WO2017214986A1 - 一种云应用伸缩方法及装置 - Google Patents

一种云应用伸缩方法及装置 Download PDF

Info

Publication number
WO2017214986A1
WO2017214986A1 PCT/CN2016/086267 CN2016086267W WO2017214986A1 WO 2017214986 A1 WO2017214986 A1 WO 2017214986A1 CN 2016086267 W CN2016086267 W CN 2016086267W WO 2017214986 A1 WO2017214986 A1 WO 2017214986A1
Authority
WO
WIPO (PCT)
Prior art keywords
action
cloud application
load
vnf
vms
Prior art date
Application number
PCT/CN2016/086267
Other languages
English (en)
French (fr)
Inventor
李飞
唐朋成
Original Assignee
华为技术有限公司
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 华为技术有限公司 filed Critical 华为技术有限公司
Priority to PCT/CN2016/086267 priority Critical patent/WO2017214986A1/zh
Publication of WO2017214986A1 publication Critical patent/WO2017214986A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]

Definitions

  • the present invention relates to the field of communications, and in particular, to a cloud application scaling method and apparatus.
  • NFV Network Functions Virtualization
  • EPC Evolved Packet Core
  • a virtual network function (VNF) system can be composed of multiple virtual machines (VMs) to support services running on the VNF system.
  • VMs virtual machines
  • This adjustment process is called Capacity expansion/reduction.
  • Capacity expansion/reduction can also be called cloud application scaling.
  • the existing cloud application scaling method is: first setting an upper threshold and a lower threshold of the system load, and when the system load exceeds a preset upper threshold, performing an action corresponding to the corresponding (for example, adding a VM); similarly, When the load of the system is lower than the preset lower threshold, the corresponding action (such as deleting a VM) is performed.
  • the upper threshold and the lower threshold are preset, cannot be changed before the user makes a change, and when the system load exceeds a preset upper threshold or is lower than a preset
  • the lower threshold is used, the number of VMs that are increased or decreased can only be a fixed value, and cannot be flexibly adjusted according to changes in system load, resulting in low resource utilization and system reliability degradation.
  • the embodiment of the present invention provides a cloud application scaling method and device, which can analyze and derive a virtual machine capacity expansion/reduction capacity strategy with the largest first value of the VNF system, thereby Increase the utilization of resources and improve the reliability of the system.
  • an embodiment of the present invention provides a cloud application scaling method, including:
  • the cloud application scaling device acquires the load of the virtual network function VNF system and the number of virtual machine VMs. Secondly, the cloud application scaling device calculates the VNF that can be obtained by performing various actions in the action set in the case of the load and the number of VMs. The first revenue value of the system, each action in the action set indicates to increase or decrease the number of VMs; finally, the cloud application expansion device acquires and executes the first action that can obtain the maximum first value of the VNF system.
  • the cloud application scaling device calculates the first revenue value of the VNF system that can be obtained by performing various actions in the action set in the case of the load and the number of VMs, and acquires the The first action that can obtain the largest first value of the VNF system, the first action is the best result of the virtual machine expansion/contraction, which can reduce the unnecessary expansion/contraction operation and greatly alleviate the ping-pong effect. It also improves the utilization of resources, improves the reliability of the system, and reduces the risk of the VNF system violating the Service-Level Agreement (SLA). It is especially suitable for application scenarios that require high SLA guarantee.
  • SLA Service-Level Agreement
  • the load and the number of VMs of the VNF system are the load and the number of VMs at the first moment of the VNF system.
  • the method further includes:
  • the cloud application expansion device obtains the predicted second time load, and the second time is later than the first time; secondly, the cloud application expansion device calculates each of the action sets in the case of the first prediction result and the number of VMs.
  • the second benefit value of the VNF system that can be obtained by the action is calculated.
  • the second benefit value of the VNF system that can be obtained by the cloud application expansion device according to each action and the first benefit value of the VNF system that can be obtained by each action Calculate the third benefit value of the VNF system that can be obtained by each action.
  • Obtain and execute the first action including:
  • the cloud application scaling device acquires and executes the second action that is maximized for the third benefit value of the VNF system.
  • the cloud application scaling method in order to ensure the stability of the NFV system, the cloud application scaling method is forward-looking, and the cloud application scaling method provided by the embodiment of the present invention can also calculate the capacity expansion/reduction capacity result. Consider the load trend at the second moment later than the first moment.
  • obtaining the first prediction result specifically includes:
  • the cloud application scaling device calculates the first prediction result according to the load of the VNF system and a preset algorithm.
  • the method further includes:
  • the cloud application scaling device receives configuration information that is sent by the network element management system EMS and includes at least a preset algorithm.
  • the configuration information includes at least an SLA indicator and a corresponding SLA threshold; a maximum/minimum VM number of the system; a maximum number of VMs added/deleted; a maximum number of PDPs/users that the system can support; a maximum load burst amount; Quantify the number of particles.
  • the method further includes:
  • the cloud application expansion device confirms that the first action is allowed to execute.
  • the cloud application scaling device can determine whether to finally perform the capacity expansion/reduction operation according to the scheduling criterion.
  • an embodiment of the present invention provides a cloud application expansion apparatus, including an acquisition module, a calculation module, and an execution module.
  • the obtaining module is configured to obtain the load of the virtual network function VNF system and the number of VMs of the virtual machine; and the calculating module is configured to calculate the load and the number of VMs after the acquiring module obtains the load and the number of VMs of the VNF system, Performing a first benefit value for the VNF system that can be obtained by each action in the action set, wherein each action in the action set indicates an increase or decrease in the number of VMs; the acquisition module is also used to calculate the load and the VM in the calculation module.
  • the first action is obtained, wherein the first action is the maximum value of the first benefit to the VNF system that can be obtained.
  • the action module is configured to execute the first action after the obtaining module obtains the first action.
  • the technical effects of the cloud application expansion device provided by the embodiment of the present invention can be referred to the above.
  • the technical effects of the cloud application expansion device described in the cloud application scaling method executed by the cloud application expansion device are not described herein.
  • the load and the number of VMs of the VNF system are the load and the number of VMs at the first moment of the VNF system.
  • the obtaining module is further configured to obtain a first prediction result before the obtaining module acquires the first action, where the first prediction result is the predicted second time load, and the second time is later than the first time; the calculation module further After the obtaining module obtains the first prediction result, calculating a second benefit value of the VNF system that can be obtained by performing each action in the action set in the case of the first prediction result and the number of VMs; and according to each action
  • the second benefit value of the VNF system and the first benefit value of the VNF system that can be obtained by each action are calculated, and the third benefit value of the VNF system that can be obtained by each action is calculated;
  • the acquiring module is specifically used for obtaining The second action, wherein the second action is an action that can obtain the third benefit value that is the largest for the VNF system; and the execution module is specifically configured to perform the second action after the acquiring module obtains the second action.
  • the acquiring device is specifically configured to calculate the first prediction result according to the load of the VNF system and a preset algorithm.
  • the cloud application expansion device further includes a receiving module.
  • the receiving module is configured to receive configuration information sent by the network element management system EMS, where the configuration information includes at least a preset algorithm.
  • the confirmation module is configured to confirm that the first action is allowed to be executed after the obtaining module acquires the first action.
  • an embodiment of the present invention further provides a cloud application expansion device, where the cloud application expansion device includes a memory, a processor, a communication interface, and a system bus.
  • the memory, the processor, and the communication interface are connected by a system bus, the memory is for storing computer instructions, and the processor is configured to execute the computer instructions of the memory storage to cause the cloud application expansion device to perform the cloud application scaling method of the first aspect.
  • the technical effects of the cloud application expansion device provided by the embodiment of the present invention can be referred to the technical effects of the cloud application expansion device described in the cloud application expansion method performed by the cloud application expansion device.
  • an embodiment of the present invention further provides a software product, where the software product includes a computer instruction for implementing a cloud application scaling method.
  • the computer instructions can be stored on a readable storage medium; the processor can read and execute the computer instructions from the readable storage medium such that the processor implements the cloud application scaling method.
  • the embodiment of the invention provides a cloud application scaling method and device, which can obtain the load of the virtual network function VNF system and the number of VMs of the virtual machine; and calculate the number of loads and the number of VMs, and obtain various actions in the action set.
  • the first revenue value of the VNF system wherein each action in the action set indicates an increase or decrease in the number of VMs; and the first action is obtained and executed, wherein the first action is the first available to the VNF system
  • the action with the largest return value since the cloud application expansion device calculates the first revenue value of the VNF system that can be obtained by performing various actions in the action set in the case of the load and the number of VMs, and acquires the pair of VNFs that can be obtained.
  • the first action with the largest first benefit value of the system is the best result of virtual machine expansion/contraction, which can reduce unnecessary expansion/contraction operations, greatly alleviate the ping-pong effect, and improve resources.
  • the utilization rate improves the reliability of the system and reduces the risk of the VNF system violating the SLA. It is especially suitable for application scenarios that require high SLA guarantee.
  • FIG. 1 is a structural diagram of an NFV system in the prior art
  • FIG. 2 is a structural diagram of an NFV system according to an embodiment of the present invention.
  • FIG. 3 is a structural diagram of another NFV system according to an embodiment of the present invention.
  • FIG. 4 is a hardware structural diagram of a cloud application expansion device according to an embodiment of the present invention.
  • FIG. 5 is a schematic flowchart 1 of a cloud application scaling method according to an embodiment of the present disclosure
  • FIG. 6 is a schematic flowchart of a cloud application scaling method according to an embodiment of the present invention. two;
  • FIG. 7 is a schematic flowchart 3 of a cloud application scaling method according to an embodiment of the present disclosure.
  • FIG. 8 is a schematic flowchart 4 of a cloud application scaling method according to an embodiment of the present disclosure.
  • FIG. 9 is a schematic flowchart of a method for updating configuration information stored in a cloud application expansion device according to an embodiment of the present disclosure.
  • FIG. 10 is a schematic structural diagram 1 of a cloud application expansion device according to an embodiment of the present invention.
  • FIG. 11 is a schematic structural diagram 2 of a cloud application expansion device according to an embodiment of the present disclosure.
  • FIG. 12 is a schematic structural diagram 3 of a cloud application expansion device according to an embodiment of the present disclosure.
  • FIG. 13 is a schematic diagram of hardware of a cloud application expansion device according to an embodiment of the present invention.
  • system and “network” are used interchangeably herein.
  • the term “and/or” in this context is merely an association describing the associated object, indicating that there may be three relationships, for example, A and / or B, which may indicate that A exists separately, and both A and B exist, respectively. B these three situations.
  • the character "/" in this article generally indicates that the contextual object is an "or" relationship.
  • FIG. 1 it is an NFV system architecture diagram in the prior art, including: Management and Orchestrator (MANO) and a service network domain, where the service network domain includes an infrastructure layer (NFV Infrastructure, NFVI). ), virtual network layer and operational support layer.
  • MANO Management and Orchestrator
  • NFVI infrastructure layer
  • virtual network layer virtual network layer
  • operational support layer NFV Infrastructure
  • MANO responsible for the management and orchestration of the entire NFVI resources, mapping and association of business networks and NFVI resources, and implementation of business resource processes for the Office of Strategic Services (OSS).
  • MANO is divided into three entities: NFV Orchestrator (NFVO), Virtual Network Function Management (VNF Manager), and Virtual Infrastructure Management (VIM), which respectively support the operation of the service network domain. Layer, virtual network layer and management of NFVI.
  • NFVI From a cloud computing perspective, NFVI is similar to a resource pool. NFVI mapping to physical infrastructure is a data center that is spread across multiple geographic locations, interconnected by high-speed communication networks. NFVI transforms physical resources into virtual resources through virtualization technology and provides all the infrastructure resources that VNF needs.
  • Virtual network layer corresponding to each telecommunication service network, each physical network element in the telecommunication service network is mapped to a single VNF, and the resources required by the VNF are decomposed into virtual computing/storage/network resources, supported by NFVI, and between VNFs.
  • the interface uses a traditional network-defined signaling interface (for example, 3GPP+ITU-T interface), and the VNF service network management system adopts a network element-network element management system-network management system (NE-EMS-NMS) system.
  • NE-EMS-NMS network element-network element management system-network management system
  • VNF is a pure software implementation of traditional network elements deployed on multiple interconnected virtual machines. Multiple VMs within the VNF are interconnected through an internal network that is invisible outside of the VNF. The VNF is interconnected with other VNFs through an external network.
  • Operation support layer It is the current OSS/Business support system (BSS). Mainly used in three areas of billing, service guarantee and service implementation.
