JP2018180773A - Management device and management method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a management device and a management method capable of appropriately autoscaling or migrating a virtual network function. SOLUTION: A management device is connected to a function management unit for adding a virtual machine to a virtual network function unit formed in a physical server or relocating the virtual network function unit, and a transfer path via the virtual network function unit. Based on the amount of increase in the input traffic per unit time, and the comparison result of the parameter related to the queue of the traffic provided in the virtual machine constituting the virtual network function unit and a predetermined threshold value, the function management unit is notified. It has an instruction unit for instructing the addition of a virtual machine to the virtual network function unit or the relocation of the virtual network function unit. [Selection diagram] FIG. 6

Description

  The present invention relates to a management apparatus and a management method.

  There is known an NFV (Network Function Virtualization) technology in which various network functions are virtually formed on a general-purpose server or the like using an application (see, for example, Patent Document 1). According to the NFV technology, network functions can be realized without the need for dedicated hardware, and therefore, the cost reduction of the network system is expected.

  With NFV, a virtual network function (VNF) is formed on a physical server. For the virtual network function, resources such as a central processing unit (CPU) and a memory are allocated as virtual machines (VMs) according to processing load, as in the cloud technology, for example. Thus, the VNF is configured by one or more VMs.

  The VNF is formed, for example, on a general-purpose server installed in a data center. The network functions implemented by the VNF include, for example, a firewall, a web proxy, and a VPN (Virtual Private Network) router.

  The management server which manages the VNF arranges and sets the VNF for the general-purpose server in accordance with a request from a user such as a company. As a result, a service chain (SC: Service Chain), which is a transfer path of data in a wide area network, for example, is set up between various VNFs such as a firewall, a web proxy, and a VPN router. The user can access the Internet and other locations of the user through the service chain.

JP, 2015-153330, A

  The management server adds a VM to the VNF, for example, according to the load of the traffic forwarding process of the VNF. This process is called, for example, auto scaling. When auto-scaling is performed, it becomes possible to evenly distribute the load of transfer processing to a plurality of VMs configuring the VNF, so the data transfer amount increases.

  However, it is difficult to specify the VNF (that is, the bottleneck VNF) that causes the decrease in the data transfer amount, for example, because it is determined not only by the status of the traffic input to the SC but also by the configuration of the SC. If autoscaling is performed on a VNF that is not the cause of the decrease in data transfer amount, the resources of the general-purpose server are wasted.

  Also, for example, there may be a case where the data transfer amount is reduced due to the performance of the virtual switch on the harperizer, which is the virtual basis of the VNF, and the hardware performance of the general-purpose server. In this case, since the data transfer amount does not increase even when the auto scaling is performed, the resource of the general-purpose server is wastefully consumed.

  Therefore, when the data transfer amount decreases due to the performance of the virtual switch and the hardware performance of the general server, the management server transfers data, for example, by transferring the VNF to another virtual switch or another general server. Improve transfer volume. This process is called, for example, migration.

  However, identifying that the cause of the decrease in data transfer amount is a virtual switch or hardware is as difficult as identifying the bottleneck VNF. If migration is executed when the cause of the decrease in the amount of data transfer is not a virtual switch or hardware, resources of the virtual switch of the transfer destination of the VNF and the general-purpose server are wasted.

  Therefore, it is an object of the present invention to provide a management apparatus and management method capable of appropriately auto-scaling or migrating virtual network functions.

  In one aspect, the management apparatus includes a function management unit that adds a virtual machine to a virtual network function unit formed in a physical server or transfers the virtual network function unit, and a transfer route passing through the virtual network function unit. The function management unit based on the amount of increase per unit time of the traffic input to the unit and the comparison result of the parameter regarding the queue of the traffic provided in the virtual machine constituting the virtual network function unit and a predetermined threshold value And an instruction unit for instructing addition of a virtual machine to the virtual network function unit or relocation of the virtual network function unit.

  In one aspect, the management apparatus is a function management unit that adds a virtual machine to a virtual network function unit formed in a physical server or transfers the virtual network function unit, and a virtual machine that configures the virtual network function unit. Addition of a virtual machine to the virtual network function unit in the function managing unit based on parameters related to an input queue storing input traffic and parameters related to an output queue storing the traffic output from the virtual machine And an instruction unit for instructing transfer of the virtual network function unit.

  In one aspect, the management method includes increasing an amount of traffic per unit time input to a transfer path via a virtual network function unit formed in a physical server, and a virtual machine configuring the virtual network function unit. It is a method of adding a virtual machine to the virtual network function unit or transferring the virtual network function unit based on the comparison result of the provided parameter regarding the queue of traffic and a predetermined threshold.

  In one aspect, the management method stores a parameter related to an input queue storing traffic input to a virtual machine configuring a virtual network function unit formed in a physical server, and the traffic output from the virtual machine It is a method of adding a virtual machine to the virtual network function unit or relocating the virtual network function unit based on parameters related to an output queue.

  In one aspect, virtual network features can be auto-scaled or migrated appropriately.

It is a block diagram which shows an example of a network management system. It is a block diagram which shows an example of a general purpose physical server. It is a block diagram which shows an example of a management server. It is a figure which shows the example of SC management table, VM management table, and virtual SW management table. It is a figure which shows operation | movement of the management server of 1st Example. It is a figure which shows operation | movement of the management server of 1st Example. It is a figure which shows the example of SC management table before and behind auto scaling, and VM management table. It is a flowchart which shows the process of the management server of 1st Example. It is a figure which shows operation | movement of the management server of 2nd Example. It is a flowchart which shows the process of the management server of 2nd Example. It is a figure which shows operation | movement of the management server of 3rd Example. It is a figure which shows the example of SC management table and VM management table before auto scaling. It is a figure which shows the state after auto scaling. It is a figure which shows the example of SC management table and VM management table after auto scaling. It is a flowchart which shows the process of the management server of 3rd Example. It is a figure which shows operation | movement of the management server of 4th Example. It is a flowchart which shows the process of the management server of 4th Example. It is a figure which shows operation | movement of the management server of 5th Example. It is a flowchart which shows the process of the management server of 5th Example. It is a figure which shows operation | movement of the management server of 6th Example. It is a figure which shows operation | movement of the management server of 6th Example. It is a figure which shows the example of VM management table before and behind migration. It is a flowchart which shows the process of the management server of 6th Example. It is a figure which shows the example of VM management table and virtual SW management table of 7th Example. It is a flowchart which shows the process of the management server of 7th Example. It is a figure which shows operation | movement of the management server of 8th Example. It is a flowchart which shows the process of the management server of 8th Example.

