众用户5接入边缘节点3,并从边缘节点3中请求获取数据流3,数据流3是边缘节点3接收中心节点B发送的数据流。
假设数据流1的数据量、数据流2的数据量和数据流3的数据量均为2M,回源流量等于边缘节点1接收的数据流1的数据量,边缘节点2接收的数据流2的数据量和边缘节点3接收的数据流3的数据量之和,即回源流量等于6M。出流流量等于边缘节点1向观众用户1发送的数据流1的数据量、边缘节点1向观众用户2发送的数据流1的数据量、边缘节点2向观众用户3发送的数据流2的数据量、边缘节点2向观众用户4发送的数据流2的数据量和边缘节点3向观众用户5发送的数据流3的数据量之和,即出流流量等于10M。回源率等于回源流量与出流流量之间的比值,即回源率等于0.6。
管理设备203包括出流流量对应的第一价格以及回源流量对应的第二价格。管理设备203基于出流流量、回流流量、第一价格和第二价格,获取内容分发网络100传输直播业务所需要的费用。
管理设备203包括两种网络资源的分配方式,该两种分配方式包括白盒分配方式和黑盒分配方式。管理设备203在接收到直播业务的业务需求信息和数据流属性信息时,基于直播业务的业务需求信息和数据流属性信息选择一种分配方式,采用选择的分配方式获取内容分发网络100传输直播业务所产生的流量。
白盒分配方式是基于管理设备203包括的第一映射关系获取该流量的方式,该白盒分配方式的详细实现过程请参见后续图4所示的方法400,在此先不详细说明。
第一映射关系用于保存观众用户数量范围、标准数据流数据量与标准边缘节点数之间的映射关系。第一映射关系定义了多个流类型,每个流类型对应的观众用户数量不同。第一映射关系包括每个流类型对应的记录,多个流类型与多个观众用户数量范围一一对应,每个流类型对应的观众用户数量范围不同。该多个流类型包括第一流类型,请求第一流类型的数据流的观众用户数量位于第一流类型对应的观众用户数量范围内。第一流类型对应的记录包括第一流类型对应的观众用户数量范围、标准数据流数量和标准边缘节点数。
第一流类型对应的标准数据流数量用于指示内容分发网络的边缘节点从内容分发网络的中心节点获取的属于第一流类型的数据流数量。第一流类型对应的标准边缘节点数用于指示请求第一流类型的每条数据流的平均边缘节点数。第一流类型对应的记录包括的各数值均为基线值。
参见下表1所示的第一映射关系,流类型1对应的记录包括观众用户数量范围“[1,3)”,标准数据流数量“103246”和标准边缘节点数“2.1”。观众用户数量范围“[1,3)”表示请求流类型1的每个数据流的观众用户数量大于或等于1且小于3。标准数据流数量“103246”表示内容分发网络的边缘节点从内容分发网络的中心节点请求的属于流类型1的103246个数据流。标准边缘节点数“2.1”表示请求流类型1的每条数据流的平均边缘节点数为2.1。对于表1所示第一映射关系中的其他记录的含义不再一一列举说明。
表1
对于直播业务的数据流,基于冷流阈值和热流阈值,将直播业务的数据流分成冷流、温流和热流三个主类型。而第一映射关系基于请求的观众用户数量将直播业务的数据流分成多个流类型,该多个流类型是主类型的子类型。该多个流类型包括第一流类型,如果第一流类型对应的观众用户数量范围的上限值未超过冷流阈值,则第一流类型为冷流的子类型。如果第一流类型对应的观众用户数量范围的下限值超过冷流阈值且第一流类型对应的观众用户数量范围的上限值未超过热流阈值,则第一流类型为温流的子类型。如果第一流类型对应的观众用户数量范围的下限值超过热流阈值,则第一流类型为热流的子类型。
黑盒分配方式是基于管理设备203包括的回源率预测模型获取该流量的方式,回源率预测模块是管理设备203事先训练出来的智能模型。该黑盒分配方式的详细实现过程请参见后续图6所示的方法600,在此先不详细说明。
参见图2,管理设备203包括决策部件2031、白盒分配部件2032和黑盒分配部件2033。可选地,管理设备203可能是计算设备集群,决策部件2031、白盒分配部件2032和黑盒分配部件2033可能是不同的计算设备。或者,管理设备203可能是计算设备,决策部件2031、白盒分配部件2032和黑盒分配部件2033可能是计算设备的不同模块。
决策部件2031用于基于直播业务的业务需求信息和数据流属性信息选择一种分配方式,在选择白盒分配方式时触发白盒分配部件2032,在选择黑盒分配方式时触发黑盒分配部件2033。
白盒分配部件2032用于在决策部件2031的触发下,采用白盒分配方式获取内容分发网络传输直播业务所产生的流量。
黑盒分配部件2033用于在决策部件2031的触发下,采用黑盒分配方式获取内容分发网络传输直播业务所产生的流量。
参见图4,本申请实施例提供了一种分配网络资源的方法400,该方法400应用于图2所示的网络架构200,该方法400由该网络架构200中的管理设备来执行。该方法400采用采用的分配方式为白盒分配方式,包括如下步骤401-404的流程。
步骤401:管理设备接收在第一时间段直播业务的业务需求信息和直播业务的数据流属性信息,该业务需求信息包括在第一时间段内需要直播业务产生的出流流量,该数据流属性信息包括直播业务的多种数据流之间的第一配比,该多种数据流包括冷流、温流和热流中的至少两个,第一时间段位于当前时间之后。
该业务需求信息还可能包括其他信息,例如,该业务需求信息还可能包括在第一时间段内需要直播业务的主播用户向内容分发网络的中心节点发送的数据流总数量。所以该业务需求信息用于描述在第一时间段内内容分发网络传输直播业务的数据流的情况。
在步骤401中,管理设备接收直播平台发送申请请求,该申请请求包括在第一时间段直播业务的业务需求信息和直播业务的数据流属性信息,该申请请求用于请求管理设备分配在第一时间段内内容分发网络传输直播业务所需要的网络资源。
管理设备还可能从调度平台获取内容分发网络的网络信息,该网络信息包括如下一个或多个,在第一时间段内内容分发网络包括的边缘节点总数量,在第一时间段内内容分发网络的冷流阈值,或者,在第一时间段内内容分发网络的热流阈值等,该冷流阈值小于该热流阈值。
在一些实施例中,管理设备接收调度平台发送的内容分发网络的网络信息。
如果在第一时间段内内容分发网络的网络信息与在第二时间段内内容分发网络的网络信息相同,调度 平台可能不会向管理设备发送内容分发网络的网络信息,也可能继续会向管理设备发送内容分发网络的网络信息,第二时间段是第一时间段的前一个时间段。如果在第一时间段内内容分发网络的网络信息与在第二时间段内内容分发网络的网络信息不同,调度平台向管理设备发送内容分发网络的网络信息。
步骤402:在管理设备历史接收的业务需求信息不包括直播业务的业务需求信息和/或管理设备历史接收的数据流属性信息不包括直播业务的数据流属性信息时,管理设备基于第一配比获取在第一时间段内直播业务对应的回源率。
在步骤402中,在管理设备历史接收的业务需求信息不包括直播业务的业务需求信息、管理设备历史接收的数据流属性信息不包括直播业务的数据流属性信息、和/或、管理设备历史接收的网络信息不包括内容分发网络的网络信息时,管理设备基于第一配比获取在第一时间段内直播业务对应的回源率。
在步骤402中,在该业务需求信息还包括数据流总数量时,管理设备基于该数据流总数量和第一配比获取在第一时间段内直播业务对应的回源率。
在步骤402中,管理设备通过如下流程获取在第一时间段内直播业务对应的回源率,该流程包括如下4021-4023的操作。可选地,参见图5,管理设备包括的决策部件接收直播业务的业务需求信息和数据流属性信息,在管理设备历史接收的业务需求信息不包括直播业务的业务需求信息和/或管理设备历史接收的数据流属性信息不包括直播业务的数据流属性信息时,触发管理设备的白盒分配部件执行如下4021-4023的操作获取该回源率。