  • the architecture of the NFV system provided by the embodiment of the present invention is as shown in FIG. 2, and includes an Element Management System (EMS), a VNF, an NFVO, a VNFM, a VIM, and a cloud application expansion device.
  • the cloud application expansion device may be deployed inside the VNF or may be deployed inside the VNFM.
  • the NFV system architecture diagram shown in FIG. 2 is a scenario where the cloud application expansion device is deployed inside the VNF, and the cloud application expansion device is deployed inside the VNFM. The situation is shown in Figure 3.
  • the VNF is a virtual network element, which is composed of multiple components deployed on one or more VMs, and virtualizes network functions in a traditional network element, including a packet data network.
  • Packet Data Network GateWay (P-GW) packet data network gateway is generally called PDN gateway, Serving GateWay (S-GW), Mobility Management Entity (MME), Home Subscriber Server (Home Subscriber) Server, HSS), etc.
  • the VNF also includes a monitor.
  • the resource load can be Central Processing Unit (CPU) utilization, memory utilization, bandwidth utilization, disk utilization, etc. Resource-related indicators such as storage/network; the service load can be the number of users, the number of sessions, the number of Packet Data Protocol (PDP) contexts, the length of the database read/write queue, and the access delay. index of.
  • the value of each type of load may be an instantaneous value, an average value, a maximum value, a minimum value, an intermediate value, etc. within a certain time window (the time window length is determined according to requirements).
  • the collected data will be periodically sent to the cloud application scaling device for decision making, and the data may be resource load and/or traffic load.
  • Monitor is an internal implementation of VNF. In reality, the functions of Monitor may be completed by other VNF functional components.
  • EMS An element management system that manages one or more network elements of a specific type and is responsible for the function and capacity of each network element.
  • NFVO Manage NFVI virtual infrastructure resources through multiple VIMs to implement resource orchestration functions and network service lifecycle management.
  • VNFM responsible for lifecycle management of VNF instances.
  • a VNFM can be associated with a single VNF instance or with multiple VNF instances of the same type or different types. Regardless of whether the automatic scaling module is deployed in the VNF or in the VNFM, the decision result is reported to the VNFM.
  • VIM Performs corresponding actions according to the decision result of the cloud application scaling device, such as adding a number of VMs or deleting a plurality of VMs, and the number of added or deleted VMs is specified by the cloud application scaling device.
  • the cloud application expansion device is composed of six parts, as shown in FIG. 4, which are a data preprocessing module, a prediction module, a feature extraction module, a learning engine module, a policy library module, and a capacity expansion/reduction capacity selection module.
  • Data preprocessing module resources reported to Monitor The load and the service load are preprocessed, the high frequency noise is eliminated, and the load change trend is extracted. The specific implementation can be smoothed by a digital filter or by extracting an envelope.
  • Prediction module using time series analysis method Load prediction, using historical data to predict short-term load, algorithms that can be used include, but are not limited to, neural network algorithms, support vector regression algorithms, integral autoregressive moving average algorithms, linear regression algorithms, exponential smoothing algorithms, and other algorithms.
  • the load prediction value is used for subsequent model learning and automatic scaling action selection;
  • Feature extraction module extracts the characteristics of the composition state set from the load reported by Monitor, including resource usage rate, current VM number, load change trend, etc.;
  • Learning engine module Using machine learning algorithms, such as enhanced learning, learning the automatic scaling strategy in each state, evaluating the different expansion/reduction actions in the current state through the income function, and obtaining different actions to the VNF system.
  • the functions of the data pre-processing module and the functions of the prediction module can be implemented by the acquisition module of the cloud application retracting device of the following embodiment; the function of the feature extraction module and the function of the learning engine module can be applied by the following embodiment cloud application
  • the calculation module of the expansion device is implemented; the function of the policy library module can be implemented by the memory of the cloud application expansion device of the following embodiment; the function of the expansion/reduction capacity selection module can be implemented by the calculation module of the cloud application expansion device of the following embodiment, The invention does not limit this.
  • the method provided by the embodiment of the present invention can be applied to the service load and the resource load of the cloud application system, and the capacity expansion/reduction capacity modeling is performed separately for different application scenarios, that is, the scenario in which the resource load cannot be obtained is suitable for use.
  • the cloud application scaling method of the service load uses a cloud application scaling method suitable for the resource load in a scenario in which the traffic load cannot be obtained, and any cloud application scaling method is configured according to the requirements when the two types of loads are available.
  • the method provided by the embodiment of the present invention can also be applied to a basic scenario and a comprehensive scenario.
  • the basic scenario is a daily busy hour idle business scenario or a sudden business scenario.
  • the main feature of the daily busy hour business scenario is that the system load will show obvious peaks and troughs with time, and the peaks and troughs will last longer, such as 8:00 to 23:00 every day is busy, other time periods are In the idle time;
  • the sudden business scenario is a short-term business burst on the basis of the daily busy hour scene, such as a short-term business peak around 13 o'clock and around 20 o'clock.
  • a synthetic scenario is a combination of two basic scenarios.
  • the embodiment of the invention provides a cloud application scaling method. As shown in FIG. 5, the method includes:
  • the cloud application expansion device obtains the load of the VNF system and the number of virtual machine VMs.
  • the cloud application scaling device determines whether it is necessary to increase/decrease the number of VMs, the cloud application scaling device first acquires the load of the VNF system and the number of virtual machine VMs.
  • the load of the VNF system and the number of virtual machine VMs are the load of the first time of the VNF system and the number of virtual machine VMs, and the first moment can be the current time.
  • the following embodiments of the present invention are described by taking the load of the VNF system and the number of virtual machine VMs as the load of the first time of the VNF system and the number of VMs.
  • the cloud application expansion device can periodically acquire the load of the VNF system and the number of virtual machine VMs according to a preset time.
  • the cloud application scaling device may also obtain a Service Level Agreement (SLA) indicator, where the SLA indicator is a performance defined between the service provider and the user to ensure the performance and reliability of the service under a certain overhead.
  • SLA Service Level Agreement
  • the method provided by the embodiment of the present invention may be applied to the service load and the resource load of the cloud application system, and the reported load may be different for different application scenarios. There is no limit to this.
  • the cloud application expansion device calculates the load and the number of VMs, and executes The first benefit value of the VNF system that can be obtained by each action in the action set, wherein each action in the action set indicates an increase or decrease in the number of VMs.
  • the cloud application retracting device may internally store a cumulative return table, where the cumulative return table includes a fourth revenue value of the VNF system that can be obtained by the cloud application retracting device in each state in the action group, and the fourth return value is The cloud application retracting device performs the historical return value to the VNF system that can be obtained by each action in the action set in each state, and the historical benefit value may be a training result or a stored historical value.
  • the state of the cloud application expansion device refers to the load corresponding to the cloud application expansion device in the case of various VM numbers; the action set includes at least one action, and the action is to increase or decrease the number of VMs, for example, adding one VM, adding two VM, etc.
  • each row of data in Table 1 represents a fourth benefit value of the VNF system obtained by the cloud application expansion device performing an action in each state, and an action of adding two VMs is taken as an example, and can be seen in the cloud application.
  • the fourth benefit value for the VNF system obtained by performing the action of adding two VMs is different.
  • the method for performing the first revenue value of the VNF system that can be obtained by each action in the action set may include, but is not limited to, the following methods:
  • the fourth benefit value of the system, and the fourth benefit value of the VNF system that can be obtained by each action is set as the first benefit value of the VNF system that can be obtained by the action;
  • the cloud application expansion device calculates the first revenue value of the VNF system that can be obtained by performing each action in the action set in the case of the load and the number of VMs at the first moment, and may also update the cumulative reward table.
  • the cloud application expansion device obtains a first action, where the first action is an action that can obtain the maximum value of the first benefit to the VNF system.
  • the cloud application expansion device calculates the load at the first time and the number of VMs at the first time, after performing the first benefit value of the VNF system that can be obtained by each action in the action set, the selected VNF is selected.
  • the first action of the system's first benefit value is the largest.
  • the cloud application expansion device performs the first action.
  • the cooling time can include: expansion cooling time and shrinkage cooling time, and expansion cooling time and shrinkage cooling time are respectively timed. That is, in the cloud application After the first action is performed, the timer starts to count. After a capacity expansion, the expansion cooling time timer is triggered. After the expansion cooling time is reached, the capacity expansion is allowed again. The timer reset restarts, and the expansion request generated in the middle will be It is ignored. Similarly, after a shrinking, the shrinkage cooling time timer will be triggered until the shrinking cooling time is reached, and the re-contraction is allowed again. The timer reset restarts and the resulting shrink request will be ignored.
  • the expansion cooling time and the shrinkage cooling time can also be timed simultaneously. That is, after the cloud application expansion device performs the first action, the timer starts to count, and after a capacity expansion/reduction, the cooling time timer is triggered until the cooling time is reached, then the capacity expansion/retraction is allowed again, and the timer reset is restarted. Timing, the expansion/reduction request generated in the middle will be ignored.
  • the cloud application scaling device calculates the first revenue value of the VNF system that can be obtained by performing various actions in the action set in the case of the load and the number of VMs, and acquires the The first action that can obtain the largest first value of the VNF system, the first action is the best result of the virtual machine expansion/contraction, which can reduce the unnecessary expansion/contraction operation and greatly alleviate the ping-pong effect. It also improves the utilization of resources, improves the reliability of the system, and reduces the risk of the VNF system violating the Service-Level Agreement (SLA). It is especially suitable for application scenarios that require high SLA guarantee.
  • SLA Service-Level Agreement
  • the method further includes steps S105-S107:
  • the cloud application expansion device obtains a first prediction result, where the first prediction result is a predicted second time load, and the second time is later than the first time.
  • step S105 may specifically include S105a:
  • the S105a and the cloud application expansion device calculate the first prediction result according to the load at the first moment of the VNF system and a preset algorithm.
  • the cloud application scaling method is forward-looking.
  • the cloud application scaling method provided by the embodiment of the present invention can also consider the future load trend when calculating the capacity expansion/reduction result.
  • the cloud application expansion device calculates the load according to the first moment and a preset algorithm.
  • the first prediction result is calculated, wherein the first prediction result is a predicted second time load, and the first time is earlier than the second time.
  • the method for the cloud application scaling device to calculate the first prediction result may include, but is not limited to, the following two methods:
  • the first prediction result is directly calculated. It can be understood that the first prediction result here is an accurate determination compared with the calculation result of the method (1). value.
  • the cloud application expansion device calculates a second benefit value for the VNF system that can be obtained by performing each action in the action set in the case of the first prediction result and the number of VMs.
  • the second income value herein refers to the second income value that can be obtained by performing each action in the action set in the case of the predicted load and the number of VMs at the second time.
  • the cloud application expansion device calculates the third benefit value of the VNF system that can be obtained by each action according to the second benefit value of the VNF system and the first benefit value of the VNF system that can be obtained by each action. Income value.
  • the first revenue value may be any one of the first revenue values calculated by the three methods provided in the foregoing step S102.
  • the method for calculating the third benefit value that can be obtained by each action may include, but is not limited to, The obtained second income value and the first income value that can be obtained by each action are added together, the two are multiplied, or the two are weighted, and the like, which is not limited by the present invention.
  • steps S103 and S104 may specifically include steps S103a and S104a:
  • the cloud application expansion device acquires a second action, wherein the second action is an action
  • the action that can be obtained by concentration is the third largest value of the VNF system.
  • the cloud application expansion device performs a second action.
  • step S108 after execution of step S103a, and before execution of step S104a, the method further includes step S108:
  • the cloud application expansion device confirms that the second action is allowed to be executed.
  • the cloud application expansion device needs to determine whether to perform the capacity expansion/reduction operation according to the scheduling criterion, and the scheduling criterion may be that the capacity expansion/reduction capacity may be expanded/reduced again. Whether to increase the VM when the load increases, whether to reduce the VM when the load is reduced, whether the capacity expansion may cause waste of resources, whether the shrinkage may cause SLA violation, and the like.
  • step S103 the process that the cloud application expansion device confirms that the first action is allowed to be executed is similar to the step S108, and is not described herein again for brevity.