  FIG. 1 is a block diagram showing an example of a network management system. The network management system includes a management server 1 which is an example of a management apparatus, and a general-purpose physical server 2 which is an example of a physical server.

  The management server 1 communicates with the general-purpose physical server 2 via a management network NWc such as a LAN (Local Area Network). In the general-purpose physical server 2, VNFs 91 to 93, which are an example of a virtual network function unit, and load balancers LB # 1 to # 3 are formed. The management server 1 communicates with the general-purpose physical server 2 to manage the VNFs 91 to 93 and the load balancers LB # 1 to # 3.

  As an example, the VNF 91 includes, as a network function, Web proxies PX # 1 and # 2 which are virtual machines (VMs), and the VNF 92 includes a plurality of firewalls FW # 1 to # 4 which are VMs. The VNF 93 includes, as a network function, VPN routers RT # 1 to # 3 which are VMs.

  In the VNFs 91 to 93, a service chain (SC), which is a transfer path of traffic in a wide area network, is set by the management server 1. Thus, the user can access another network NWb from the network NWa via the general-purpose physical server 2. The networks NWa and NWb include a LAN, an intranet, and the Internet.

  At the input side of each of the VNFs 91 to 93, load balancers LB # 1 to # 3 are provided to distribute the traffic load among the VMs. The load balancer LB # 1 evenly distributes the traffic input from the network NWa to the Web proxies PX # 1 and # 2 of the VNF 91. The traffic processed by each of the Web proxies PX # 1 and # 2 is input to the load balancer LB # 2.

  The load balancer LB # 2 evenly distributes the traffic input from the Web proxies PX # 1 and # 2 to the firewalls FW # 1 to FW4 of the VNF 92. The traffic processed by the firewalls FW # 1 to # 4 is input to the load balancer LB # 3.

  The load balancer LB # 3 equally distributes the traffic input from the firewalls FW # 1 to # 4 to the VPN routers RT # 1 to # 3 of the VNF 93. The traffic processed by each of the VPN routers RT # 1 to # 3 is transmitted to the network NWb.

  The management server 1 adds a VM to the VNF, for example, according to the load of the forwarding process of the VNF traffic. When the load of transfer processing of the VNF 92 is large, the management server 1 adds a firewall FW to the VNF 92 by executing auto scaling.

  More specifically, the management server 1 identifies a VM that is a bottleneck in traffic transfer processing from the VMs of VNF 91 to 93, and performs autoscaling on the VNF 91 to 93 including the VM. In order to specify a VM, parameters relating to an input queue and an output queue of traffic provided in the VM are used. Thus, the management server 1 can appropriately autoscale VNF 91 to 93. The configuration of the general-purpose physical server 2 will be described first, and then the configuration of the management server 1 will be described.

  FIG. 2 is a block diagram showing an example of the general-purpose physical server 2. For example, in the data center, a plurality of general-purpose physical servers 2 are installed.

  Each general-purpose physical server 2 is connected to a network NWd (for example, LAN) in the data center via a ToR (Top-Of-Rack Switch) 3. Further, a router 4 for communicating with the networks NWa and NWb is connected to the network NWd. Therefore, each general-purpose physical server 2 can communicate with the networks NWa and NWb via the router 4. Also, each general-purpose physical server 2 can communicate with the management server 1 via the management network NWc.

  The general-purpose physical server 2 includes a plurality of central processing units (CPUs) 20, a memory 21, a chipset 22, a storage 23, and a network interface card (NIC) 24. The NIC 24 handles communication with the management network NWc and the network NWd. Each CPU 20 communicates with the management server 1 and the networks NWa and NWb via the NIC 24.

  The CPU 20 is connected to the memory 21, the chipset 22, and the storage 23. The memory 21 functions as a working memory of the CPU 20. The chipset 22 includes hardware such as a bus controller, for example.

  The storage 23 is, for example, a storage device such as a hard disk drive. The storage 23 stores an OS (Operating System) for driving the CPU 20 and various software. In the CPU 20, a hypervisor 200 and a VM 201 are formed as functions. The CPUs 20 operate in cooperation with each other.

  The hypervisor 200 is an OS that is a basis for operating each VM 201. The hypervisor 200 has virtual switches (virtual SW) 210 and 211 as functions. The virtual switches 210 and 211 distribute the traffic to each VM 201 according to the setting of the service chain. The virtual switches 210 and 211 are formed in different CPUs 20 as an example.

  The VM 201 is configured by virtual hardware. The VM 201 includes a virtual CPU (vCPU) 213, a virtual NIC (vNIC) 212, a virtual memory (v memory) 214, and a virtual storage (v storage) 215. With this configuration, the VM 201 executes various network functions.

  FIG. 3 is a block diagram showing an example of the management server 1. The management server 1 includes a CPU 10, a read only memory (ROM) 11, a random access memory (RAM) 12, a hard disk drive (HDD) 13, a communication port 14, an input device 15, and an output device 16. The CPU 10 is connected to the ROM 11, the RAM 12, the HDD 13, the communication port 14, the input device 15, and the output device 16 via the bus 19 so that signals can be input to and output from each other.

  The ROM 11 stores an OS for driving the CPU 10 and a management program. The management program implements the management method described below. As ROM11, EPROM (Erasable Programmable ROM) etc. are mentioned, for example.

  The RAM 12 functions as a working memory of the CPU 10. The communication port 14 is, for example, a wireless LAN card or a NIC, and is connected to the management network NWc. The CPU 10 communicates with the general-purpose physical server 2 via the communication port 14.

  The input device 15 is a device that inputs information to the management server 1. Examples of the input device 15 include a keyboard, a mouse, and a touch panel. The input device 15 outputs the input information to the CPU 10 via the bus 19.

  The output device 16 is a device that outputs information of the management server 1. Examples of the output device 16 include a display, a touch panel, a speaker, and a printer. The output device 16 acquires and outputs information from the CPU 10 via the bus 19.

  When the CPU 10 reads a program from the ROM 11, a VNF determination unit 100, an auto scaling execution unit 101, a migration execution unit 102, and a path setting unit 103 are formed as functions. The HDD 13 also stores an SC management table (TBL) 130, a VM management table (TBL) 131, and a virtual SW management table (TBL) 132.