4021:管理设备基于多个流类型中的每个流类型对应的标准数据流数量、直播业务的标准配比以及第一配比,获取在第一时间段内每个流类型对应的目标数据流数量,该多个流类型包括第一流类型,第一流类型对应的目标数据流数量是内容分发网络的边缘节点从内容分发网络的中心节点请求发送的第一流类型的数据流数量,该标准配比是冷流、温流和热流中的至少两个之间的配比。
管理设备包括第一映射关系和该标准配比,第一映射关系包括每个流类型对应的标准数据流数量。直播业务的冷流、温流和热流之间的标准配比表示为Rc0:Rw0:Rh0;直播业务的冷流、温流和热流之间的第一配比表示为Rc:Rw:Rh。
在4021中,基于内容分发网络的冷流阈值和热流阈值,以及第一映射关系中的第一流类型对应的观众用户数量范围,确定第一流类型属于的主类型,即确定第一流类型的数据流是冷流、温流还是热流。如果第一流类型的主类型是冷流,则第一流类型对应的目标数据流数量等于S*Rc/Rc0,S为第一流类型对应的标准数据流数量。如果第一流类型的主类型是温流,则第一流类型对应的目标数据流数量等于S*Rw/Rw0。如果第一流类型的主类型是热流,则第一流类型对应的目标数据流数量等于S*Rh/Rh0。
接下来列举第一实例,在第一实例中管理设备包括如下表2所示的第一映射关系,基于内容分发网络的冷流阈值和热流阈值,以及表2中的每个流类型对应的观众用户数量范围,确定流类型1的主类型和流类型2的主类型为冷流,流类型3的主类型和流类型4的主类型为温流,流类型5的主类型和流类型6的主类型为热流。参见表3,基于Rc0、Rc和流类型1对应的标准数据流数量S1,获取流类型1对应的目标数据流数量为S1*Rc/Rc0,同理得到流类型2对应的目标数据流数量为S2*Rc/Rc0。基于Rw0、Rw和流类型3对应的标准数据流数量S3,获取流类型3对应的目标数据流数量为S3*Rw/Rw0,同理得到流类型4对应的目标数据流数量为S4*Rw/Rw0。基于Rh0、Rh和流类型5对应的标准数据流数量S5,获取流类型5对应的目标数据流数量为S5*Rh/Rh0,同理得到流类型6对应的目标数据流数量为S6*Rh/Rh0。
表2
表3
在步骤4021中,管理设备包括直播业务的标准数据流总数量。如果该业务需求信息还包括数据流总数量时,管理设备基于每个流类型对应的标准数据流数量、标准配比、第一配比、该数据流总数据和标准数据流总数量,获取在第一时间段内每个流类型对应的目标数据流数量。
例如,该多个流类型包括第一流类型,如果第一流类型的主类型是冷流,则第一流类型对应的目标数据流数量等于S*Q*Rc/(Q0*Rc0),S为第一流类型对应的标准数据流数量,Q为该数据流总数量,Q0为标准数据流总数量。如果第一流类型的主类型是温流,则第一流类型对应的目标数据流数量等于S*Q*Rw/(Q0*Rw0)。如果第一流类型的主类型是热流,则第一流类型对应的目标数据流数量等于S*Q*Rh/(Q0*Rh0)。
接下来列举第二实例,在第二实例中管理设备包括如上表2所示的第一映射关系,基于内容分发网络的冷流阈值和热流阈值,以及表2中的每个流类型对应的观众用户数量范围,确定流类型1的主类型和流类型2的主类型为冷流,流类型3的主类型和流类型4的主类型为温流,流类型5的主类型和流类型6的主类型为热流。参见表4,基于Rc0、Rc、Q、Q0和流类型1对应的标准数据流数量S1,获取流类型1对应的目标数据流数量为S1*Q*Rc/(Q0*Rc0),同理得到流类型2对应的目标数据流数量为S2*Q*Rc/(Q0*Rc0)。基于Rw0、Rw、Q、Q0和流类型3对应的标准数据流数量S3,获取流类型3对应的目标数据流数量为S3*Q*Rw/(Q0*Rw0),同理得到流类型4对应的目标数据流数量为S4*Q*Rw/(Q0*Rw0)。基于Rh0、Q、Q0、Rh和流类型5对应的标准数据流数量S5,获取流类型5对应的目标数据流数量为S5*Q*Rh/(Q0*Rh0),同理得到流类型6对应的目标数据流数量为S6*Q*Rh/(Q0*Rh0)。
表4
4022:管理设备获取在第一时间段内每个流类型对应的目标边缘节点数,第一流类型对应的目标边缘节点数是获取第一流类型的数据流的平均边缘节点数。
第一映射关系定义了每个流类型的主类型。在上述操作4021中,管理设备基于该冷流阈值和热流阈值,确定第一流类型的主类型是冷流、温流或者热流。如果第一流类型的主类型由第一映射关系定义的冷流变为温流,或者由第一映射关系定义的温流变为热流,则减少第一流类型对应的标准边缘节点数。如果第一流类型的数据流由第一映射关系定义的温流变为冷流,或者由第一映射关系定义的热流变为温流,则增加第一流类型对应的标准边缘节点数。如果第一流类型的主类型与第一映射关系定义的主类型相同,则不用变动第一流类型对应的标准边缘节点数。
第一时间段内内容分发网络的网络信息包括第一时间段内内容分发网络的边缘节点总数量,管理设备包括标准边缘节点总数量。管理设备计算该边缘节点总数量与该标准边缘节点总数之间的比值,将该比值 分别与每个流类型对应的标准边缘节点数相乘,得到每个流类型对应的目标边缘节点数。
在一些实施例中,管理设备包括大于1的第一系数以及小于1的第二系数。
在一些实施例中,管理设备增加第一流类型对应的标准边缘节点数的操作为:计算第一流类型对应的标准边缘节点数与第一系数之间乘积,得到第一数值,将第一映射关系保存的第一流类型对应的标准边缘节点数替换为第一数值。
在一些实施例中,管理设备增加第一流类型对应的标准边缘节点数的操作为:计算第一流类型对应的标准边缘节点数与第二系数之间乘积,得到第二数值,将第一映射关系保存的第一流类型对应的标准边缘节点数替换为第二数值。
例如,假设第一系数为x,x大于1,第二系数为y,y小于1,在上述列举的第一实例或第二实例中,流类型2的主类型由第一映射关系定义的温流变为冷流。参见上述表3或表4,管理设备将上述表2中的流类型2对应的标准边缘节点数替换为K2*x,以实现增加流类型2对应的标准边缘节点数。流类型5的主类型由第一映射关系定义的温流变为热流,将上述表2中的流类型5对应的标准边缘节点数替换为K5*y,以实现减少流类型5对应的标准边缘节点数。
假设管理设备计算边缘节点总数量与标准边缘节点总数之间的比值为z。在上述列举的第一实例中,将该比值z分别与表3中的每个流类型对应的标准边缘节点数相乘,得到每个流类型对应的目标边缘节点数,得到的结果见如下表5所示。在上述列举的第二实例中,将该比值z分别与表4中的每个流类型对应的标准边缘节点数相乘,得到每个流类型对应的目标边缘节点数,得到的结果见如下表6所示。
表5
表6
4023:基于标准回源率,每个流类型对应的目标数据流数量和目标边缘节点数,以及每个流类型对应的标准数据流数量和标准边缘节点数,获取在第一时间段内直播业务对应的回源率。
在4023中,基于标准回源率,每个流类型对应的目标数据流数量和目标边缘节点数,以及每个流类型对应的标准数据流数量和标准边缘节点数,通过如下第一公式获取在第一时间段内直播业务对应的回源率。
第一公式:
在第一公式中,M为流类型总数,streami为第i个流类型对应的目标数据流数量,nodei为第i个流类型对应的目标边缘节点数,Streami为第i个流类型对应的标准数据流数量,Nodei为第i个流类型对应的标准边缘节点数,C为回源率变化倍数,E0为第一映射关系定义的标准回源率,E为在第一时间段内直播业务对应的回源率。
步骤403:管理设备基于该回源率和该出流流量获取回源流量。
在步骤403中,管理设备将该回源率与该出流流量相乘,得到回源流量。
步骤404:管理设备基于该回源流量和该出流流量,在内容分发网络中分配网络资源。
在步骤404中,管理设备基于该回源流量分配内容分发网络的中心节点到内容分发网络的边缘节点之间的网络资源,该网络资源包括带宽资源等,这样内容分发网络会预留分配的该网络资源,该网络资源用于传输边缘节点向中心节点请求发送的数据流。
管理设备基于该出流流量分配内容分发网络的边缘节点的网络资源,该网络资源包括带宽资源等,这样内容分发网络会预留分配的该网络资源,该网络资源用于传输边缘节点向直播业务的观众用户发送的观众用户请求的数据流。