  • step S104a if the scheduling criterion determines that the expansion/reduction operation is finally allowed to be performed, step S104a is continued; if the scheduling criterion determines that the expansion/reduction operation is not allowed to be performed, the process of the present invention is terminated.
  • the cloud application expansion device stores configuration information, and the configuration information can be updated.
  • the update method includes:
  • the cloud application expansion device confirms that the configuration information needs to be updated.
  • the cloud application expansion device can be deployed in the VNF or in the VNFM, the actual execution of the configuration information may be VNF or VNFM, and the present invention is not limited. .
  • the configuration information may include but is not limited to the following information:
  • the SLA indicator and the corresponding SLA threshold are used to select a new SLA indicator.
  • the current usage rate is the throughput, and the configuration is changed to the delay.
  • the SLA threshold is used as the system performance benchmark and is also used to calculate the current In the state, select an SLA performance part of the action immediate return;
  • the system maximum/minimum VM number keeps the system VM number between the minimum VM number and the maximum VM number, and the minimum VM number is the minimum number of VMs required by the system to maintain the service, and
  • the maximum number of VMs represents the resources that the system can provide to the maximum extent possible.
  • the maximum number of VMs added/deleted specifies the range of actions that can be selected for one expansion/reduction; the maximum PDP that the system can support Number/user number (or similar business load indicator, other data similar to the number of users, etc.), when the service load indicator is an elastic expansion decision indicator, the parameter specifies the maximum load that the system can support, usually the parameter and system
  • the maximum number of VMs corresponds to; the maximum load burst, including the aspect of the traffic load indicator and the performance compliance indicator, represents the maximum load burst that the system can handle, and is used to quantify the load burst, so that the system can cope with various Load burst; quantify the number of particles, including resource utilization and load burst. Since the state set includes resource usage, VM number, and load burst, the number of states will be too large, which will seriously affect the performance of the algorithm. The number of particles can effectively control the number of states.
  • the elastic scaling decision algorithm supports the use of the business load indicator, the separate use of the resource load indicator, the integrated service load indicator, and the resource load indicator for capacity expansion/reduction.
  • the user can Flexible configuration. When an metric is configured for unsatisfactory decision-making, you can configure other metrics for elastic scaling decision to achieve the purpose of securing services. For example, the current CPU usage is used for decision making, and now reconfigured to use random access memory. (RAM, Random Access Memory) Make a decision, or configure it as a user.
  • RAM Random Access Memory
  • the cloud application expansion device sends configuration information update request information to the EMS.
  • the EMS receives the configuration information update request information sent by the cloud application expansion device.
  • the EMS sends the configuration information update request information to the OSS.
  • OSS herein may refer to an operation and maintenance personnel operating the OSS, and may also refer to the OSS itself.
  • the OSS receives configuration information update request information sent by the EMS.
  • the OSS sends configuration information update feedback information to the EMS, where the configuration information update feedback information includes configuration information.
  • the EMS receives the configuration information update feedback information sent by the OSS.
  • the EMS sends the configuration information update feedback information to the cloud application expansion device.
  • the cloud application expansion device receives configuration information update feedback information sent by the EMS.
  • the cloud application expansion device updates configuration information.
  • the cloud application expansion device sends update completion information to the EMS.
  • the EMS receives the update completion information sent by the cloud application expansion device.
  • the EMS sends the update completion information to the OSS.
  • the OSS receives the update completion information sent by the EMS.
  • steps S211-S214 are optional steps, and steps S211-S214 may not be performed during a specific execution process.
  • the embodiment of the invention provides a cloud application scaling method, which acquires the load of the virtual network function VNF system and the number of virtual machine VMs; and calculates the pairs that can be obtained by performing various actions in the action set in the case of the load and the number of VMs.
  • a first revenue value of the VNF system wherein each action in the action set indicates an increase or decrease in the number of VMs; and the first action is obtained and executed, wherein the first action is a first benefit value of the VNF system that can be obtained The biggest action.
  • the cloud application expansion device calculates the first revenue value of the VNF system that can be obtained by performing various actions in the action set in the case of the load and the number of VMs, and acquires the pair of VNFs that can be obtained.
  • the first action with the largest first benefit value of the system the first action is the best result of virtual machine expansion/contraction, which can reduce unnecessary expansion/contraction operations, greatly alleviate the ping-pong effect, and improve resources.
  • the utilization rate improves the reliability of the system and reduces the risk of the VNF system violating the SLA. It is especially suitable for application scenarios that require high SLA guarantee.
  • the embodiment of the present invention provides a cloud application expansion device.
  • the cloud application expansion device is configured to execute the steps performed by the cloud application expansion device in the above method.
  • the cloud application expansion device may include modules corresponding to the corresponding steps.
  • the cloud application scaling device may include an obtaining module 10, a computing module 11 and an executing module 12.
  • the obtaining module 10 is configured to acquire the load and virtuality of the virtual network function VNF system The number of VMs.
  • the calculation module 11 is configured to calculate, after the acquisition module 10 acquires the load and the number of VMs of the VNF system, the first revenue value of the VNF system that can be obtained by performing each action in the action set in the case of the load and the number of VMs. Wherein, each action in the action set indicates an increase or decrease in the number of VMs.
  • the obtaining module 10 is further configured to: after the calculating module 11 calculates the first revenue value of the VNF system that can be obtained by each action in the action set, in the case that the load and the number of VMs are calculated, the first action is obtained, where An action is the action that maximizes the first benefit value of the VNF system that can be obtained.
  • the executing module 12 is configured to perform the first action after the obtaining module 10 acquires the first action.
  • the load and the number of VMs of the VNF system are the load and the number of VMs at the first moment of the VNF system.
  • the obtaining module 10 is further configured to obtain a first prediction result before the obtaining module 10 obtains the first action, where the first prediction result is a predicted second time load, and the second time is later than the first time.
  • the calculation module 11 is further configured to: after the obtaining the first prediction result by the obtaining module 10, calculate a second benefit value of the VNF system that can be obtained by performing each action in the action set in the case of the first prediction result and the number of VMs And calculating a third benefit value for the VNF system that can be obtained by each action according to the second benefit value of the VNF system and the first benefit value of the VNF system that can be obtained by each action.
  • the obtaining module 10 is specifically configured to obtain a second action, wherein the second action is an action that can obtain the third benefit value that is the largest for the VNF system.
  • the executing module 12 is specifically configured to perform the second action after the obtaining module 10 acquires the second action.
  • the acquiring device 10 is configured to calculate a first prediction result according to a load of the VNF system and a preset algorithm.
  • the cloud application expansion device further includes a receiving module 13 .
  • the receiving module 13 is configured to receive configuration information sent by the network element management system EMS.
  • the configuration information includes at least a preset algorithm.
  • the cloud application expansion device further includes a confirmation module 14.
  • the confirmation module 14 is configured to confirm that the first action is allowed to be executed after the obtaining module 10 acquires the first action.
  • the cloud application expansion apparatus of the embodiment may correspond to the cloud application expansion apparatus in the cloud application expansion method of the embodiment of any one of the foregoing 5 to FIG. 9, and the cloud application expansion apparatus of the embodiment
  • the division and/or function of each module in the process is to implement the method flow shown in any one of FIG. 5 to FIG. 9.
  • no further details are provided herein.
  • the functions of the obtaining module 10, the calculating module 11, the executing module 12, and the confirming module 14 may be implemented by a processor, and the function of the receiving module 13 may be implemented by a receiver.
  • the embodiment of the invention provides a cloud application expansion device. Based on the description of the above embodiment, since the cloud application expansion device calculates the first revenue value of the VNF system that can be obtained by performing various actions in the action set in the case of the load and the number of VMs, and acquires the pair of VNFs that can be obtained.
  • the first action with the largest first benefit value of the system the first action is the best result of virtual machine expansion/contraction, which can reduce unnecessary expansion/contraction operations, greatly alleviate the ping-pong effect, and improve resources.
  • the utilization rate improves the reliability of the system and reduces the risk of the VNF system violating the SLA. It is especially suitable for application scenarios that require high SLA guarantee.
  • the embodiment of the present invention further provides a cloud application expansion device.
  • the cloud application expansion device includes a memory 20, a processor 21, a communication interface 22, and a system bus 23.
  • the memory 20, the processor 21 and the communication interface 22 are connected by a system bus 23 for storing some computer instructions, and the processor 21 is configured to execute computer instructions to enable the cloud application expansion device to perform any one of FIGS. 5-9.
  • Cloud application scaling method For the specific cloud application scaling method, refer to the foregoing as shown in any one of FIG. 5 to FIG. Related descriptions in the embodiments are not described herein again.
  • the processor 21 can be a central processing unit (CPU).
  • the processor 21 can also be other general purpose processors, digital signal processing (DSP), application specific integrated circuit (ASIC), field-programmable gate array (FPGA) or Other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc.
  • the general purpose processor may be a microprocessor or the processor or any conventional processor or the like.
  • the processor 21 may be a dedicated processor, which may include at least one of a baseband processing chip, a radio frequency processing chip, and the like. Further, the dedicated processor may also include a chip having other dedicated processing functions of the cloud application scaling device.
  • the memory 20 may include a volatile memory such as a random-access memory (RAM); the memory 20 may also include a non-volatile memory such as a read-only memory (read) -only memory, ROM), flash memory, hard disk drive (HDD) or solid-state drive (SSD); the memory 20 may also include a combination of the above types of memories.
  • RAM random-access memory
  • ROM read-only memory
  • HDD hard disk drive
  • SSD solid-state drive
  • the system bus 23 can include a data bus, a power bus, a control bus, and a signal status bus. For the sake of clarity in the present embodiment, various buses are illustrated as the system bus 23 in FIG.
  • Communication interface 22 may include a receiver and a transmitter.
  • the receiver and the transmitter may be specifically a transceiver on the cloud application expansion device.
  • the transceiver can be a wireless transceiver.
  • each step in the method flow shown in any of the above-mentioned FIG. 5 to FIG. 9 can be implemented by hardware execution of a computer-executed instruction in the form of software. To avoid repetition, we will not repeat them here.
  • the embodiment of the invention provides a cloud application expansion device. Based on the description of the above embodiment, since the cloud application expansion device calculates the first revenue value of the VNF system that can be obtained by performing various actions in the action set in the case of the load and the number of VMs, and acquires the The first action that can obtain the largest first value of the VNF system, the first action is the best result of the virtual machine expansion/contraction, which can reduce the unnecessary expansion/contraction operation and greatly alleviate the ping-pong effect. It also improves the utilization of resources, improves the reliability of the system, and reduces the risk of the VNF system violating the SLA. It is especially suitable for application scenarios that require high SLA guarantee.
  • the embodiment of the invention further provides a software product, which may include computer instructions for implementing a cloud application scaling method.
  • the computer instructions can be stored on a readable storage medium; the processor can read and execute the computer instructions from the readable storage medium such that the processor implements the cloud application scaling method.
  • the disclosed apparatus and method may be implemented in other manners.
  • the device embodiments described above are merely illustrative.
  • the division of the modules or units is only a logical function division.
  • there may be another division manner for example, multiple units or components may be used. Combinations can be integrated into another system, or some features can be ignored or not executed.
  • the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interface, device or unit, and may be in an electrical, mechanical or other form.
  • the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Can be based on reality It is necessary to select some or all of the units to achieve the purpose of the solution of the embodiment.
  • each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
  • the above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.
  • the integrated unit if implemented in the form of a software functional unit and sold or used as a standalone product, may be stored in a computer readable storage medium.
  • the technical solution of the present invention which is essential or contributes to the prior art, or all or part of the technical solution, may be embodied in the form of a software product stored in a storage medium.
  • a number of instructions are included to cause a computer device (which may be a personal computer, server, or network device, etc.) or a processor to perform all or part of the steps of the methods described in various embodiments of the present invention.
  • the foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and the like. .