  FIG. 4 is a diagram showing an example of the SC management table 130, the VM management table 131, and the virtual SW management table 132. When forming the VNF in the general-purpose physical server 2, the CPU 10 generates the SC management table 130, the VM management table 131, and the virtual SW management table 132.

  The SC management table 130 includes an SC identifier “SC # i” (i is a positive integer) for identifying an SC, a transmission source address “ADsr” and a destination address “ADds” of the SC, and the number of VNFs constituting the SC “n “(I is a positive integer) and its configuration information of VNFs # 1 to #n are registered. For each VNF # 1 to #n configuration information, the function name "FW" of the VNF (for example, FW = firewall, PRX = Web proxy, RT = VPN router), the number of VMs "m" (m is a positive integer) And an LB address "ADbc" indicating a load balancer.

  In the VM management table 131, an SC identifier "SC # i" for identifying an SC and management information of VNFs # 1 to #n are registered. Each management information of VNF # 1 to #n includes ID "FW # 1" (for function name FW) of VM # 1 to #m, management address "ADmg", and transfer address "ADtr". It is done.

  In the virtual SW management table 132, the identifiers “SWa” and “SWb” of the virtual switches 210 and 211 (virtual SW # 1 and # 2) formed in the hypervisor 200 are registered. The VM management table 131 and the virtual SW management table 132 also include parameters other than the above depending on the embodiment.

  Referring again to FIG. 3, the auto scaling execution unit 101 performs auto scaling of VNF. The auto scaling execution unit 101 adds a VM to the VNF in accordance with the instruction of the VNF determination unit 100.

  The migration execution unit 102 executes VNF migration. The migration execution unit 102 relocates the VNF in accordance with the instruction of the VNF determination unit 100. The auto scaling execution unit 101 and the migration execution unit 102 are an example of a function management unit.

  The VNF determination unit 100 determines, among the VMs 201 of one or more VNFs through which the SC passes, the VM 201 which is a bottleneck of transfer processing. The VNF determination unit 100 is an example of an instruction unit.

  The VNF determination unit 100 compares the amount of increase per unit time of traffic input to the SC (hereinafter referred to as “input traffic”), and the comparison result of the traffic queue parameter provided in the VM 201 and a predetermined threshold value. And identify the VM 201 of the bottleneck. For example, even if the stored amount of traffic (for example, Byte) in the input queue of the VM 201 increases due to a temporary increase in input traffic, for example, the VNF determination unit 100 does not specify the VM 201 as a bottleneck.

  This is because when the input traffic temporarily increases, the storage amount of the input queue increases regardless of the transfer processing performance of the VM 201. Therefore, the VNF determination unit 100 identifies the VM 201 whose storage amount of the input queue has increased when the input traffic has not increased as a bottleneck.

  As described above, the VNF determination unit 100 identifies the VM 201 of the bottleneck. The VNF determination unit 100 instructs the auto scaling execution unit 101 to execute auto scaling or instructs the migration execution unit 102 to execute migration based on the identification result.

  More specifically, when the VNF determination unit 100 identifies the bottleneck VM 201, the VNF determination unit 100 instructs the auto scaling execution unit 101 to execute auto scaling of the VM 201. When the VNF determination unit 100 identifies not the VM 201 but the virtual switches 210 and 211 as a bottleneck, the VNF determination unit 100 instructs the migration execution unit 102 to migrate the VM 201 connected to the virtual switches 210 and 211.

  As described above, the VNF determination unit 100 adds the VM 201 to the VNF to the auto scaling execution unit 101 based on the increase amount per unit time of the input traffic and the comparison result of the parameter related to the queue of the VM 201 and the predetermined threshold. Instruct the migration execution unit 102 to relocate the VNF. Therefore, the VNF determination unit 100 can appropriately autoscale or migrate the VNF.

  Further, in another embodiment, the VNF determination unit 100 determines a bottle of transfer processing based on a parameter related to an input queue storing traffic input to the VM 201 and a parameter related to output queue storing traffic output from the VM 201. The VM 201 that is the bottleneck is determined. For example, the VNF determination unit 100 identifies the VM 201 of the bottleneck from the storage amounts of the input queue and the output queue of the VM 201.

  When the storage amount of the input queue of the VM 201 is large but the storage amount of the output queue is small, the VNF determination unit 100 identifies the VM 201 as a bottleneck. This is because if the transfer processing performance of the VM 201 is high, even if the storage amount of the input queue of the VM 201 increases, the storage amount of the output queue decreases.

  Also in this case, the VNF determination unit 100 instructs the auto-scaling execution unit 101 to execute auto scaling or instructs the migration execution unit 102 to execute migration based on the bottleneck identification result.

  As described above, the VNF determination unit 100 instructs the auto-scaling execution unit 101 to add the VM 201 to the VNF based on the parameters related to the input queue of the VM 201 and the parameters related to the output queue of the VM 201, and to the migration execution unit 102. Direct VNF relocation. Therefore, the VNF determination unit 100 can appropriately autoscale or migrate the VNF.

  Further, the path setting unit 103 sets the SC according to the instruction of the auto scaling execution unit 101 or the migration execution unit 102. More specifically, the path setting unit 103 sets a new SC to the VM 201 and the virtual switches 210 and 211 when auto scaling or migration is performed. Therefore, the general-purpose physical server 2 can transfer traffic even after execution of auto scaling or migration.

  Next, the operation of the management server 1 in each embodiment will be described.

(First embodiment)
5 and 6 are diagrams showing the operation of the management server 1 according to the first embodiment. FIG. 5 shows a state before execution of auto scaling, and FIG. 6 shows a state after execution of auto scaling. In FIGS. 5 and 6, the same components as those in FIGS. 2 and 3 are designated by the same reference numerals, and the description thereof will be omitted.

  Referring to FIG. 5, in the general-purpose physical server 2, a load balancer LB # 1 and a firewall FW # 1 are formed as the VM 201. The load balancer LB # 1 and the firewall FW # 1 input / output traffic to / from the virtual switch (# 1) 210 of the hypervisor 200.