在步骤404中,管理设备还可能获取在第一时间段内内容分发网络传输直播业务所需要的费用。
管理设备包括出流流量对应的第一价格以及回源流量对应的第二价格。管理设备基于出流流量、回流流量、第一价格和第二价格,获取内容分发网络传输直播业务所需要的费用。
在一些实施例中,在获取到直播业务对应的回源率时,获取至少一个基础特征,该至少一个基础特征包括该回流流量和第一配比。将该至少一个基础特征和该回源率组成一个训练样本。可选地,该至少一个基础特征还包括如下一个或多个,该边缘节点总数量或者该数据流总数量。
在一些实施例中,基于该至少一个基础特征获取至少一个衍生特征,将该至少一个基础特征、该至少一个衍生特征和该回源率组成一个训练样本。可选地,获取该至少一个衍生特征的详细实现过程,将在后续实施例进行详细说明,在此先不介绍。
在本申请实施例中,管理设备接收在第一时间段内直播业务的业务需求信息和直播业务的数据流属性信息。管理设备基于第一配比获取在第一时间段内直播业务对应的回源率,基于回源率和业务需求信息包括的出流流量获取在第一时间段内直播业务产生的回源流量。基于该回源流量和出流流量在内容分发网络中分配网络资源。由于获取直播业务的数据流属性信息,基于直播业务的业务属性信息和业务需求信息获取在第一时间段内直播业务产生的回源流量和出流流量,这样在获取流量时参考了直播业务的数据流属性,不同时间段直播业务的数据流属性可能不同,不同的数据流属性对回源流量产生不同的影响,基于不同时间段的回源流量和出流流量分配网络资源时,能够提高分配网络资源的灵活性。此外,基于不同时间段的回源流量和出流流量获取内容分发网络传输直播业务所需要的费用,从而提高了计费灵活性。另外,在管理设备历史接收的业务需求信息不包括直播业务的业务需求信息和/或管理设备历史接收的数据流属性信息不包括直播业务的数据流属性信息时,管理设备基于第一配比获取在第一时间段内直播业务对应的回源率,这样提高获取回源率的精度。
参见图6,本申请实施例提供了一种分配网络资源的方法600,该方法600应用于图2所示的网络架构200,该方法600由该网络架构200中的管理设备来执行。该方法600采用采用的分配方式为黑盒分配方式,包括如下步骤601-604的流程。
步骤601:与图4所示方法400的步骤401相同,在此不再详细说明。
步骤602:在管理设备历史接收的业务需求信息包括直播业务的业务需求信息和管理设备历史接收的数据流属性信息包括直播业务的数据流属性信息时,管理设备基于该回流流量、第一配比和回源率预测模型,获取在第一时间段内直播业务对应的回源率。
参见图7,管理设备包括的决策部件接收直播业务的业务需求信息和数据流属性信息,在管理设备历史接收的业务需求信息包括直播业务的业务需求信息和管理设备历史接收的数据流属性信息包括直播业务的数据流属性信息时,触发管理设备的黑盒分配部件基于该回流流量、第一配比和回源率预测模型,获取在第一时间段内直播业务对应的回源率。
在步骤602中,在管理设备历史接收的业务需求信息包括直播业务的业务需求信息、管理设备历史接收的数据流属性信息包括直播业务的数据流属性信息以及管理设备历史接收的网络信息包括内容分发网络的网络信息时,管理设备基于该回流流量、第一配比和回源率预测模型,获取在第一时间段内直播业务对应的回源率。
在一些实施例中,在该业务需求信息还包括数据流总数量时,管理设备基于该数据流总数量、回流流量、第一配比和回源率预测模型,获取在第一时间段内直播业务对应的回源率。
在一些实施例中,在管理设备接收到内容分发网络的网络信息时,该网络信息包括在第一时间段内内容分发网络包括的边缘节点总数量,管理设备可能基于该边缘节点总数量、该数据流总数量、回流流量、第一配比和回源率预测模型,获取在第一时间段内直播业务对应的回源率。
在步骤602中,管理设备通过如下方式一和方式二获取在第一时间段内直播业务对应的回源率。接下来对方式一和方式二进行详细说明。
方式一,管理设备获取至少一个基础特征,该至少一个基础特征包括该回流流量和第一配比。将该至少一个基础特征输入到回源率预测模型中,使回源率预测模型基于该至少一个基础特征得到在第一时间段内直播业务对应的回源率。获取回源预测模型输出的该回源率。
在一些实施例中,该至少一个基础特征还包括如下一个或多个,该边缘节点总数量或者该数据流总数量。
方式二,管理设备获取至少一个基础特征,该至少一个基础特征包括该回流流量和第一配比,基于该至少一个基础特征获取至少一个衍生特征。将该至少一个基础特征和该至少一个衍生特征输入到回源率预测模型中,使回源率预测模型基于该至少一个基础特征和该至少一个衍生特征得到在第一时间段内直播业务对应的回源率。获取回源预测模型输出的该回源率。
在一些实施例中,该至少一个基础特征还包括如下一个或多个,该边缘节点总数量或者该数据流总数量。
该至少一个衍生特征包括如下一个或多个,在第一时间段内直播业务的观众用户总数量,直播业务的主播用户发送的数据流对应的平均观众用户数量,直播业务的主播用户发送的数据流对应的平均出流流量,在第一时间段内边缘节点发送的平均数据流数量,在第一时间段内向边缘节点请求数据流的平均观众用户数量,或者,在第一时间段内边缘节点发送的平均出流流量。
接下来列举了管理设备获取至少一个衍生特征的实现实例。当然,还有其他获取该至少一个衍生特征的实现实例,在此不再一一列举。该实现实例为:
在一些实施例中,基于该出流流量和直播业务的数据流的标准数据量,获取观众用户总数量。该标准数据量是一个固定值,可选地,该观众用户总数据量等于该出流流量与该标准数据量之间的比值。
在一些实施例中,基于该观众用户总数量和该数据流总数量,获取平均观众用户数量。可选地,该平均观众用户数量等于该观众用户总数量与该数据流总数量之间的比值。
在一些实施例中,基于该出流流量和该数据流总数量,获取平均出流流量。可选地,该平均出流流量等于该出流流量与该数据流总数量之间的比值。
在一些实施例中,基于该数据流总数量和该边缘节点总数量,获取平均数据流数量。可选地,该平均数据流数量等于该数据流总数量与该边缘节点总数量之间的比值。
在一些实施例中,基于该观众用户总数量和该边缘节点总数量,获取平均观众用户数量。可选地,该平均观众用户数量等于该观众用户总数量与该边缘节点总数量之间的比值。
在一些实施例中,基于该出流流量和该边缘节点总数量,获取平均出流流量。可选地,该平均出流流量等于该出流流量与该边缘节点总数量之间的比值。
该回源率预测模型是事先训练得到的,该训练过程为:基于至少一个训练样本,获取回源率预测模型,每个训练样本包括历史获取的直播业务的回源流量、历史获取的直播业务的冷流、温流和热流之间的配比,以及基于历史获取的回源流量和该配比得到的回源率。
也就是说,每个训练样本包括至少一个基础特征和与该至少一个基础特征对应的回源率,或者,每个训练样本包括至少一个基础特征、至少一个衍生特征、与该至少一个基础特征和该至少一个衍生特征对应的回源率。
在一些实施例中,基于至少一个训练样本,通过如下6021-6023的操作,获取回源率预测模型。
6021:基于至少一个训练样本和待训练回源率预测模型,识别每个训练样本对应的回源率。
对于任一个训练样本,将该训练样本包括的至少一个基础特征输入到待训练回源率预测模型中,使待训练回源率预测模型基于该至少一个基础特征得到回源率,获取待训练回源预测模型输出的该回源率,将该回源率作为该训练样本对应的回源率。或者,
将该训练样本包括的至少一个基础特征和至少一个衍生特征输入到待训练回源率预测模型中,使待训练回源率预测模型基于该至少一个基础特征和该至少一个衍生特征得到回源率,获取待训练回源预测模型输出的该回源率,将该回源率作为该训练样本对应的回源率。
6022:基于每个训练样本包括的回源率和每个训练样本对应的回源率,通过损失函数计算损失值,基于该损失值调整待训练回源率预测模型的参数。
6023:确定是否继续训练待训练回源率预测模型,如果确定继续训练待训练回源率预测模型,返回步骤6021,如果确定不继续训练待训练回源率预测模型,将待训练回源率预测模型作为回源率预测模型。