Landscapes

  • Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Debugging And Monitoring (AREA)

Abstract

本发明实施例提供一种云应用伸缩方法及装置，涉及通信领域，能够分析得出对VNF系统的第一收益值最大的虚拟机扩容/缩容结果，从而提高了资源的利用率，提升系统的可靠性。该云应用伸缩方法包括：获取虚拟网络功能VNF系统的负载和虚拟机VM个数；计算在负载和VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第一收益值，其中，动作集中的各个动作均指示增加或者减少VM的个数；获取并执行第一动作，其中，第一动作为所能获得的对VNF系统的第一收益值最大的动作。

Description

一种云应用伸缩方法及装置 技术领域
本发明涉及通信领域，尤其涉及一种云应用伸缩方法及装置。
背景技术
网络功能虚拟化(Network Functions Virtualization，NFV)是一种将传统的演进的分组核心(Evolved Packet Core，EPC)网元部署在云网络中，使得NFV能够利用计算机服务器及虚拟化技术来承载很多功能的软件处理的技术，通过软硬件解耦及功能抽象，使网络设备的功能不依赖于专用硬件，不仅降低了设备成本，还能实现资源的充分灵活共享及新业务的快速开发和部署。
通常的，一个虚拟网络功能(Virtual Network Function，VNF)系统可以由多个虚拟机(Virtual Machine，VM)组成，用以支持VNF系统上运行的业务。同时，VM的数量也会随着业务的变化而变化(如当业务量增大的时候VM的数量会增加，当业务量减小的时候VM的数量会减少)，这种调整过程被称为扩容/缩容，也可以称为云应用伸缩。现有的云应用伸缩方法为：首先设定系统负载的上限阈值和下限阈值，当系统负载超过预先设定的上限阈值时，执行与此相对应的动作(如增加一个VM)；类似的，当系统的负载低于预先设定的下限阈值时，执行与此相应的动作(如删除一个VM)。
然而，现有的云应用伸缩方法中，上限阈值和下线阈值都是预先设定好的，在用户进行变更之前无法更改，并且当系统负载超过预先设定的上限阈值或者低于预先设定的下限阈值时，增加或者减少的VM的个数只能是固定值，无法根据系统负载的变化灵活调整，造成资源的利用率低，系统可靠性下降的问题。
发明内容
本发明的实施例提供一种云应用伸缩方法及装置，能够分析得出对VNF系统的第一收益值最大的虚拟机扩容/缩容策略，从而提 高了资源的利用率，提升系统的可靠性。
为达到上述目的，本发明的实施例采用如下技术方案：
第一方面，本发明实施例提供一种云应用伸缩方法，包括：
首先，云应用伸缩装置获取虚拟网络功能VNF系统的负载和虚拟机VM个数；其次，云应用伸缩装置计算在负载和VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第一收益值，动作集中的各个动作均指示增加或者减少VM的个数；最后，云应用伸缩装置获取并执行所能获得的对VNF系统的第一收益值最大的第一动作。
本发明实施例提供的云应用伸缩方法中，由于云应用伸缩装置计算在负载和VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第一收益值，并获取所能获得的对VNF系统的第一收益值最大的第一动作，第一动作是虚拟机扩容/缩容的最佳结果，能够减少不必要的扩容/缩容操作，极大地缓解了乒乓效应，并且提高了资源的利用率，提升系统的可靠性，减少了VNF系统违反服务等级协议(Service-Level Agreement，SLA)的风险，尤其适合需要高SLA保障的应用场景。
进一步地，VNF系统的负载和VM个数为VNF系统第一时刻的负载和VM个数。
在获取并执行第一动作前，方法还包括：
首先，云应用伸缩装置获取预测到的第二时刻的负载，第二时刻晚于第一时刻；其次，云应用伸缩装置计算在第一预测结果和VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第二收益值；最后，云应用伸缩装置根据各个动作所能获得的对VNF系统的第二收益值和各个动作所能获得的对VNF系统的第一收益值，计算各个动作所能获得的对VNF系统的第三收益值。
获取并执行第一动作，具体包括：
云应用伸缩装置获取并执行所能获得的对VNF系统的第三收益值最大的第二动作。
本发明实施例提供的云应用伸缩方法中，为了保证NFV系统的稳定性，使得云应用伸缩方法具有一定的前瞻性，本发明实施例提供的云应用伸缩方法还可以在计算扩容/缩容结果时考虑晚于第一时刻的第二时刻的负载走势。
进一步地，获取第一预测结果，具体包括：
云应用伸缩装置根据VNF系统的负载和预设算法，计算第一预测结果。
进一步地，方法还包括：
云应用伸缩装置接收网元管理系统EMS发送的至少包括预设算法的配置信息。
可选的，配置信息至少包括SLA指标及相应的SLA阈值；系统最大/最小VM数；增加/删除的VM的最大数目；系统所能支持的最大PDP数/用户数；最大负载突发量；量化颗粒数。
可选的，在获取第一动作后，方法还包括：
云应用伸缩装置确认第一动作允许执行。
本发明实施例提供的云应用伸缩方法中，云应用伸缩装置能够根据调度准则判断最终是否执行扩容/缩容动作。
第二方面，本发明实施例提供一种云应用伸缩装置，包括获取模块，计算模块和执行模块。
获取模块，用于获取虚拟网络功能VNF系统的负载和虚拟机VM个数；计算模块，用于在获取模块获取VNF系统的负载和VM个数后，计算在负载和VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第一收益值，其中，动作集中的各个动作均指示增加或者减少VM的个数；获取模块，还用于在计算模块计算在负载和VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第一收益值后，获取第一动作，其中，第一动作为所能获得的对VNF系统的第一收益值最大的动作；执行模块，用于在获取模块获取第一动作后，执行第一动作。
本发明实施例提供的云应用伸缩装置的技术效果可以参见上述 第一方面云应用伸缩装置执行的云应用伸缩方法中描述的云应用伸缩装置的技术效果，此处不再赘述。
进一步地，VNF系统的负载和VM个数为VNF系统第一时刻的负载和VM个数。
获取模块，还用于在获取模块获取第一动作前，获取第一预测结果，其中，第一预测结果为预测到的第二时刻的负载，第二时刻晚于第一时刻；计算模块，还用于在获取模块获取第一预测结果后，计算在第一预测结果和VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第二收益值；并根据各个动作所能获得的对VNF系统的第二收益值和各个动作所能获得的对VNF系统的第一收益值，计算各个动作所能获得的对VNF系统的第三收益值；获取模块，具体用于获取第二动作，其中，第二动作为所能获得的对VNF系统的第三收益值最大的动作；执行模块，具体用于在获取模块获取第二动作后，执行第二动作。
进一步地，获取装置，具体用于根据VNF系统的负载和预设算法，计算第一预测结果。
进一步地，云应用伸缩装置还包括接收模块。
接收模块，用于接收网元管理系统EMS发送的配置信息，其中，配置信息至少包括预设算法。
进一步地，确认模块，用于在获取模块获取第一动作后，确认第一动作允许执行。
第三方面，本发明实施例还提供一种云应用伸缩装置，云应用伸缩装置包括存储器、处理器、通信接口和系统总线。
存储器、处理器和通信接口通过系统总线连接，存储器用于存储计算机指令，处理器用于执行存储器存储的计算机指令，以使云应用伸缩装置执行如上述第一方面的云应用伸缩方法。
本发明实施例提供的云应用伸缩装置的技术效果可以参见上述第一方面云应用伸缩装置执行的云应用伸缩方法中描述的云应用伸缩装置的技术效果，此处不再赘述。
第四方面，本发明实施例还提供一种软件产品，软件产品包括实现云应用伸缩方法的计算机指令。
计算机指令可以存储在可读存储介质上；处理器可以从该可读存储介质上读取到计算机指令并执行，使得处理器实现云应用伸缩方法。
本发明实施例提供一种云应用伸缩方法及装置，通过获取虚拟网络功能VNF系统的负载和虚拟机VM个数；计算在负载和VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第一收益值，其中，动作集中的各个动作均指示增加或者减少VM的个数；获取并执行第一动作，其中，第一动作为所能获得的对VNF系统的第一收益值最大的动作。基于上述实施例的描述，由于云应用伸缩装置计算在负载和VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第一收益值，并获取所能获得的对VNF系统的第一收益值最大的第一动作，第一动作是虚拟机扩容/缩容的最佳结果，能够减少不必要的扩容/缩容操作，极大地缓解了乒乓效应，并且提高了资源的利用率，提升系统的可靠性，减少了VNF系统违反SLA的风险，尤其适合需要高SLA保障的应用场景。
附图说明
为了更清楚地说明本发明实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的一些实施例。
图1为现有技术中的一种NFV系统架构图；
图2为本发明实施例提供的一种NFV系统架构图；
图3为本发明实施例提供的另一种NFV系统架构图；
图4为本发明实施例提供的一种云应用伸缩装置的硬件结构图；
图5为本发明实施例提供的一种云应用伸缩方法的流程示意图一；
图6为本发明实施例提供的一种云应用伸缩方法的流程示意图 二；
图7为本发明实施例提供的一种云应用伸缩方法的流程示意图三；
图8为本发明实施例提供的一种云应用伸缩方法的流程示意图四；
图9为本发明实施例提供的一种云应用伸缩装置里存储的配置信息的更新方法的流程示意图；