  The firewall FW # 1 is provided with an input queue 80 and an output queue 81 for traffic. The input queue 80 stores the traffic input from the virtual switch 210 to the firewall FW # 1 in units of packets (see rectangular frame). The output queue 81 stores the traffic output from the firewall FW # 1 to the virtual switch 210 in units of packets. For example, in the case of network processing software "vhost", the input queue 80 corresponds to "Rx ring" and the output queue 81 corresponds to "Tx ring". The input queue 80 and the output queue 81 are examples of queues.

  The code R1 indicates the path of the SC. Traffic passes through the load balancer LB # 1 and the firewall FW # 1 in this order. More specifically, traffic is input from the router 4 to the load balancer LB # 1 via the virtual switch 210. The traffic output from the load balancer # 1 is input to the input queue 80 via the virtual switch 210.

  The firewall FW # 1 reads traffic from the input queue 80 on a packet basis and processes it. The firewall FW # 1 writes the processed traffic to the output queue 81 on a packet basis. The virtual switch 210 reads traffic from the output queue 81 in units of packets and outputs the traffic to the router 4. In this way, traffic is forwarded along the SC.

  The router 4 is provided with a traffic monitoring unit 40 that monitors SC input traffic. The router 4 is configured by, for example, a CPU circuit similar to the management server 1, and the traffic monitoring unit 40 can be formed by software as a function of the CPU, for example. The traffic monitoring unit 40 notifies the VNF determination unit 100 of the input traffic monitoring result via the general-purpose physical server 2.

  Further, in the CPU 20 of the general-purpose physical server 2, a queue monitoring unit 220 is formed as a function of software. The queue monitoring unit 220 monitors the storage amount of the input queue 80 and notifies the VNF determination unit 100 of the monitoring result.

  The VNF determination unit 100 periodically acquires input traffic monitoring results from the traffic monitoring unit 40. For example, an SC identifier is added to the monitoring result of the input traffic, and the VNF determination unit 100 manages the monitoring result of the input traffic for each SC based on the SC identifier. In addition, instead of or in addition to the SC identifier, the source address and the destination address of the SC may be used to manage the monitoring result of the input traffic.

  More specifically, the VNF determination unit 100 acquires an increase amount ΔTsc per unit time of input traffic. At this time, the traffic monitoring unit 40 may notify the increase amount ΔTsc according to a periodic instruction from the VNF determination unit 100, or periodically notify the increase amount ΔTsc without an instruction from the VNF determination unit 100. It is also good.

  The traffic monitoring unit 40 measures an input traffic amount every unit time (for example, one second), and calculates a difference from the previously measured input traffic amount. The traffic monitoring unit 40 reports the difference as the amount of increase ΔTsc per unit time of the input traffic. The increase amount ΔTsc may be calculated by the VNF determination unit 100. In this case, the traffic monitoring unit 40 notifies the VNF determination unit 100 of the input traffic amount per unit time.

  Further, the VNF determination unit 100 periodically acquires the storage amount Lin of the input queue 80 from the queue monitoring unit 220. For example, the ID of the corresponding VM 201 is assigned to the storage amount Lin of the input queue 80, and the VNF determination unit 100 manages the storage amount Lin of the input queue 80 for each VM 201 based on the ID of the VM 201. At this time, the queue monitoring unit 220 may notify the storage amount Lin of the input queue 80 in accordance with the periodic instruction from the VNF determination unit 100, or the input queue 80 may be periodically notified without the instruction from the VNF determination unit 100. The storage amount Lin of may be notified.

  The VNF determination unit 100 instructs the auto-scaling execution unit 101 to execute auto scaling based on the increase amount ΔTsc per unit time of input traffic and the comparison result of the storage amount Lin of the input queue 80 and the predetermined threshold THa. Do. That is, in this example, the storage amount Lin of the input queue 80 is used as a parameter related to the traffic queue.

  More specifically, when the storage amount Lin of the input queue 80 is larger than the predetermined threshold THa and the increase amount ΔTsc per unit time of the input traffic is equal to or smaller than the predetermined threshold THb, the VNF determination unit 100 Is instructed to the auto scaling execution unit 101. That is, when the input traffic is not increasing and the storage amount Lin of the input queue 80 is increasing, the VNF determination unit 100 identifies the VM 201 as a bottleneck, and as shown in FIG. Instruct to add a new VM.

  The auto scaling execution unit 101 adds a new firewall FW # 2 in accordance with the instruction of the VNF determination unit 100. More specifically, the auto-scaling execution unit 101 identifies the VNF # 1 included in the firewall FW # 1 by referring to the VM management table 131, and adds the firewall FW # 2 to the VNF # 1. In the same manner as the firewall FW # 1, the firewall FW # 2 is provided with an input queue 82 and an output queue 83.

  FIG. 7 is a diagram showing an example of the SC management table 130 and the VM management table 131 before and after auto scaling. After the addition of the firewall FW # 2, the auto-scaling execution unit 101 updates the number of VMs of VNF # 1 in the SC management table 130 from 1 to 2 as indicated by a dotted line frame.

  Further, the auto-scaling execution unit 101 adds management information of the firewall FW # 2, which is the new VM # 2, to the VM management table 131, as indicated by a dotted line frame. Thus, the SC management table 130 and the VM management table 131 are updated.

  Referring again to FIG. 6, after the addition of the firewall FW # 2, the auto-scaling execution unit 101 requests the path setting unit 103 to set up the SC via the firewall FW # 2. The path setting unit 103 sets the SC for the load balancer LB # 1, the firewall FW # 2, and the virtual switch 210 in response to the request for setting the SC. As a result, SC is set as a route through which traffic is distributed from the load balancer LB # 1 to the firewalls FW # 1 and FW # 2 as indicated by a symbol R2.

  The load balancer LB # 1 equally distributes traffic to the firewalls FW # 1 and FW # 2. Therefore, the load on the forwarding process of the firewall FW # 1 is reduced, and the SC bottleneck is eliminated. Therefore, the traffic transfer amount of the entire SC increases.

  FIG. 8 is a flowchart showing the process of the management server 1 of the first embodiment. This process is performed, for example, regularly.

  The VNF determination unit 100 selects the VM 201 to be determined (step St1). At this time, the VNF determination unit 100 selects, for example, the VM 201 having the highest priority. The VNF determination unit 100 sets the priority of the VM 201 in advance according to a predetermined standard.