在一些实施例中,当对待训练回源率预测模型进行训练的次数达到指定次数时,确定不继续对待训练回源率预测模型进行训练。或者,
使用多个校验样本获取待训练回源率预测模型获取回源率的精度,在该精度超过指定阈值,确定不继续对待训练回源率预测模型进行训练。在实现时:
获取多个校验样本,每个校验样本包括至少一个基础特征和与该至少一个基础特征相对应的回源率,或者,每个校验样本包括至少一个基础特征、至少一个衍生特征、与该至少一个基础特征和该至少一个衍生特征相对应的回源率。基于待训练回源率预测模型获取每个校验样本对应的回源率。基于每个校验样本包括的回源率和每个校验样本对应的回源率,计算获取回源率的精度。在该精度未超过指定阈值,确定继续对待训练回源率预测模型进行训练,在该精度超过指定阈值,确定不继续对待训练回源率预测模型进行训练。
参见图8,黑盒分配部件包括特征转换模块,模型训练模块和模型推理模块,黑盒分配部件将历史接收的业务需求信息和数据流属性信息,以及历史获取的回源率输入到特征转换模块。特征转换模块基于业务需求信息和数据流属性信息得到训练样本,模型训练模块基于该训练样本训练回源率预测模块。模型推理模块基于第一时间段内的业务需求信息、数据流属性信息和回源率预测模型,获取第一时间段内直播业务的回源率。
步骤603:管理设备基于该回源率和该出流流量获取回源流量。
在步骤603中,管理设备将该回源率与该出流流量相乘,得到回源流量。
步骤604:管理设备基于该回源流量和该出流流量,在内容分发网络中分配网络资源。
管理设备分配网络资源的详细实现过程,参见图4所示的方法400的步骤404中的相关内容,在此不再详细说明。
在步骤604中,管理设备还可能获取在第一时间段内内容分发网络传输直播业务所需要的费用。
管理设备包括出流流量对应的第一价格以及回源流量对应的第二价格。管理设备基于出流流量、回流流量、第一价格和第二价格,获取内容分发网络传输直播业务所需要的费用。
在本申请实施例中,管理设备接收在第一时间段内直播业务的业务需求信息和直播业务的数据流属性信息。管理设备基于该回流流量、第一配比和回源率预测模型,获取在第一时间段内直播业务对应的回源率,基于回源率和业务需求信息包括的出流流量获取在第一时间段内直播业务产生的回源流量。基于该回源流量和出流流量在内容分发网络中分配网络资源。由于获取直播业务的数据流属性信息,基于直播业务的业务属性信息和业务需求信息获取在第一时间段内直播业务产生的回源流量和出流流量,这样在获取流量时参考了直播业务的数据流属性,不同时间段直播业务的数据流属性可能不同,从而提高了分配网络资源的灵活性。另外,在管理设备历史接收的业务需求信息包括直播业务的业务需求信息和管理设备历史接收的数据流属性信息不包括直播业务的数据流属性信息时,管理设备基于该回流流量、第一配比和回源率预测模型,获取在第一时间段内直播业务对应的回源率,这样提高获取回源率的精度,同时通过回源率预 测模型来获取回源率,提高获取回源率的效率,进而提高了分配网络资源的效率。
参见图9,本申请实施例提供了一种分配网络资源的装置900,该装置900部署在图2所示网络架构200的管理设备上,图4所示方法400的管理设备上或图6所示的方法600的管理设备上。该装置900包括:
信息获取模块901,用于获取在第一时间段内直播业务的业务需求信息和直播业务的数据流属性信息,业务需求信息用于描述在第一时间段内内容分发网络传输直播业务的数据流的情况,第一时间段位于当前时间之后,内容分发网络包括中心节点和边缘节点,中心节点用于向边缘节点发送来自直播业务的主播用户的数据流,边缘节点用于向请求该数据流的观众用户发送该数据流;
流量获取模块902,用于基于业务需求信息和数据流属性信息,获取在第一时间段内直播业务产生的回源流量和出流流量,该回源流量是中心节点发送的该数据流产生的流量,该出流流量是边缘节点发送的该数据流产生的流量;
资源分配模块903,用于基于该回源流量和该出流流量在内容分发网络中分配网络资源。
可选地,信息获取模块901获取业务需求信息和数据流属性信息的详细实现过程参见图4所示方法400的步骤401和图6所示方法600的步骤601中的相关内容,在此不再详细说明。
可选地,流量获取模块902获取在第一时间段内直播业务产生的回源流量和出流流量的详细实现过程参见图4所示方法400的步骤402-403和图6所示方法600的步骤602-603中的相关内容,在此不再详细说明。
可选地,资源分配模块903分配网络资源的详细实现过程参见图4所示方法400的步骤404和图6所示方法600的步骤604中的相关内容,在此不再详细说明。
可选地,业务需求信息包括该出流流量,数据流属性信息包括直播业务的冷流、温流和热流之间的第一配比;
流量获取模块902,用于:
基于第一配比获取在第一时间段内直播业务对应的回源率,该回源率是回源流量与出流流量之间的比例;
基于该回源率和出流流量,获取回源流量。
可选地,流量获取模块902基于第一配比获取在第一时间段内直播业务对应的回源率的详细实现过程参见图4所示方法400的步骤402中的相关内容,在此不再详细说明。
可选地,冷流是请求的观众用户数量不超过冷流阈值的数据流,温流是请求的观众用户数量超过冷流阈值且未超过热流阈值的数据流,热流是请求的观众用户数量超过热流阈值的数据流,热流阈值大于冷流阈值。
可选地,流量获取模块902基于该回源率和出流流量,获取回源流量的详细实现过程参见图4所示方法400的步骤403中的相关内容,在此不再详细说明。
可选地,业务需求信息还包括在第一时间段内需要直播业务的主播用户向中心节点发送的数据流总数量,
流量获取模块902,用于基于数据流总数量和第一配比获取在第一时间段内直播业务对应的回源率。
可选地,流量获取模块902,用于:
基于多个流类型中的每个流类型对应的标准边缘节点数,获取在第一时间段内每个流类型对应的边缘节点数,请求所述每个流类型的数据流的观众用户数量不同;
基于每个流类型对应的标准数据流数量和第一配比,获取在第一时间段内每个流类型对应的数据流数量;
基于每个流类型对应的数据流数量和边缘节点数,获取在第一时间段内直播业务对应的回源率。
可选地,流量获取模块902获取在第一时间段内每个流类型对应的边缘节点数的详细实现过程参见图4所示方法400的操作4022中的相关内容,在此不再详细说明。
可选地,流量获取模块902获取在第一时间段内每个流类型对应的数据流数量的详细实现过程参见图4所示方法400的操作4021中的相关内容,在此不再详细说明。
可选地,流量获取模块902获取在第一时间段内直播业务对应的回源率的详细实现过程参见图4所示 方法400的操作4023中的相关内容,在此不再详细说明。
可选地,信息获取模块901,还用于获取内容分发网络的网络信息,该网络信息包括如下一个或多个,在第一时间段内内容分发网络包括的边缘节点总数量,在第一时间段内内容分发网络的冷流阈值,或者,在第一时间段内内容分发网络的热流阈值;
流量获取模块902,用于基于该网络信息和每个流类型对应的标准边缘节点数,获取在第一时间段内每个流类型对应的边缘节点数。
可选地,信息获取模块901获取内容分发网络的网络信息的详细实现过程参见图4所示方法400的步骤401和图6所示方法600的步骤601中的相关内容,在此不再详细说明。
可选地,流量获取模块902基于该网络信息和每个流类型对应的标准边缘节点数,获取在第一时间段内每个流类型对应的边缘节点数的详细实现过程参见图4所示方法400的步骤402中的相关内容,在此不再详细说明。
可选地,流量获取模块902,用于在历史获取的业务需求信息中不包括直播业务的业务需求信息和/或历史获取的数据流属性信息不包括直播业务的数据流属性信息时,基于第一配比获取在第一时间段内直播业务对应的回源率。
可选地,业务需求信息包括该出流流量,数据流属性信息包括直播业务包括的冷流、温流和热流之间的第一配比;