图10为本发明实施例提供的一种云应用伸缩装置的结构示意图一；
图11为本发明实施例提供的一种云应用伸缩装置的结构示意图二；
图12为本发明实施例提供的一种云应用伸缩装置的结构示意图三；
图13为本发明实施例提供的一种云应用伸缩装置的硬件示意图。
具体实施方式
下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本发明一部分实施例，而不是全部的实施例。
另外，本文中术语“系统”和“网络”在本文中常被可互换使用。本文中术语“和/或”，仅仅是一种描述关联对象的关联关系，表示可以存在三种关系，例如，A和/或B，可以表示：单独存在A，同时存在A和B，单独存在B这三种情况。另外，本文中字符“/”，一般表示前后关联对象是一种“或”的关系。
需要说明的是，本发明的技术方案可以应用于NFV系统中。如图1所示，为现有技术中的一种NFV系统架构图，包括：管理编排域(Management and Orchestrator，MANO)和业务网络域，其中，业务网络域包括基础设施层(NFV Infrastructure，NFVI)、虚拟网络层和运营支撑层。
MANO：负责对整个NFVI资源的管理和编排，业务网络和NFVI资源的映射与关联，运营支持系统(Office of Strategic Services，OSS)业务资源流程的实施等。MANO内部分为虚拟网络功能协调(NFV Orchestrator，NFVO)，虚拟网络功能管理(VNF Manager，VNFM)和虚拟基础设施管理(Virtual Infrastructure Management，VIM)三个实体，分别完成对业务网络域的运营支撑层、虚拟网络层和NFVI的管理。
NFVI：从云计算的角度看，NFVI类似一个资源池。NFVI映射到物理基础设施就是分散在多个地理位置上的数据中心，数据中心之间通过高速通信网络互联。NFVI将物理资源通过虚拟化技术转换为虚拟资源，并提供VNF需要的所有基础设施资源。
虚拟网络层：对应各个电信业务网络，电信业务网络中的每个物理网元映射为单个VNF，VNF所需要的资源分解为虚拟的计算/存储/网络等资源，由NFVI提供支撑，VNF之间的接口采用传统网络定义的信令接口(例如3GPP+ITU-T接口)，VNF的业务网管采用网元-网元管理系统-网络管理系统(NE-EMS-NMS)体制。VNF是传统网元的纯软件实现，部署在多个互联的虚拟机上。VNF内部的多个VM通过内部网络互联，内部网络在VNF外部是不可见的。VNF与其他VNF之间通过外部网络互联。
运营支撑层：是目前的OSS/业务支持系统(Business support system，BSS)。主要用于计费、服务保障与服务实施三个领域。
本发明实施例提供的NFV系统架构图如图2所示，包括网元管理系统(Element Management System，EMS)、VNF、NFVO、VNFM、VIM和云应用伸缩装置。其中，云应用伸缩装置可以部署在VNF内部，也可以部署在VNFM内部，其中，图2所示的NFV系统架构图是云应用伸缩装置部署在VNF内部的情况，云应用伸缩装置部署在VNFM内部的情况如图3所示。
具体的，VNF：是虚拟网元，由部署在一个或多个VM上的多个组件组成，将传统网元中的网络功能虚拟化，包括分组数据网络 网关(Packet Data Network GateWay，P-GW)，分组数据网络网关通常被称PDN网关，服务网关(Serving GateWay，S-GW)，网络节点(Mobility Management Entity，MME)，归属签约用户服务器(Home Subscriber Server，HSS)等。同时，VNF还包括监控器(Monitor)。
Monitor：负责监控与收集组成VNF的各个VM的资源负载和业务负载，其中资源负载可以是中央处理器(Central Processing Unit，CPU)利用率、内存利用率、带宽利用率、磁盘利用率等与计算/存储/网络等资源相关的指标；业务负载可以是当前系统中的用户数、会话数、分组数据协议(Packet Data Protocol，PDP)上下文数、数据库读写队列长度、访问时延等与业务相关的指标。每一种负载的取值可以是瞬时值，一定时窗(时间窗长度根据需求确定)内的平均值、最大值、最小值、中间值等。采集到的数据将周期性的发送给云应用伸缩装置用以决策，数据可以是资源负载和/或业务负载。Monitor是VNF的内部实现，实际中可能由其他VNF功能组件完成Monitor的功能。
EMS：网元管理系统，管理特定类型的一个或多个网元，负责每个网元的功能和容量。
NFVO：通过多个VIM管理NFVI虚拟基础设施资源，实现资源编排功能，网络服务的生命周期管理。
VNFM：负责VNF实例的生命周期管理。一个VNFM可以关联到单个VNF实例，也可以关联到多个同类型或不同类型的VNF实例。无论自动伸缩模块部署在VNF中还是VNFM中，决策结果都会上报给VNFM。
VIM：根据云应用伸缩装置的决策结果执行相应的动作，如增加若干个VM或删除若干个VM，所增加或删除的VM个数由云应用伸缩装置指定。
云应用伸缩装置由6部分构成，如图4所示，分别是数据预处理模块、预测模块、特征提取模块、学习引擎模块、策略库模块及扩容/缩容选择模块。(1)数据预处理模块：对Monitor上报的资源 负荷与业务负荷进行预处理，消除高频噪声，提取出负载变化趋势，具体实施时可以通过数字滤波器进行平滑，也可以通过提取包络实现；(2)预测模块：使用时间序列分析方法进行负载预测，利用历史数据预测短期内的负载，可以使用的算法包括但不限于神经网络算法、支持向量回归算法、积分自回归移动平均算法、线性回归算法、指数平滑算法以及其他算法。负载预测值用于后续模型学习与自动伸缩动作选择；(3)特征提取模块：从Monitor上报的负载中提取出组成状态集的特征，包括资源使用率、当前VM数、负载变化趋势等；(4)学习引擎模块：利用机器学习算法，如增强学习，学习每个状态下的自动伸缩策略，通过收益函数对当前状态下选择不同的扩容/缩容动作进行评估，得到执行不同动作对VNF系统的收益，根据计算出来的结果对策略库模块中的内容进行更新；支持利用历史数据进行离线学习和利用实时数据进行在线学习；(5)策略库模块：保存扩容/缩容策略，即每一状态下选择一个扩容/缩容动作的收益；(6)扩容/缩容选择模块：根据策略库模块中的扩容/缩容策略与预测模块给出的负载预测值选择当前状态下回报最大的扩容/缩容动作。
可以理解的是，数据预处理模块的功能和预测模块的功能可以由下述实施例云应用伸缩装置的获取模块实现；特征提取模块的功能和学习引擎模块的功能可以由下述实施例云应用伸缩装置的计算模块实现；策略库模块的功能可以由下述实施例云应用伸缩装置的存储器实现；扩容/缩容选择模块的功能可以由下述实施例云应用伸缩装置的计算模块实现，本发明对此不做限制。
还需要说明的是，本发明实施例提供的方法可以适用于云应用系统的业务负荷和资源负荷，针对不同的应用场景分别进行扩容/缩容建模，即在不能获取资源负荷的场景使用适合业务负荷的云应用伸缩方法，在不能获取业务负荷的场景使用适合资源负荷的云应用伸缩方法，在两种负荷都能获取的情况根据需求配置使用任一种云应用伸缩方法。
进一步地，本发明实施例提供的方法还可以适用于基本场景和综合场景。其中，基本场景为日常忙时闲时业务场景或者突发业务场景。日常忙时闲时业务场景的主要特征是系统负载会随着时间呈现出明显的波峰与波谷，并且波峰与波谷的持续时间较长，如每天8点到23点为忙时，其他时间段为闲时；突发业务场景是在日常忙时闲时场景的基础上出现短时业务突发，如在13点左右、20点左右出现了短时的业务波峰。综合场景是两个基本场景的组合。
实施例1
本发明实施例提供一种云应用伸缩方法，如图5所示，该方法包括：
S101、云应用伸缩装置获取VNF系统的负载和虚拟机VM个数。
当云应用伸缩装置判断是否需要增加/减少VM个数时，云应用伸缩装置首先会获取VNF系统的负载和虚拟机VM个数。通常的，VNF系统的负载和虚拟机VM个数为VNF系统第一时刻的负载和虚拟机VM个数，第一时刻可以为当前时刻。为了便于理解，本发明下述实施例均以VNF系统的负载和虚拟机VM个数为VNF系统第一时刻的负载和虚拟机VM个数为例进行说明。
可以理解的是，为了保证云应用伸缩装置的性能，云应用伸缩装置能够根据预设的时间，周期性地获取VNF系统的负载和虚拟机VM个数。
进一步地，云应用伸缩装置还可以获取服务等级协议(Service-Level Agreement，SLA)指标，其中，SLA指标是在一定开销下为保障服务的性能和可靠性，服务提供商与用户间定义的一种双方认可的协定，通常这个开销是驱动提供服务质量的主要因素。
可以理解的是，根据不同的应用需求，由于本发明实施例提供的方法可以适用于云应用系统的业务负荷和资源负荷，针对不同的应用场景，上报的负荷也会相应有所不同，本发明对此不做限制。
S102、云应用伸缩装置计算在负载和VM个数的情况下，执行 动作集中的各个动作所能获得的对VNF系统的第一收益值，其中，动作集中的各个动作均指示增加或者减少VM的个数。
云应用伸缩装置内部可以存储有累积回报表，其中，累积回报表包括云应用伸缩装置在各个状态下执行动作集中的各个动作所能获得的对VNF系统的第四收益值，第四收益值为云应用伸缩装置在各个状态下执行动作集中的各个动作所能获得的对VNF系统的历史收益值，该历史收益值可以是训练得出的，也可以是存储的历史值。云应用伸缩装置的状态是指云应用伸缩装置在各种VM个数的情况下对应的负载；动作集包括至少一个动作，动作为增加或者减少VM的个数，例如增加一个VM、增加两个VM等。
示例性的，累积回报表如表1所示：
表1
Figure PCTCN2016086267-appb-000001
具体的，表1中的每一行数据代表云应用伸缩装置在各个状态下执行一个动作所获得的对VNF系统的第四收益值，以增加2个VM的动作为例，可以看出在云应用伸缩装置的各个状态下，执行增加2个VM的动作所获得的对VNF系统的第四收益值不同。