  For example, the VNF determination unit 100 may set the priority of the VM 201 having the largest storage amount Lin of the input queue 80 high, or the priority of the VM 201 having the storage amount Lin exceeding the predetermined threshold THa at the earliest time. May be set high. Further, the VNF determination unit 100 may set the priority of the VM 201 having the smallest difference between the maximum storage amount of the input queue 80 and the storage amount Lin high, or the priority of the VM 201 located closest to the SC exit side. May be set high. Furthermore, the VNF determination unit 100 may set the priority based on the ID of the VM 201.

  Next, the VNF determination unit 100 acquires the storage amount Lin of the input queue 80 from the queue monitoring unit 220 (step St2). Next, the VNF determination unit 100 acquires an increase amount ΔTsc per unit time of input traffic from the traffic monitoring unit 40 (step St3). At this time, the VNF determination unit 100 may obtain the input traffic amount from the traffic monitoring unit 40 and calculate the increase amount ΔTsc of the input traffic amount per unit time.

  Next, the VNF determination unit 100 compares the storage amount Lin of the input queue 80 with a predetermined threshold THa (step St4). At this time, the VNF determination unit 100 may obtain, from the queue monitoring unit 220, the result of the comparison processing performed by the queue monitoring unit 220.

  When Lin ≦ THa is satisfied (No in step St4), the VNF determination unit 100 ends the process. When Lin> THa is satisfied (Yes in step St4), the VNF determination unit 100 compares the increase amount ΔTsc per unit time of the input traffic with a predetermined threshold THb (step St5). At this time, the VNF determination unit 100 may obtain, from the traffic monitoring unit 40, the result of the comparison processing performed by the traffic monitoring unit 40.

  If ΔTsc> THb is satisfied (Yes in step St5), the VNF determination unit 100 ends the process. When Tsc ≦ THb is satisfied (No in step St5), the VNF determining unit 100 instructs the auto-scaling execution unit 101 to add the VM 201 to the VNF (step St6). The auto scaling execution unit 101 adds the VM 201 to the VNF in accordance with the instruction of the VNF determination unit 100.

  Next, the auto-scaling execution unit 101 requests the path setting unit 103 to set SC (Step St7). The path setting unit 103 sets the SC in response to a request from the auto scaling execution unit 101. Thus, the processing of the management server 1 is executed. The present process is an example of a management method executed by the management server 1.

  As described above, the VNF determination unit 100 performs auto scaling on the auto scaling execution unit 101 based on the increase amount ΔTsc per unit time of input traffic and the comparison result of the storage amount Lin of the input queue 80 and the predetermined threshold THa. To direct. Therefore, the VNF determination unit 100 can suppress unnecessary addition of the VM 201 when the storage amount Lin of the input queue 80 increases due to the increase of the input traffic. As a result, unnecessary consumption of resources and the setting process of the SC and the load balancer are avoided.

Second Embodiment
When the storage amount Lin of the input queue 80 is used for determination processing as in the first embodiment, if the maximum storage amount of the input queue 80 is different for each VM 201, it is necessary to set an individual threshold THa for each VM 201. , It takes time and effort of setting process. Therefore, the VNF determination unit 100 may use the increase amount per unit time of the storage amount Lin of the input queue 80 instead of the storage amount Lin of the input queue 80 for the determination process. In this case, even if the maximum storage amount is different for each VM 201, it suffices to set a common threshold according to the load state of the VM 201, so the time and effort for setting processing is reduced.

  FIG. 9 is a diagram showing the operation of the management server 1 of the second embodiment. FIG. 9 shows a state before execution of autoscaling. In FIG. 9, the same components as in FIG. 5 are assigned the same reference numerals and descriptions thereof will be omitted.

  Every time the VNF determination unit 100 acquires the storage amount Lin of the input queue 80 from the queue monitoring unit 220, the storage amount Lin of the input queue 80 is registered in the VNF management information of the VM management table 131 as the previous storage amount Lin '. Do. The VNF determination unit 100 calculates the difference ΔLin (= Lin−Lin ′) of the storage amount Lin of the input queue 80 acquired from the queue monitoring unit 220 and the storage amount Lin ′ of the previous input queue 80 acquired from the VM management table 131. calculate. The VNF determination unit 100 uses the difference ΔLin as the increase amount per unit time of the storage amount Lin of the input queue 80 for the determination process. The queue monitoring unit 220 may calculate the increase amount ΔLin per unit time of the storage amount Lin of the input queue 80 and notify the VNF determination unit 100 of the increase amount ΔLin.

  More specifically, in the VNF determination unit 100, the increase amount ΔLin per unit time of the storage amount Lin of the input queue 80 is larger than a predetermined threshold THa ′, and the increase amount ΔTsc per unit time of input traffic is predetermined If the threshold value THb is less than or equal to the threshold value THb, the automatic scaling execution unit 101 is instructed to add a VM. That is, when the increase amount per unit time of the storage amount Lin of the input queue 80 is large when the input traffic is not increasing, the VNF determination unit 100 specifies the corresponding VM 201 as a bottleneck, as shown in FIG. , And instructs to add a new VM 201 to the VNF.

  FIG. 10 is a flow chart showing processing of the management server of the second embodiment. In FIG. 10, the processing common to FIG. 8 is assigned the same reference numeral, and the description thereof is omitted.

  After acquiring the storage amount Lin of the input queue 80 (Step St2), the VNF determining unit 100 acquires the previous storage amount Lin 'of the input queue 80 from the VM management table 131 (Step St2a). At this time, the VNF determination unit 100 searches the storage amount Lin ′ of the corresponding input queue 80 from the VM management table 131 based on the ID of the VM 201.

  Next, the VNF determination unit 100 calculates an increase amount ΔLin per unit time of the storage amount Lin of the input queue 80 (step St2 b). Note that the VNF determination unit 100 may acquire the increase amount ΔLin per unit time of the storage amount Lin of the input queue 80 from the queue monitoring unit 220.

  After acquiring the increase amount ΔTsc per unit time of the input traffic (step St3), the VNF determination unit 100 compares the increase amount ΔLin per unit time of the storage amount Lin of the input queue 80 with a predetermined threshold THa ′ Step St4a). When ΔLin ≦ THa ′ is satisfied (No in step St4a), the VNF determination unit 100 updates the previous storage amount Lin ′ of the VM management table 131 to the newly acquired storage amount Lin (step St8), and performs processing. Finish.