流量获取模块902,用于:
基于回流流量、第一配比和回源率预测模型,获取在第一时间段内直播业务对应的回源率,该回源率是回源流量与出流流量之间的比例;
基于该回源率和出流流量,获取回源流量。
可选地,流量获取模块902基于回流流量、第一配比和回源率预测模型,获取在第一时间段内直播业务对应的回源率的详细实现过程参见图6所示方法600的步骤602中的相关内容,在此不再详细说明。
可选地,流量获取模块902基于该回源率和出流流量,获取回源流量的详细实现过程参见图6所示方法600的步骤603中的相关内容,在此不再详细说明。
可选地,业务需求信息还包括在第一时间段内需要直播业务的主播用户向中心节点发送的数据流总数量,
流量获取模块902,用于基于数据流总数量、回流流量、第一配比和回源率预测模型,获取在第一时间段内直播业务对应的回源率。
可选地,流量获取模块902,用于:
获取在第一时间段内内容分发网络包括的边缘节点总数量;
基于边缘节点总数量、数据流总数量、回流流量、第一配比和回源率预测模型,获取在第一时间段内直播业务对应的回源率。
可选地,流量获取模块902,用于:
基于出流流量、数据流总数量和/或边缘节点总数量,获取直播业务对应的至少一个衍生特征;
基于至少一个衍生特征、边缘节点总数量、数据流总数量、回流流量、第一配比和回源率预测模型,获取在第一时间段内直播业务对应的回源率。
可选地,流量获取模块902获取直播业务对应的至少一个衍生特征的详细实现过程参见图6所示方法600的步骤602中的相关内容,在此不再详细说明。
可选地,流量获取模块902基于至少一个衍生特征、边缘节点总数量、数据流总数量、回流流量、第一配比和回源率预测模型,获取在第一时间段内直播业务对应的回源率的详细实现过程参见图6所示方法600的步骤603中的相关内容,在此不再详细说明。
可选地,至少一个衍生特征包括如下一个或多个,在第一时间段内直播业务的观众用户总数量,直播业务的主播用户发送的数据流对应的平均观众用户数量,直播业务的主播用户发送的数据流对应的平均出流流量,在第一时间段内边缘节点发送的平均数据流数量,在第一时间段内向边缘节点请求数据流的平均观众用户数量,或者,在第一时间段内边缘节点发送的平均出流流量;
流量获取模块902,用于:
基于出流流量和直播业务的数据流的标准数据量,获取观众用户总数量;和/或,
基于观众用户总数量和数据流总数量,获取平均观众用户数量;和/或,
基于出流流量和数据流总数量,获取平均出流流量;和/或,
基于数据流总数量和边缘节点总数量,获取平均数据流数量;和/或,
基于观众用户总数量和边缘节点总数量,获取平均观众用户数量;和/或,
基于出流流量和边缘节点总数量,获取平均出流流量。
可选地,流量获取模块902,用于在历史获取的业务需求信息包括直播业务的需求信息以及历史获取的数据流属性信息包括直播业务的数据流属性信息时,基于回流流量、第一配比和回源率预测模型,获取在第一时间段内直播业务对应的回源率。
可选地,所述装置900还包括:
模型获取模块904,用于基于至少一个训练样本,获取回源率预测模型,每个训练样本包括历史获取的回源流量、历史获取的第一配比,以及基于历史获取的回源流量和第一配比得到的回源率。
可选地,模型获取模块904基于至少一个训练样本,获取回源率预测模型的详细实现过程参见图6所示方法600的操作6021-6023中的相关内容,在此不再详细说明。
可选地,所述装置900还包括:
费用获取模块905,用于基于该回源流量和该出流流量,获取内容分发网络传输直播业务所需要的费用。
在本申请实施例中,信息获取模块获取在第一时间段内直播业务的业务需求信息和直播业务的数据流属性信息,业务需求信息用于描述在第一时间段内内容分发网络传输直播业务的数据流的情况,第一时间段位于当前时间之后。流量获取模块基于业务需求信息和数据流属性信息,获取在第一时间段内直播业务产生的回源流量和出流流量。资源分配模块基于该回源流量和该出流流量在内容分发网络中分配网络资源。由于信息获取模块获取直播业务的数据流属性信息,流量获取模块基于直播业务的业务属性信息和业务需求信息获取在第一时间段内直播业务产生的回源流量和出流流量,这样在获取流量时参考了直播业务的数据流属性,不同时间段直播业务的数据流属性可能不同,基于不同时间段的数据流属性得到的回源流量也不同,从而提高了分配网络资源的灵活性。
参见图10,本申请实施例提供了一种计算机设备1000示意图。该计算机设备1000可以是上述任意实施例中的管理设备。例如该计算机设备1000可以是上述图2所示网络架构200中的管理设备,或者,是上述图4所示方法400或图6所示方法600中的管理设备。该计算机设备1000包括至少一个处理器1001,内部连接1002,存储器1003以及至少一个收发器1004。
该计算机设备1000是一种硬件结构的装置,可以用于实现图9所述的装置900中的功能模块。例如,本领域技术人员可以想到图9所示的装置900中的信息获取模块901、流量获取模块902、资源分本模块903、模型获取模块904和费用获取模块905可以通过该至少一个处理器1001调用存储器1003中的代码来实现。
可选的,该计算机设备1000还可用于实现上述任一实施例中管理设备的功能。
可选的,上述处理器1001可以是一个通用中央处理器(central processing unit,CPU),网络处理器(network processor,NP),微处理器,特定应用集成电路(application-specific integrated circuit,ASIC),或一个或多个用于控制本申请方案程序执行的集成电路。
上述内部连接1002可包括一通路,在上述组件之间传送信息。可选的,内部连接1002为单板或总线等。
上述收发器1004,用于与其他设备或通信网络通信。
上述存储器1003可以是只读存储器(read-only memory,ROM)或可存储静态信息和指令的其他类型的静态存储设备,随机存取存储器(random access memory,RAM)或者可存储信息和指令的其他类型的动态存储设备,也可以是电可擦可编程只读存储器(electrically erasable programmable read-only memory,EEPROM)、只读光盘(compact disc read-only memory,CD-ROM)或其他光盘存储、光碟存储(包括压缩光碟、激光碟、光碟、数字通用光碟、蓝光光碟等)、磁盘存储介质或者其他磁存储设备、或者能够用于携带或存储具有指令或数据结构形式的期望的程序代码并能够由计算机存取的任何其他介质,但不限于此。存储器可以是独立存在,通过总线与处理器相连接。存储器也可以和处理器集成在一起。
其中,存储器1003用于存储执行本申请方案的应用程序代码,并由处理器1001来控制执行。处理器1001用于执行存储器1003中存储的应用程序代码,以及配合至少一个收发器1004,从而使得该计算机设备1000实现本专利方法中的功能。
在具体实现中,作为一种实施例,处理器1001可以包括一个或多个CPU,例如图10中的CPU0和CPU1。
在具体实现中,作为一种实施例,该计算机设备1000可以包括多个处理器,例如图10中的处理器1001和处理器1007。这些处理器中的每一个可以是一个单核(single-CPU)处理器,也可以是一个多核(multi-CPU)处理器。这里的处理器可以指一个或多个设备、电路、和/或用于处理数据(例如计算机程序指令)的处理核。
本领域普通技术人员可以理解实现上述实施例的全部或部分步骤可以通过硬件来完成,也可以通过程序来指令相关的硬件完成,所述的程序可以存储于一种计算机可读存储介质中,上述提到的存储介质可以是只读存储器,磁盘或光盘等。
以上所述仅为本申请的可选实施例,并不用以限制本申请,凡在本申请的原则之内,所作的任何修改、等同替换、改进等,均应包含在本申请的保护范围之内。