云应用伸缩装置计算在第一时刻的负载和VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第一收益值的方法可以包括但不限于以下几种方式：
(1)：直接在累积回报表中查找到第一时刻对应的负载和VM个数，并获取该负载和VM个数对应的那一列数据，即执行动作集中的各个动作所能获得的对VNF系统的第四收益值，并将每个动作所能获得的对VNF系统的第四收益值设定为该动作所能获得的对VNF系统的第一收益值；
(2)：根据预设的收益值计算方法计算在第一时刻的负载和VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第一收益值；
(3)：结合上述方法(1)和方法(2)，并对方法(1)和方法(2)获得的两个收益值做运算(该运算可以包括但并不局限于两者相加，两者相乘，或者两者做加权运算等)，得到对VNF系统的第一收益值。
可选的，云应用伸缩装置计算在第一时刻的负载和VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第一收益值后，还可以更新累积回报表中第一时刻的负载和第一时刻的VM个数的情况下的对VNF系统的第四收益值，以供后续计算使用。
S103、云应用伸缩装置获取第一动作，其中，第一动作为所能获得的对VNF系统的第一收益值最大的动作。
云应用伸缩装置计算在第一时刻的负载和第一时刻的VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第一收益值后，选取所能获得的对VNF系统的第一收益值最大的第一动作。
S104、云应用伸缩装置执行第一动作。
需要补充的是，在云应用伸缩方法中还可以引入必要的冷却时间来保证系统的可靠性，冷却时间可以包括：扩容冷却时间和缩容冷却时间，扩容冷却时间与缩容冷却时间分别计时。即在云应用伸 缩装置执行完第一动作后，计时器开始计时，一次扩容之后会触发扩容冷却时间计时器，直至达到扩容冷却时间后才允许再次扩容，计时器复位重新开始计时，中间产生的扩容请求将会被忽略；同理，一次缩容之后会触发缩容冷却时间计时器，直至达到缩容冷却时间后才允许再次缩容，计时器复位重新开始计时，中间产生的缩容请求将会被忽略。可选的，扩容冷却时间与缩容冷却时间也能够同时计时。即在云应用伸缩装置执行完第一动作后，计时器开始计时，在一次扩容/缩容之后会触发冷却时间计时器，直至达到冷却时间后才允许再次扩容/缩容，计时器复位重新开始计时，中间产生的扩容/缩容请求将会被忽略。
本发明实施例提供的云应用伸缩方法中，由于云应用伸缩装置计算在负载和VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第一收益值，并获取所能获得的对VNF系统的第一收益值最大的第一动作，第一动作是虚拟机扩容/缩容的最佳结果，能够减少不必要的扩容/缩容操作，极大地缓解了乒乓效应，并且提高了资源的利用率，提升系统的可靠性，减少了VNF系统违反服务等级协议(Service-Level Agreement，SLA)的风险，尤其适合需要高SLA保障的应用场景。
进一步地，如图6所示，在步骤S103执行之前，方法还包括步骤S105-S107：
S105、云应用伸缩装置获取第一预测结果，其中，第一预测结果为预测到的第二时刻的负载，第二时刻晚于第一时刻。
具体地，如图7所示，步骤S105具体可以包括S105a：
S105a、云应用伸缩装置根据VNF系统第一时刻的负载和预设算法，计算第一预测结果。
为了保证NFV系统的稳定性，使得云应用伸缩方法具有一定的前瞻性，本发明实施例提供的云应用伸缩方法还可以在计算扩容/缩容结果时考虑未来的负载走势。
具体地，云应用伸缩装置根据第一时刻的负载和预设算法，计 算第一预测结果，其中，第一预测结果为预测到的第二时刻的负载，且第一时刻早于第二时刻。
可选的，云应用伸缩装置计算第一预测结果的方法可以包括但不限于以下两种方式：
(1)：根据预设算法和第一时刻的负载，计算负载变化趋势，然后在负载变化趋势上读取第二时刻的负载值，即第一预测结果，可以理解的是，由于是在负载变化趋势是一个负载大致走势图，因此这里的第一预测结果为一个大概估算值；
(2)：根据预设算法和第一时刻的负载，直接计算第一预测结果，可以理解的是，这里的第一预测结果与方法(1)计算出的结果相比，是一个精确的定值。
S106、云应用伸缩装置计算在第一预测结果和VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第二收益值。
需要说明的是，这里的第二收益值是指在预测到的第二时刻的负载和VM个数的情况下，执行动作集中的各个动作所能获得的第二收益值。
S107、云应用伸缩装置根据各个动作所能获得的对VNF系统的第二收益值和各个动作所能获得的对VNF系统的第一收益值，计算各个动作所能获得的对VNF系统的第三收益值。
其中，第一收益值可以是通过上述步骤S102提供的三种方法计算出的任意一种第一收益值。
同时，根据各个动作所能获得的第二收益值和各个动作所能获得的第一收益值，计算各个动作所能获得的第三收益值的方法可以包括但并不局限于将各个动作所能获得的第二收益值和各个动作所能获得的第一收益值两者相加，两者相乘，或者两者做加权运算等，本发明对此并不做限制。
进一步地，若本发明实施例提供的云应用伸缩方法执行了步骤S105-S107，则步骤S103和S104具体可以包括步骤S103a和S104a：
S103a、云应用伸缩装置获取第二动作，其中，第二动作为动作 集中所能获得的对VNF系统的第三收益值最大的动作。
S104a、云应用伸缩装置执行第二动作。
进一步地，如图8所示，在步骤S103a执行之后，且在步骤S104a执行之前，方法还包括步骤S108：
S108、云应用伸缩装置确认第二动作允许执行。
需要说明的是，在云应用伸缩装置获取第二动作后，云应用伸缩装置需要根据调度准则判断最终是否执行扩容/缩容动作，调度准则可以是扩容/缩容后是否可能会再次扩容/缩容、是否在负载增加时增加VM、是否在负载减少时减少VM、扩容是否可能造成资源浪费、缩容是否可能造成SLA违反等。
还需要说明的是，在步骤S103执行之后，且在步骤S104执行之前，云应用伸缩装置确认第一动作允许执行的过程和步骤S108类似，为了简洁，此处不再赘述。
相应的，若调度准则判断最终允许执行扩容/缩容动作，则继续执行步骤S104a；若调度准则判断最终不允许执行扩容/缩容动作，则终止本发明流程。
还需要补充的是，本发明实施例中云应用伸缩装置里存储有配置信息，配置信息能够进行更新，如图9所示，该更新方法包括：
S201、云应用伸缩装置确认配置信息需要更新。
需要说明的是，由于云应用伸缩装置可以部署在VNF内部，也可以部署在VNFM内部，因此在实际的应用中，更新配置信息的执行主体可以是VNF，也可以是VNFM，本发明不做限制。
具体的，配置信息可以包括但不限于以下信息：
(1)：SLA指标及相应的SLA阈值，其作用是选择新的SLA指标，例如当前使用的是吞吐率，经过配置改成时延，SLA阈值是作为系统性能基准，也用于计算在当前状态下选择一个动作即时回报的SLA性能部分；系统最大/最小VM数，使系统VM数维持在最小VM数和最大VM数之间，最小VM数是系统维持服务所需要VM的最小数，而最大VM数则代表了系统最大限度所能提供的资源， 也用于计算在当前状态下选择一个动作即时回报的资源部分；增加/删除的VM的最大数目，规定了一次扩容/缩容所能选择的动作的取值范围；系统所能支持的最大PDP数/用户数(或类似业务负荷指标、与用户数类似的其他数据等)，以业务负荷指标为弹性伸缩决策指标时，该参数规定了系统所能支持的最大负载量，通常该参数与系统最大VM数相对应；最大负载突发量，包括业务负荷指标方面和性能符合指标方面，代表系统所能应对的最大负载突发，用于对负载突发量进行量化，使得系统能够应对各种负载突发；量化颗粒数，包含资源利用和负载突发量两方面，由于状态集中包含资源使用率、VM数、负载突发量，状态的数量过大将会严重影响算法性能，通过设定量化颗粒数可以有效控制状态的数量。
(2)：重新配置用来做决策的负载，弹性伸缩决策算法支持单独使用业务负荷指标、单独使用资源负荷指标、综合业务负荷指标和资源负荷指标做扩容/缩容决策，使用者可以根据需求灵活配置，当配置某个指标进行决策效果不理想时，可以配置使用其他的指标进行弹性伸缩决策已达到保障服务的目的，例如当前使用CPU使用率做决策，现在重新配置为使用随机存取存储器(RAM，Random Access Memory)做决策，或者配置为用户。
(3)：配置是否使用负载预测功能。
S202、云应用伸缩装置发送配置信息更新请求信息至EMS。
S203、EMS接收云应用伸缩装置发送的配置信息更新请求信息。
S204、EMS发送配置信息更新请求信息至OSS。
需要说明的是，这里的OSS可以指操作OSS的运维人员，也可以指OSS本身。
S205、OSS接收EMS发送的配置信息更新请求信息。
S206、OSS发送配置信息更新反馈信息至EMS，其中，配置信息更新反馈信息包括配置信息。
S207、EMS接收OSS发送的配置信息更新反馈信息。
S208、EMS发送配置信息更新反馈信息至云应用伸缩装置。
S209、云应用伸缩装置接收EMS发送的配置信息更新反馈信息。
S210、云应用伸缩装置更新配置信息。
S211、云应用伸缩装置发送更新完成信息至EMS。
S212、EMS接收云应用伸缩装置发送的更新完成信息。
S213、EMS发送更新完成信息至OSS。
S214、OSS接收EMS发送的更新完成信息。
需要说明的是，步骤S211-S214为可选步骤，在具体的执行过程中，可以不执行步骤S211-S214。