  Further, when Lin> THa 'is satisfied (Yes in step St4a), the VNF determination unit 100 executes the processing in step St5 and the above. Thus, the management server 1 executes the process. The present process is an example of a management method executed by the management server 1.

  As described above, the VNF determination unit 100 compares the increase amount ΔTsc per unit time of input traffic and the increase amount ΔLin per unit time of the storage amount Lin of the input queue 80 with a predetermined threshold THa ′. The auto scaling execution unit 101 is instructed to perform auto scaling. Therefore, the VNF determination unit 100 can suppress unnecessary addition of the VM 201 when the increase amount of the storage amount Lin of the input queue 80 is increased due to the increase of the input traffic. As a result, unnecessary consumption of resources and the setting process of the SC and the load balancer are avoided.

  Furthermore, as described above, the VNF determination unit 100 can use the threshold value THa 'common to the VM 201 for the increase amount ΔLin in the determination process, so the process of setting the threshold value THa' can be reduced.

Third Embodiment
In the first embodiment and the second embodiment, the VNF determination unit 100 executes the determination process for each VM 201, but even if a plurality of VMs 201 on a common SC are specified by a single determination process, the VM 201 of the bottleneck is identified. Good. Thereby, the VNF determination unit 100 can easily identify the VM 201 of the bottleneck for each SC.

  FIG. 11 is a diagram showing the operation of the management server 1 of the third embodiment. FIG. 11 shows a state before execution of autoscaling. In FIG. 11, the same components as in FIG. 5 will be assigned the same reference numerals and descriptions thereof will be omitted.

  In the general-purpose physical server 2, in addition to the load balancer LB # 1 and the firewall FW # 1 similar to those in FIG. 5, the load balancer LB # 2 and the Web proxy PRX # 1 are formed. The load balancer LB # 2 exchanges traffic with the virtual switch (# 1) 210 of the hypervisor 200, and the Web proxy PRX # 1 exchanges with the virtual switch (# 2) 211 of the hypervisor 200. Input and output traffic.

  The web proxy PRX # 1 is provided with an input queue 84 and an output queue 85 for traffic. The input queue 84 stores traffic input from the virtual switch 211 to the Web proxy PRX # 1 in units of packets. The output queue 85 stores the traffic output from the Web proxy PRX # 1 to the virtual switch 211 in units of packets. The input queue 84 and the output queue 85 are examples of queues.

  The code R3 shows the path of SC. Traffic passes through the load balancer LB # 1, the firewall FW # 1, the load balancer LB # 2, and the web proxy PRX # 1 in this order. More specifically, traffic is input from the router 4 to the load balancer LB # 1 via the virtual switch 210. The traffic output from the load balancer LB # 1 is input to the input queue 80 via the virtual switch 210.

  The firewall FW # 1 reads traffic from the input queue 80 on a packet basis and processes it. The firewall FW # 1 writes the processed traffic to the output queue 81 on a packet basis. The virtual switch 210 reads traffic from the output queue 81 in units of packets and outputs the traffic to the load balancer LB # 2. The traffic output from the load balancer LB # 2 is input to the input queue 84 via the virtual switches 210 and 211.

  The Web proxy PRX # 1 reads out traffic from the input queue 84 on a packet basis and processes it. The Web proxy PRX # 1 writes the processed traffic to the output queue 85 on a packet basis. The virtual switch 211 reads the traffic from the output queue 85 in units of packets and outputs the traffic to the router 4. In this way, traffic is forwarded along the SC.

  FIG. 12 is a diagram showing an example of the SC management table 130 and the VM management table 131 after auto scaling. In the SC management table 130, configuration information of the firewall FW # 1 configuring the VNF # 1 and configuration information of the Web proxy PRX # 1 configuring the VNF # 2 are registered. In the VM management table 131, management information of the firewall FW # 1 configuring the VNF # 1 and management information of the Web proxy PRX # 1 configuring the VNF # 2 are registered.

  Thus, a plurality of VMs 201 are formed on the SC of this example.

  The VNF determination unit 100 periodically acquires the storage amount Lin of the input queues 80 and 84 from the queue monitoring unit 220. For example, the ID of the VM 201 is attached to the storage amount Lin of the input queues 80 and 84, and the VNF determination unit 100 manages the storage amount Lin of the input queues 80 and 84 for each VM 201 based on the ID of the VM 201. At this time, the queue monitoring unit 220 may notify the storage amount Lin of the input queues 80 and 84 in accordance with the periodic instruction from the VNF determination unit 100, or may periodically input without the instruction from the VNF determination unit 100. The storage amount Lin of the queues 80 and 84 may be notified.

  The VNF determination unit 100 executes the above-described determination process for each VM 201 on the common SC. At this time, the VNF determination unit 100 determines the determination order of the VM 201 according to the above-described priority. For example, if the priority of the firewall FW # 1 is higher than the priority of the Web proxy PRX # 1, the VNF determination unit 100 first executes the determination processing of the firewall FW # 1, and then determines the Web proxy PRX # 1. Execute the process

  The VNF determination unit 100 executes auto scaling in the auto scaling execution unit 101 based on the increase amount ΔTsc of input traffic per unit time and the comparison result between the storage amount Lin of the input queues 80 and 84 and the predetermined threshold THa. To indicate. That is, in this example, the storage amount Lin of the input queues 80 and 84 is used as a parameter related to the traffic queue. As in the second embodiment, an increase amount ΔLin per unit time of the storage amount Lin of the input queues 80 and 84 may be used as a parameter related to the traffic queue.

  When the VNF determination unit 100 identifies the firewall FW # 1 as the bottleneck VM 201 as a result of the determination process, the VNF determination unit 100 instructs the auto-scaling execution unit 101 to perform auto-scaling of VNF # 1. As a result, a new firewall FW # 2 is added to the VNF # 1.

  FIG. 13 is a diagram showing a state after auto scaling. In FIG. 13, the same components as in FIG. 11 will be assigned the same reference numerals and descriptions thereof will be omitted.

  The auto scaling execution unit 101 adds a new firewall FW # 2 in accordance with the instruction of the VNF determination unit 100. More specifically, the auto-scaling execution unit 101 identifies the VNF # 1 included in the firewall FW # 1 by referring to the VM management table 131, and adds the firewall FW # 2 to the VNF # 1.