Claims (29)

  1. 一种分配网络资源的方法,其特征在于,所述方法包括:
    管理设备获取在第一时间段内直播业务的业务需求信息和所述直播业务的数据流属性信息,所述业务需求信息用于描述在所述第一时间段内内容分发网络传输所述直播业务的数据流的情况,所述第一时间段位于当前时间之后,所述内容分发网络包括中心节点和边缘节点,所述中心节点用于向所述边缘节点发送来自所述直播业务的主播用户的数据流,所述边缘节点用于向请求所述数据流的观众用户发送所述数据流;
    所述管理设备基于所述业务需求信息和所述数据流属性信息,获取在所述第一时间段内所述直播业务产生的回源流量和出流流量,所述回源流量是所述中心节点发送的所述数据流产生的流量,所述出流流量是所述边缘节点发送的所述数据流产生的流量;
    所述管理设备基于所述回源流量和所述出流流量在所述内容分发网络中分配网络资源。
  2. 如权利要求1所述的方法,其特征在于,所述业务需求信息包括所述出流流量,所述数据流属性信息包括所述直播业务的冷流、温流和热流之间的第一配比;
    所述基于所述业务需求信息和所述数据流属性信息,获取在所述第一时间段内的所述直播业务产生的回源流量,包括:
    基于所述第一配比获取在所述第一时间段内所述直播业务对应的回源率,所述回源率是所述回源流量与所述出流流量之间的比例;
    基于所述回源率和所述出流流量,获取所述回源流量。
  3. 如权利要求2所述的方法,其特征在于,所述业务需求信息还包括在所述第一时间段内需要所述直播业务的主播用户向所述中心节点发送的数据流总数量,
    所述基于所述第一配比获取在所述第一时间段内所述直播业务对应的回源率,包括:
    基于所述数据流总数量和所述第一配比获取在所述第一时间段内所述直播业务对应的回源率。
  4. 如权利要求2或3所述的方法,其特征在于,所述方法还包括:
    基于多个流类型中的每个流类型对应的标准边缘节点数,获取在所述第一时间段内所述每个流类型对应的边缘节点数,请求所述每个流类型的数据流的观众用户数量不同;
    所述基于所述第一配比获取在所述第一时间段内所述直播业务对应的回源率,包括:
    基于所述每个流类型对应的标准数据流数量和所述第一配比,获取在所述第一时间段内所述每个流类型对应的数据流数量;
    基于所述每个流类型对应的数据流数量和边缘节点数,获取在所述第一时间段内所述直播业务对应的回源率。
  5. 如权利要求4所述的方法,其特征在于,所述基于多个流类型中的每个流类型对应的标准边缘节点数,获取在所述第一时间段内所述每个流类型对应的边缘节点数,包括:
    基于所述内容分发网络的网络信息和所述每个流类型对应的标准边缘节点数,获取在所述第一时间段内所述每个流类型对应的边缘节点数。
  6. 如权利要求2-5任一项所述的方法,其特征在于,所述基于所述第一配比获取在所述第一时间段内所述直播业务对应的回源率,包括:
    在历史获取的业务需求信息中不包括所述直播业务的业务需求信息和/或历史获取的数据流属性信息不包括所述直播业务的数据流属性信息时,基于所述第一配比获取在所述第一时间段内所述直播业务对应的回源率。
  7. 如权利要求1所述的方法,其特征在于,所述业务需求信息包括所述出流流量,所述数据流属性信 息包括所述直播业务包括的冷流、温流和热流之间的第一配比;
    所述基于所述业务需求信息和所述数据流属性信息,获取所述第一时间段内的所述直播业务产生的回源流量,包括:
    基于所述回流流量、所述第一配比和回源率预测模型,获取在所述第一时间段内所述直播业务对应的回源率,所述回源率是所述回源流量与所述出流流量之间的比例;
    基于所述回源率和所述出流流量,获取所述回源流量。
  8. 如权利要求7所述的方法,其特征在于,所述业务需求信息还包括在所述第一时间段内需要所述直播业务的主播用户向所述中心节点发送的数据流总数量,
    所述基于所述回流流量、所述第一配比和回源率预测模型,获取在所述第一时间段内所述直播业务对应的回源率,包括:
    基于所述数据流总数量、所述回流流量、所述第一配比和所述回源率预测模型,获取在所述第一时间段内所述直播业务对应的回源率。
  9. 如权利要求8所述的方法,其特征在于,所述方法还包括:
    获取在所述第一时间段内所述内容分发网络包括的边缘节点总数量;
    所述基于所述数据流总数量、所述回流流量、所述第一配比和所述回源率预测模型,获取在所述第一时间段内所述直播业务对应的回源率,包括:
    基于所述边缘节点总数量、所述数据流总数量、所述回流流量、所述第一配比和所述回源率预测模型,获取在所述第一时间段内所述直播业务对应的回源率。
  10. 如权利要求9所述的方法,其特征在于,所述方法还包括:
    基于所述出流流量、所述数据流总数量和/或所述边缘节点总数量,获取所述直播业务对应的至少一个衍生特征;
    所述基于所述边缘节点总数量、所述数据流总数量、所述回流流量、所述第一配比和所述回源率预测模型,获取在所述第一时间段内所述直播业务对应的回源率,包括:
    基于所述至少一个衍生特征、所述边缘节点总数量、所述数据流总数量、所述回流流量、所述第一配比和所述回源率预测模型,获取在所述第一时间段内所述直播业务对应的回源率。
  11. 如权利要求7-10任一项所述的方法,其特征在于,所述基于所述回流流量、所述第一配比和回源率预测模型,获取在所述第一时间段内所述直播业务对应的回源率,包括:
    在历史获取的业务需求信息包括所述直播业务的需求信息以及历史获取的数据流属性信息包括所述直播业务的数据流属性信息时,基于所述回流流量、所述第一配比和所述回源率预测模型,获取在所述第一时间段内所述直播业务对应的回源率。
  12. 如权利要求7-11任一项所述的方法,其特征在于,所述方法还包括:
    基于至少一个训练样本,获取所述回源率预测模型,每个训练样本包括历史获取的回源流量、历史获取的第一配比,以及基于所述历史获取的回源流量和第一配比得到的回源率。
  13. 如权利要求1-12任一项所述的方法,其特征在于,所述方法还包括:
    基于所述回源流量和所述出流流量,获取所述内容分发网络传输所述直播业务所需要的费用。
  14. 一种分配网络资源的装置,其特征在于,所述装置包括:
    信息获取模块,用于获取在第一时间段内直播业务的业务需求信息和所述直播业务的数据流属性信息,所述业务需求信息用于描述在所述第一时间段内内容分发网络传输所述直播业务的数据流的情况,所述第一时间段位于当前时间之后,所述内容分发网络包括中心节点和边缘节点,所述中心节点用于向所述边缘节点发送来自所述直播业务的主播用户的数据流,所述边缘节点用于向请求所述数据流的观众用户发送所 述数据流;
    流量获取模块,用于基于所述业务需求信息和所述数据流属性信息,获取在所述第一时间段内所述直播业务产生的回源流量和出流流量,所述回源流量是所述中心节点发送的所述数据流产生的流量,所述出流流量是所述边缘节点发送的所述数据流产生的流量;
    资源分配模块,用于基于所述回源流量和所述出流流量在所述内容分发网络中分配网络资源。
  15. 如权利要求14所述的装置,其特征在于,所述业务需求信息包括所述出流流量,所述数据流属性信息包括所述直播业务的冷流、温流和热流之间的第一配比;
    所述流量获取模块,用于:
    基于所述第一配比获取在所述第一时间段内所述直播业务对应的回源率,所述回源率是所述回源流量与所述出流流量之间的比例;
    基于所述回源率和所述出流流量,获取所述回源流量。
  16. 如权利要求15所述的装置,其特征在于,所述业务需求信息还包括在所述第一时间段内需要所述直播业务的主播用户向所述中心节点发送的数据流总数量,
    所述流量获取模块,用于基于所述数据流总数量和所述第一配比获取在所述第一时间段内所述直播业务对应的回源率。
  17. 如权利要求15或16所述的装置,其特征在于,所述流量获取模块,用于:
    基于多个流类型中的每个流类型对应的标准边缘节点数,获取在所述第一时间段内所述每个流类型对应的边缘节点数,请求所述每个流类型的数据流的观众用户数量不同;
    基于所述每个流类型对应的标准数据流数量和所述第一配比,获取在所述第一时间段内所述每个流类型对应的数据流数量;
    基于所述每个流类型对应的数据流数量和边缘节点数,获取在所述第一时间段内所述直播业务对应的回源率。
  18. 如权利要求17所述的装置,其特征在于,所述流量获取模块,用于基于所述内容分发网络的网络信息和所述每个流类型对应的标准边缘节点数,获取在所述第一时间段内所述每个流类型对应的边缘节点数。
  19. 如权利要求15-18任一项所述的装置,其特征在于,所述流量获取模块,用于在历史获取的业务需求信息中不包括所述直播业务的业务需求信息和/或历史获取的数据流属性信息不包括所述直播业务的数据流属性信息时,基于所述第一配比获取在所述第一时间段内所述直播业务对应的回源率。
  20. 如权利要求14所述的装置,其特征在于,所述业务需求信息包括所述出流流量,所述数据流属性信息包括所述直播业务包括的冷流、温流和热流之间的第一配比;
    所述流量获取模块,用于:
    基于所述回流流量、所述第一配比和回源率预测模型,获取在所述第一时间段内所述直播业务对应的回源率,所述回源率是所述回源流量与所述出流流量之间的比例;
    基于所述回源率和所述出流流量,获取所述回源流量。
  21. 如权利要求20所述的装置,其特征在于,所述业务需求信息还包括在所述第一时间段内需要所述直播业务的主播用户向所述中心节点发送的数据流总数量,
    所述流量获取模块,用于基于所述数据流总数量、所述回流流量、所述第一配比和所述回源率预测模型,获取在所述第一时间段内所述直播业务对应的回源率。
  22. 如权利要求21所述的装置,其特征在于,所述流量获取模块,用于:
    获取在所述第一时间段内所述内容分发网络包括的边缘节点总数量;
    基于所述边缘节点总数量、所述数据流总数量、所述回流流量、所述第一配比和所述回源率预测模型,获取在所述第一时间段内所述直播业务对应的回源率。
  23. 如权利要求22所述的装置,其特征在于,所述流量获取模块,用于:
    基于所述出流流量、所述数据流总数量和/或所述边缘节点总数量,获取所述直播业务对应的至少一个衍生特征;
    基于所述至少一个衍生特征、所述边缘节点总数量、所述数据流总数量、所述回流流量、所述第一配比和所述回源率预测模型,获取在所述第一时间段内所述直播业务对应的回源率。
  24. 如权利要求20-23任一项所述的装置,其特征在于,所述流量获取模块,用于在历史获取的业务需求信息包括所述直播业务的需求信息以及历史获取的数据流属性信息包括所述直播业务的数据流属性信息时,基于所述回流流量、所述第一配比和所述回源率预测模型,获取在所述第一时间段内所述直播业务对应的回源率。
  25. 如权利要求20-24任一项所述的装置,其特征在于,所述装置还包括:
    模型获取模块,用于基于至少一个训练样本,获取所述回源率预测模型,每个训练样本包括历史获取的回源流量、历史获取的第一配比,以及基于所述历史获取的回源流量和第一配比得到的回源率。
  26. 如权利要求14-25任一项所述的装置,其特征在于,所述装置还包括:
    费用获取模块,用于基于所述回源流量和所述出流流量,获取所述内容分发网络传输所述直播业务所需要的费用。
  27. 一种计算机设备,其特征在于,包括至少一个处理器,所述至少一个处理器用于与存储器耦合,读取并执行所述存储器中的指令,以实现如权利要求1-13任一项所述的方法。
  28. 一种计算机可读存储介质,其上存储有计算机程序,其特征在于,所述计算机程序被计算机执行时,实现如权利要求1-13任一项所述的方法。
  29. 一种计算机程序产品,其特征在于,所述计算机程序产品包括在计算机可读存储介质中存储的计算机程序,并且所述计算程序通过处理器进行加载来实现如权利要求1-13任一项所述的方法。
PCT/CN2023/111572 2022-09-06 2023-08-07 分配网络资源的方法、装置及存储介质 WO2024051424A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202211085847.5A CN117714755A (zh) 2022-09-06 2022-09-06 分配网络资源的方法、装置及存储介质
CN202211085847.5 2022-09-06