本发明实施例提供一种云应用伸缩方法，通过获取虚拟网络功能VNF系统的负载和虚拟机VM个数；计算在负载和VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第一收益值，其中，动作集中的各个动作均指示增加或者减少VM的个数；获取并执行第一动作，其中，第一动作为所能获得的对VNF系统的第一收益值最大的动作。基于上述实施例的描述，由于云应用伸缩装置计算在负载和VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第一收益值，并获取所能获得的对VNF系统的第一收益值最大的第一动作，第一动作是虚拟机扩容/缩容的最佳结果，能够减少不必要的扩容/缩容操作，极大地缓解了乒乓效应，并且提高了资源的利用率，提升系统的可靠性，减少了VNF系统违反SLA的风险，尤其适合需要高SLA保障的应用场景。
实施例2
本发明实施例提供一种云应用伸缩装置，如图10所示，云应用伸缩装置用于执行以上方法中的云应用伸缩装置所执行的步骤。云应用伸缩装置可以包括相应步骤所对应的模块。示例性的，云应用伸缩装置可以包括获取模块10，计算模块11和执行模块12。
获取模块10，用于获取虚拟网络功能VNF系统的负载和虚拟 机VM个数。
计算模块11，用于在获取模块10获取VNF系统的负载和VM个数后，计算在负载和VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第一收益值，其中，动作集中的各个动作均指示增加或者减少VM的个数。
获取模块10，还用于在计算模块11计算在负载和VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第一收益值后，获取第一动作，其中，第一动作为所能获得的对VNF系统的第一收益值最大的动作。
执行模块12，用于在获取模块10获取第一动作后，执行第一动作。
可选的，VNF系统的负载和VM个数为VNF系统第一时刻的负载和VM个数。
获取模块10，还用于在获取模块10获取第一动作前，获取第一预测结果，其中，第一预测结果为预测到的第二时刻的负载，第二时刻晚于第一时刻。
计算模块11，还用于在获取模块10获取第一预测结果后，计算在第一预测结果和VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第二收益值；并根据各个动作所能获得的对VNF系统的第二收益值和各个动作所能获得的对VNF系统的第一收益值，计算各个动作所能获得的对VNF系统的第三收益值。
获取模块10，具体用于获取第二动作，其中，第二动作为所能获得的对VNF系统的第三收益值最大的动作。
执行模块12，具体用于在获取模块10获取第二动作后，执行第二动作。
可选的，获取装置10，具体用于根据VNF系统的负载和预设算法，计算第一预测结果。
可选的，如图11所示，云应用伸缩装置还包括接收模块13。
接收模块13，用于接收网元管理系统EMS发送的配置信息， 其中，配置信息至少包括预设算法。
可选的，如图12所示，云应用伸缩装置还包括确认模块14。
确认模块14，用于在获取模块10获取第一动作后，确认第一动作允许执行。
可以理解的是，本实施例的云应用伸缩装置可对应于上述如图5-图9任意之一的实施例的云应用伸缩方法中的云应用伸缩装置，并且本实施例的云应用伸缩装置中的各个模块的划分和/或功能等均是为了实现如图5-图9任意之一所示的方法流程，为了简洁，在此不再赘述。
可选的，作为本发明的另一个实施例，获取模块10、计算模块11、执行模块12和确认模块14的功能可以由处理器实现，接收模块13的功能可以由接收器实现。
本发明实施例提供一种云应用伸缩装置。基于上述实施例的描述，由于云应用伸缩装置计算在负载和VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第一收益值，并获取所能获得的对VNF系统的第一收益值最大的第一动作，第一动作是虚拟机扩容/缩容的最佳结果，能够减少不必要的扩容/缩容操作，极大地缓解了乒乓效应，并且提高了资源的利用率，提升系统的可靠性，减少了VNF系统违反SLA的风险，尤其适合需要高SLA保障的应用场景。
实施例3
本发明实施例还提供一种云应用伸缩装置，如图13所示，该云应用伸缩装置包括：存储器20、处理器21、通信接口22和系统总线23。
存储器20、处理器21和通信接口22通过系统总线23连接，存储器20用于存储一些计算机指令，处理器21用于执行计算机指令，以使云应用伸缩装置执行如图5-图9任意之一的云应用伸缩方法。具体的云应用伸缩方法可参见上述如图5-图9任意之一所示的 实施例中的相关描述，此处不再赘述。
处理器21可以为中央处理器(central processing unit，CPU)。处理器21还可以为其他通用处理器、数字信号处理器(digital signal processing，DSP)、专用集成电路(application specific integrated circuit，ASIC)、现场可编程门阵列(field-programmable gate array，FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件等。通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等。
处理器21可以为专用处理器，该专用处理器可以包括基带处理芯片、射频处理芯片等中的至少一个。进一步地，该专用处理器还可以包括具有云应用伸缩装置其他专用处理功能的芯片。
存储器20可以包括易失性存储器(volatile memory)，例如随机存取存储器(random-access memory，RAM)；存储器20也可以包括非易失性存储器(non-volatile memory)，例如只读存储器(read-only memory，ROM)，快闪存储器(flash memory)，硬盘(hard disk drive，HDD)或固态硬盘(solid-state drive，SSD)；存储器20还可以包括上述种类的存储器的组合。
系统总线23可以包括数据总线、电源总线、控制总线和信号状态总线等。本实施例中为了清楚说明，在图13中将各种总线都示意为系统总线23。
通信接口22可以包括接收器和发送器。并且在云应用伸缩装置的具体实现中，接收器和发送器具体可以是云应用伸缩装置上的收发器。该收发器可以为无线收发器。
在具体实现过程中，上述如图5-图9任意之一所示的方法流程中的各步骤均可以通过硬件执行软件形式的计算机执行指令实现。为避免重复，此处不再赘述。
本发明实施例提供一种云应用伸缩装置。基于上述实施例的描述，由于云应用伸缩装置计算在负载和VM个数的情况下，执行动作集中的各个动作所能获得的对VNF系统的第一收益值，并获取所 能获得的对VNF系统的第一收益值最大的第一动作，第一动作是虚拟机扩容/缩容的最佳结果，能够减少不必要的扩容/缩容操作，极大地缓解了乒乓效应，并且提高了资源的利用率，提升系统的可靠性，减少了VNF系统违反SLA的风险，尤其适合需要高SLA保障的应用场景。
实施例4
本发明实施例还提供一种软件产品，该软件产品可以包括实现云应用伸缩方法的计算机指令。
计算机指令可以存储在可读存储介质上；处理器可以从该可读存储介质上读取到计算机指令并执行，使得处理器实现云应用伸缩方法。
所属领域的技术人员可以清楚地了解到，为描述的方便和简洁，仅以上述各功能模块的划分进行举例说明，实际应用中，可以根据需要而将上述功能分配由不同的功能模块完成，即将装置的内部结构划分成不同的功能模块，以完成以上描述的全部或者部分功能。上述描述的系统，装置和单元的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。
在本申请所提供的几个实施例中，应该理解到，所揭露的装置和方法，可以通过其它的方式实现。例如，以上所描述的装置实施例仅仅是示意性的，例如，所述模块或单元的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，例如多个单元或组件可以结合或者可以集成到另一个系统，或一些特征可以忽略，或不执行。另一点，所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口，装置或单元的间接耦合或通信连接，可以是电性，机械或其它的形式。
所述作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实 际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。
另外，在本发明各个实施例中的各功能单元可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现，也可以采用软件功能单元的形式实现。
所述集成的单元如果以软件功能单元的形式实现并作为独立的产品销售或使用时，可以存储在一个计算机可读取存储介质中。基于这样的理解，本发明的技术方案本质上或者说对现有技术做出贡献的部分或者该技术方案的全部或部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质中，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)或处理器(processor)执行本发明各个实施例所述方法的全部或部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(ROM，Read-Only Memory)、随机存取存储器(RAM，Random Access Memory)、磁碟或者光盘等各种可以存储程序代码的介质。
以上所述，仅为本发明的具体实施方式，但本发明的保护范围并不局限于此，任何熟悉本技术领域的技术人员在本发明揭露的技术范围内，可轻易想到变化或替换，都应涵盖在本发明的保护范围之内。因此，本发明的保护范围应所述以权利要求的保护范围为准。