  FIG. 14 is a diagram showing an example of the SC management table 130 and the VM management table 131 after auto scaling. After the addition of the firewall FW # 2, the auto-scaling execution unit 101 updates the number of VMs of VNF # 1 in the SC management table 130 from 1 to 2 as indicated by a dotted line frame.

  Further, as shown by the dotted line frame, the auto-scaling execution unit 101 adds the management information of the firewall FW # 2, which is the new VM # 2, to the item of VNF # 1 of the VM management table 131. Thus, the SC management table 130 and the VM management table 131 are updated.

  Referring again to FIG. 13, after the addition of the firewall FW # 2, the auto-scaling execution unit 101 requests the path setting unit 103 to set up the SC via the firewall FW # 2. The path setting unit 103 sets the SC for the load balancer LB # 1, the firewall FW # 2, and the virtual switch 210 in response to the request for setting the SC. As a result, the SC is set to a path through which traffic is distributed from the load balancer LB # 1 to the firewalls FW # 1 and FW # 2 as indicated by a symbol R4.

  The load balancer LB # 1 equally distributes traffic to the firewalls FW # 1 and FW # 2. Therefore, the load on the forwarding process of the firewall FW # 1 is reduced, and the SC bottleneck is eliminated. Therefore, the traffic transfer amount of the entire SC increases.

  FIG. 15 is a flowchart showing processing of the management server 1 of the third embodiment. The VNF determination unit 100 selects an SC to be determined from the SC management table 130 (step St21). At this time, the VNF determination unit 100 selects, for example, the SC designated by the input device 15 by the user.

  Next, the VNF determination unit 100 acquires, from the SC management table 130, the number of VMs N (a positive integer) passed by the SC (Step St22). At this time, the VNF determination unit 100 obtains the number N of VMs by summing up the number m of VMs of each of the VNFs # 1 to #n in the SC management table 130.

  Next, the VNF determination unit 100 sets 1 to a variable k (positive integer) (step St23). The variable k is an identifier that identifies the VM 201 through which the currently selected SC passes. The VNF determination unit 100 executes the determination process of each VM 201 in the order of priority by assigning the variable k to each VM 201 according to the order of priority of the VMs 201.

  Next, the VNF determination unit 100 selects the VM [k] 201 (step St24). VM [k] 201 is the VM 201 to which the variable k is assigned.

  Next, the VNF determination unit 100 acquires the storage amount Lin of the input queues 80 and 84 of the VM [k] 201 from the queue monitoring unit 220 (Step St25). Next, the VNF determination unit 100 acquires an increase amount ΔTsc per unit time of input traffic from the traffic monitoring unit 40 (step St26).

  Next, the VNF determination unit 100 compares the storage amount Lin of the input queues 80 and 84 of the VM [k] 201 with a predetermined threshold THa (step St27). For example, when the variable k = 1 is assigned to the firewall FW # 1 and the variable k = 2 is assigned to the Web proxy PRX # 1, the VNF determination unit 100 stores the input queue 80 in the process of the first step St27. The amount Lin is compared with the predetermined threshold THa, and the storage amount Lin of the input queue 84 is compared with the predetermined threshold THa in the process of the next step St27.

  If Lin ≦ THa is satisfied (No in step St27), the VNF determining unit 100 adds 1 to the variable k (step St31). Next, the VNF determination unit 100 compares the variable k with the number of VMs N (step St32). If k> N is satisfied (Yes in step St32), the VNF determining unit 100 determines that the determination processing of all the VMs 201 via which the SC passes has ended, and ends the processing. Further, when k ≦ N is satisfied (No in Step St32), the VNF determining unit 100 executes the processes after Step St24 again for the other VMs 201.

  When Lin> THa is satisfied (Yes in step St27), the VNF determination unit 100 compares the increase amount ΔTsc per unit time of input traffic with a predetermined threshold THb (step St28). If ΔTsc> THb is satisfied (Yes in step St28), the VNF determining unit 100 ends the process.

  Further, when Tsc ≦ THb is satisfied (No in step St28), the VNF determination unit 100 instructs the auto scaling execution unit 101 to add the VM 201 to the VNF (step St29). Next, the auto-scaling execution unit 101 requests the path setting unit 103 to set SC (Step St30). Thus, the processing of the management server 1 is executed. The present process is an example of a management method executed by the management server 1.

  As described above, the VNF determination unit 100 determines that the increase amount of input traffic per unit time is larger than a predetermined amount among the plurality of VNFs on the common SC, and the storage amount Lin of the input queues 80 and 84 is a predetermined threshold THa. Instruct to add VM 201 to larger VNF. Thus, not only the above-described effect can be obtained, but the VNF determination unit 100 can easily identify the VM 201 of the bottleneck for each SC. The same effect can be obtained when the storage amount Lin of the input queues 80 and 84 is replaced by the increase amount ΔLin per unit time of the storage amount Lin.

Fourth Embodiment
In each of the embodiments described above, the VNF determination unit 100 uses the increase amount ΔTsc of input traffic per unit time for the determination process, but is not limited to this. The VNF determination unit 100 may specify the VM 201 which is a bottleneck based on the parameters related to the input queues 80 and 84 of the VM 201 and the parameters related to the output queues 81 and 85 of the VM 201.

  FIG. 16 is a diagram showing the operation of the management server 1 of the fourth embodiment. FIG. 16 shows a state before execution of autoscaling. In FIG. 16, the same components as in FIG. 5 will be assigned the same reference numerals and descriptions thereof will be omitted.

  The queue monitoring unit 220 monitors the input queue 80 and the output queue 81, and periodically notifies the VNF determination unit 100 of the storage amount Lin of the input queue 80 and the storage amount Lout of the output queue 81. Thus, the VNF determination unit 100 periodically acquires the storage amount Lin of the input queue 80 and the storage amount Lout of the output queue 81 from the queue monitoring unit 220.

  For example, the ID of the corresponding VM 201 is assigned to the storage amount Lin of the input queue 80 and the storage amount Lout of the output queue 81, and the VNF determination unit 100 stores the storage amount Lin of the input queue 80 and the storage of the output queue 81. The amount Lout is managed for each VM 201 based on the ID of the VM 201. At this time, the queue monitoring unit 220 may notify the storage amount Lin of the input queue 80 and the storage amount Lout of the output queue 81 according to the periodic instruction from the VNF determination unit 100, or the instruction from the VNF determination unit 100. Instead, the storage amount Lin of the input queue 80 and the storage amount Lout of the output queue 81 may be notified periodically.