Publications (1)

Publication Number Publication Date
WO2024051424A1 true WO2024051424A1 (zh) 2024-03-14

Family

ID=90157540

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/111572 WO2024051424A1 (zh) 2022-09-06 2023-08-07 分配网络资源的方法、装置及存储介质

Country Status (2)

Country Link
CN (1) CN117714755A (zh)
WO (1) WO2024051424A1 (zh)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2007241667A (ja) * 2006-03-08 2007-09-20 Nec Corp 業務フロー制御システム、業務フロー制御方法および制御用プログラム
CN108366020A (zh) * 2018-02-02 2018-08-03 网宿科技股份有限公司 一种发送数据资源的获取请求的方法和系统
CN110572687A (zh) * 2019-08-09 2019-12-13 北京达佳互联信息技术有限公司 直播回源聚合的方法、装置、系统、设备及存储介质
CN111327461A (zh) * 2020-01-23 2020-06-23 华为技术有限公司 一种基于cdn系统的域名管理方法、装置、设备及介质
CN112929676A (zh) * 2019-12-06 2021-06-08 北京金山云网络技术有限公司 一种直播数据流获取方法、装置、节点及系统
CN114501073A (zh) * 2022-02-16 2022-05-13 上海哔哩哔哩科技有限公司 直播回源方法及装置

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2007241667A (ja) * 2006-03-08 2007-09-20 Nec Corp 業務フロー制御システム、業務フロー制御方法および制御用プログラム
CN108366020A (zh) * 2018-02-02 2018-08-03 网宿科技股份有限公司 一种发送数据资源的获取请求的方法和系统
CN110572687A (zh) * 2019-08-09 2019-12-13 北京达佳互联信息技术有限公司 直播回源聚合的方法、装置、系统、设备及存储介质
CN112929676A (zh) * 2019-12-06 2021-06-08 北京金山云网络技术有限公司 一种直播数据流获取方法、装置、节点及系统
CN111327461A (zh) * 2020-01-23 2020-06-23 华为技术有限公司 一种基于cdn系统的域名管理方法、装置、设备及介质
CN114501073A (zh) * 2022-02-16 2022-05-13 上海哔哩哔哩科技有限公司 直播回源方法及装置

Also Published As

Publication number Publication date
CN117714755A (zh) 2024-03-15

Similar Documents

Publication Publication Date Title
US9112809B2 (en) Method and apparatus for controlling utilization in a horizontally scaled software application
US8612615B2 (en) Systems and methods for identifying usage histories for producing optimized cloud utilization
US9953351B1 (en) Managing resource requests that exceed reserved resource capacity
CN111641973B (zh) 一种雾计算网络中基于雾节点协作的负载均衡方法
US20140229210A1 (en) System and Method for Network Resource Allocation Considering User Experience, Satisfaction and Operator Interest
CN110022269B (zh) 通信数据处理方法、装置和设备
Nan et al. Queueing model based resource optimization for multimedia cloud
CN109819057A (zh) 一种负载均衡方法及系统
CN110069341A (zh) 边缘计算中结合功能按需配置的有依赖关系任务的调度方法
CN111614754B (zh) 面向雾计算的成本效率优化的动态自适应任务调度方法
Alencar et al. Dynamic microservice allocation for virtual reality distribution with qoe support
CN113452566A (zh) 一种云边端协同资源管理方法及系统
CN104994150B (zh) 一种面向云视频服务的请求分配方法
CN109254726A (zh) 分布式存储系统中服务质量保障方法、控制节点及系统
US20140143427A1 (en) Providing Resources in a Cloud
WO2017075967A1 (zh) 在线媒体服务的带宽分配方法及系统
CN108965884A (zh) 一种转码任务的分配方法及调度设备、转码设备
WO2020177255A1 (zh) 无线接入网的资源分配方法及装置
TW202121274A (zh) 雲端資源管理方法、裝置、電子設備及電腦可讀儲存媒體
CN107733805A (zh) 业务负载调度方法和装置
WO2023108761A1 (zh) 监控业务带宽分配方法、装置、电子设备及存储介质
CN111611076A (zh) 任务部署约束下移动边缘计算共享资源公平分配方法
Shen et al. Deadline-aware rate allocation for IoT services in data center network
WO2021057981A1 (zh) 云计费方法、装置、云管理平台、系统及存储介质
CN113778675A (zh) 一种基于面向区块链网络的计算任务分配系统及方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23862122

Country of ref document: EP

Kind code of ref document: A1