Claims (11)

  1. 一种云应用伸缩方法，其特征在于，包括：
    获取虚拟网络功能VNF系统的负载和虚拟机VM个数；
    计算在所述负载和所述VM个数的情况下，执行动作集中的各个动作所能获得的对所述VNF系统的第一收益值，其中，所述动作集中的各个动作均指示增加或者减少所述VM的个数；
    获取并执行第一动作，其中，所述第一动作为所能获得的对所述VNF系统的第一收益值最大的动作。
  2. 根据权利要求1所述的云应用伸缩方法，其特征在于，所述VNF系统的负载和VM个数为所述VNF系统第一时刻的负载和VM个数；
    在所述获取并执行第一动作前，所述方法还包括：
    获取第一预测结果，其中，所述第一预测结果为预测到的第二时刻的负载，所述第二时刻晚于所述第一时刻；
    计算在所述第一预测结果和所述VM个数的情况下，执行动作集中的各个动作所能获得的对所述VNF系统的第二收益值；
    根据各个动作所能获得的对所述VNF系统的第二收益值和各个动作所能获得的对所述VNF系统的第一收益值，计算各个动作所能获得的对所述VNF系统的第三收益值；
    所述获取并执行第一动作，具体包括：
    获取并执行第二动作，其中，所述第二动作为所能获得的对所述VNF系统的第三收益值最大的动作。
  3. 根据权利要求2所述的云应用伸缩方法，其特征在于，所述获取第一预测结果，具体包括：
    根据所述所述VNF系统的负载和预设算法，计算第一预测结果。
  4. 根据权利要求3所述的云应用伸缩方法，其特征在于，所述方法还包括：
    接收网元管理系统EMS发送的配置信息，其中，所述配置信息至少包括所述预设算法。
  5. 根据权利要求1-4中任意一项所述的云应用伸缩方法，其特征在于，在所述获取第一动作后，所述方法还包括：
    确认所述第一动作允许执行。
  6. 一种云应用伸缩装置，其特征在于，包括获取模块，计算模块和执行模块；
    所述获取模块，用于获取虚拟网络功能VNF系统的负载和虚拟机VM个数；
    所述计算模块，用于在所述获取模块获取VNF系统的负载和VM个数后，计算在所述负载和所述VM个数的情况下，执行动作集中的各个动作所能获得的对所述VNF系统的第一收益值，其中，所述动作集中的各个动作均指示增加或者减少所述VM的个数；
    所述获取模块，还用于在所述计算模块计算在所述负载和所述VM个数的情况下，执行动作集中的各个动作所能获得的对所述VNF系统的第一收益值后，获取第一动作，其中，所述第一动作为所能获得的对所述VNF系统的第一收益值最大的动作；
    所述执行模块，用于在所述获取模块获取第一动作后，执行所述第一动作。
  7. 根据权利要求6所述的云应用伸缩装置，其特征在于，所述VNF系统的负载和VM个数为所述VNF系统第一时刻的负载和VM个数；
    所述获取模块，还用于在所述获取模块获取第一动作前，获取第一预测结果，其中，所述第一预测结果为预测到的第二时刻的负载，所述第二时刻晚于所述第一时刻；
    所述计算模块，还用于在所述获取模块获取第一预测结果后，计算在所述第一预测结果和所述VM个数的情况下，执行动作集中的各个动作所能获得的对所述VNF系统的第二收益值；并根据各个动作所能获得的对所述VNF系统的第二收益值和各个动作所能获得的对所述VNF系统的第一收益值，计算各个动作所能获得的对所述VNF系统的第三收益值；
    所述获取模块，具体用于获取第二动作，其中，所述第二动作为所能获得的对所述VNF系统的第三收益值最大的动作；
    所述执行模块，具体用于在所述获取模块获取所述第二动作后，执行所述第二动作。
  8. 根据权利要求7所述的云应用伸缩装置，其特征在于，
    所述获取装置，具体用于根据所述所述VNF系统的负载和预设算法，计算第一预测结果。
  9. 根据权利要求8所述的云应用伸缩装置，其特征在于，所述云应用伸缩装置还包括接收模块；
    所述接收模块，用于接收网元管理系统EMS发送的配置信息，其中，所述配置信息至少包括所述预设算法。
  10. 根据权利要求6-9中任意一项所述的云应用伸缩装置，其特征在于，所述云应用伸缩装置还包括确认模块；
    所述确认模块，用于在所述获取模块获取第一动作后，确认所述第一动作允许执行。
  11. 一种云应用伸缩装置，其特征在于，所述云应用伸缩装置包括存储器、处理器、通信接口和系统总线；
    所述存储器、所述处理器和所述通信接口通过所述系统总线连接，所述存储器用于存储计算机指令，所述处理器用于执行所述存储器存储的计算机指令，以使所述云应用伸缩装置执行权利要求1-5任意一项所述的云应用伸缩方法。
PCT/CN2016/086267 2016-06-17 2016-06-17 一种云应用伸缩方法及装置 WO2017214986A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2016/086267 WO2017214986A1 (zh) 2016-06-17 2016-06-17 一种云应用伸缩方法及装置

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2016/086267 WO2017214986A1 (zh) 2016-06-17 2016-06-17 一种云应用伸缩方法及装置

Publications (1)

Publication Number Publication Date
WO2017214986A1 true WO2017214986A1 (zh) 2017-12-21

Family

ID=60662903

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/086267 WO2017214986A1 (zh) 2016-06-17 2016-06-17 一种云应用伸缩方法及装置

Country Status (1)

Country Link
WO (1) WO2017214986A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109815092A (zh) * 2019-01-28 2019-05-28 中国工商银行股份有限公司 云平台自动伸缩方法及系统
CN111385142A (zh) * 2020-02-14 2020-07-07 重庆邮电大学 基于Kubernetes的自适应伸缩web容器的方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102214117A (zh) * 2010-04-07 2011-10-12 中兴通讯股份有限公司 虚拟机管理方法、系统及虚拟机管理服务器
CN103425535A (zh) * 2013-06-05 2013-12-04 浙江大学 云环境下的敏捷弹性伸缩方法
CN103984602A (zh) * 2014-05-20 2014-08-13 华为技术有限公司 一种vm资源调度方法、装置及系统
CN104038392A (zh) * 2014-07-04 2014-09-10 云南电网公司 一种云计算资源服务质量评估方法
WO2015153242A1 (en) * 2014-03-31 2015-10-08 Microsoft Technology Licensing, Llc Dynamically identifying target capacity when scaling cloud resources

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102214117A (zh) * 2010-04-07 2011-10-12 中兴通讯股份有限公司 虚拟机管理方法、系统及虚拟机管理服务器
CN103425535A (zh) * 2013-06-05 2013-12-04 浙江大学 云环境下的敏捷弹性伸缩方法
WO2015153242A1 (en) * 2014-03-31 2015-10-08 Microsoft Technology Licensing, Llc Dynamically identifying target capacity when scaling cloud resources
CN103984602A (zh) * 2014-05-20 2014-08-13 华为技术有限公司 一种vm资源调度方法、装置及系统
CN104038392A (zh) * 2014-07-04 2014-09-10 云南电网公司 一种云计算资源服务质量评估方法

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109815092A (zh) * 2019-01-28 2019-05-28 中国工商银行股份有限公司 云平台自动伸缩方法及系统
CN111385142A (zh) * 2020-02-14 2020-07-07 重庆邮电大学 基于Kubernetes的自适应伸缩web容器的方法
CN111385142B (zh) * 2020-02-14 2022-04-12 重庆邮电大学 基于Kubernetes的自适应伸缩web容器的方法

Similar Documents

Publication Publication Date Title
WO2020258920A1 (zh) 一种网络切片资源管理方法及设备
CN107566194B (zh) 一种实现跨域虚拟网络映射的方法
Ren et al. Dynamic auto scaling algorithm (DASA) for 5G mobile networks
US10289451B2 (en) Method, apparatus, and system for adjusting deployment location of virtual machine
JP7159347B2 (ja) モデル更新方法および装置、ならびにシステム
CN109343965A (zh) 资源调整方法、装置、云平台和服务器
CN106533723B (zh) 虚拟资源调度方法、装置及系统
CN113037786B (zh) 智能算力调度方法、装置和系统
Yousaf et al. Cost analysis of initial deployment strategies for virtualized mobile core network functions
US20120221730A1 (en) Resource control system and resource control method
CN104901989B (zh) 一种现场服务提供系统及方法
WO2017128980A1 (zh) 云平台中管理资源的方法和装置
CN104038392A (zh) 一种云计算资源服务质量评估方法
WO2016086406A1 (zh) 一种网络资源部署方法和设备
CN112703774B (zh) 管理电信网络中的处理资源的方法和电信网络及存储介质
CN107483292B (zh) 用于云平台的动态监控方法
US10949315B2 (en) Performance measurements related to virtualized resources
CN108021447B (zh) 一种基于分布式数据确定最优资源策略的方法及系统
CN110430068A (zh) 一种特征工程编排方法及装置
CN112365366A (zh) 一种基于智能化5g切片的微电网管理方法及系统
CN110647392A (zh) 一种基于容器集群的智能弹性伸缩方法
Phung-Duc et al. Design and analysis of deadline and budget constrained autoscaling (DBCA) algorithm for 5G mobile networks
CN111078404A (zh) 一种计算资源确定方法、装置、电子设备及介质
CN114780244A (zh) 容器云资源弹性分配方法、装置、计算机设备及介质
KR101630125B1 (ko) 클라우드 컴퓨팅 자원관리 시스템에서의 자원 요구량 예측 방법

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: 16905100

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 16905100

Country of ref document: EP

Kind code of ref document: A1