  The VNF determination unit 100 automatically performs the auto scaling execution unit 101 based on the comparison result of the storage amount Lin of the input queue 80 and the predetermined threshold THa, and the comparison result of the storage amount Lout of the output queue 81 and the predetermined threshold THd. Instructs execution of scaling. That is, in this example, the storage amount Lin of the input queue 80 is used as a parameter related to the input queue 80, and the storage amount Lout of the output queue 81 is used as a parameter related to the output queue 81.

  More specifically, when the storage amount Lin of the input queue 80 is larger than the predetermined threshold THa and the storage amount Lout of the output queue 81 is the predetermined threshold THd or less, the VNF determination unit 100 adds VM. It instructs the auto scaling execution unit 101. That is, when the storage amount Lin of the input queue 80 of the VM 201 is large but the storage amount Lout of the output queue 81 is small, the VNF determination unit 100 identifies the VM 201 as a bottleneck and newly adds the VNF as shown in FIG. To add a new VM.

  FIG. 17 is a flowchart showing processing of the management server 1 of the fourth embodiment. In FIG. 17, the processing common to FIG. 8 is given the same reference numeral, and the description thereof is omitted.

  After acquiring the storage amount Lin of the input queue 80 (step St2), the VNF determining unit 100 acquires the storage amount Lout of the output queue 81 from the queue monitoring unit 220 (step St2c). When Lin> THa is satisfied (Yes in step St4), the VNF determination unit 100 compares the storage amount Lout of the output queue 81 with a predetermined threshold THd (step St5a). At this time, the VNF determination unit 100 may obtain, from the queue monitoring unit 220, the result of the comparison processing performed by the queue monitoring unit 220.

  When Lout> THd is satisfied (Yes in step St5a), the VNF determination unit 100 inputs a large amount of traffic (packets) to the VM 201, but also a large amount of traffic transferred by the VM 201. The process ends without determining that it is a neck. Further, when Lout ≦ THd is satisfied (No in step St5a), the VNF determination unit 100 determines that the VM 201 is a bottleneck because the traffic transferred by the VM 201 is small, and the auto scaling execution unit 101 transmits the VNF to the VNF. The addition of the VM 201 is instructed (step St6).

  Next, the auto-scaling execution unit 101 requests the path setting unit 103 to set SC (Step St7). Thus, the processing of the management server 1 is executed. The present process is an example of a management method executed by the management server 1.

  Thus, the VNF determination unit 100 performs autoscaling based on the comparison result of the storage amount Lin of the input queue 80 and the predetermined threshold THa, and the comparison result of the storage amount Lout of the output queue 81 and the predetermined threshold THd. It instructs the unit 101 to execute auto scaling. Therefore, the VNF determination unit 100 can reliably identify the VM 201 of the bottleneck based on the states of the input queue 80 and the output queue 81. For this reason, addition of unnecessary resources is suppressed and auto scaling is appropriately performed.

  Note that the VNF determination unit 100 may use the increase amount ΔLin per unit time of the storage amount Lin instead of the storage amount Lin of the input queue 80 for the determination processing, or store it instead of the storage amount Lout of the output queue 81. The increase amount ΔLout per unit time of the amount Lout may be used in the determination process. Also in this case, the same effect as described above can be obtained.

Fifth Embodiment
In the fourth embodiment, the VNF determination unit 100 executes the determination processing for each VM 201. However, as in the third embodiment, the VM 201 of the bottleneck is specified from a plurality of VMs 201 on a common SC by one determination processing. You may Thereby, the VNF determination unit 100 can easily identify the VM 201 of the bottleneck for each SC.

  FIG. 18 is a diagram showing the operation of the management server 1 of the fifth embodiment. FIG. 18 shows a state before execution of autoscaling. In FIG. 18, the same components as in FIG. 11 will be assigned the same reference numerals and descriptions thereof will be omitted.

  The queue monitoring unit 220 monitors the input queues 80 and 84 and the output queues 81 and 85, and periodically notifies the VNF determination unit 100 of the storage amount Lin of the input queues 80 and 84 and the storage amount Lout of the output queues 81 and 85. Do. As a result, the VNF determination unit 100 periodically acquires the storage amount Lin of the input queues 80 and 84 and the storage amount Lout of the output queues 81 and 85 from the queue monitoring unit 220.

  For example, the ID of the corresponding VM 201 is assigned to the storage amount Lin of the input queues 80 and 84 and the storage amount Lout of the output queues 81 and 85, and the VNF determination unit 100 determines the storage amount Lin of the input queues 80 and 84. The storage amount Lout of the output queues 81 and 85 is managed for each VM 201 based on the ID of the VM 201. At this time, the queue monitoring unit 220 may notify the storage amount Lin of the input queues 80 and 84 and the storage amount Lout of the output queues 81 and 85 according to the periodic instruction from the VNF determination unit 100, or the VNF determination unit The storage amount Lin of the input queues 80 and 84 and the storage amount Lout of the output queues 81 and 85 may be notified periodically without an instruction from 100.

  The VNF determination unit 100 executes the same determination processing as that of the fourth embodiment for each VM 201 on the common SC. At this time, the VNF determination unit 100 determines the determination order of the VM 201 according to the above-described priority. For example, if the priority of the firewall FW # 1 is higher than the priority of the Web proxy PRX # 1, the VNF determination unit 100 first executes the determination processing of the firewall FW # 1, and then determines the Web proxy PRX # 1. Execute the process

  The VNF determination unit 100 executes auto scaling based on the comparison result of the storage amount Lin of the input queues 80 and 84 and the predetermined threshold THa, and the comparison result of the storage amount Lout of the output queues 81 and 85 and the predetermined threshold THd. It instructs the unit 101 to execute auto scaling. That is, in this example, the storage amount Lin of the input queues 80 and 84 is used as a parameter related to the input queues 80 and 84, and the storage amount Lout of the output queues 81 and 85 is used as a parameter related to the output queues 81 and 85.

  As in the second embodiment, an increase amount ΔLin per unit time of the storage amount Lin of the input queues 80 and 84 may be used as a parameter related to the input queues 80 and 84. Further, an increase amount ΔLout per unit time of the storage amount Lout of the output queues 81 and 85 may be used as a parameter related to the output queues 81 and 85.

  When the VNF determination unit 100 identifies the firewall FW # 1 as the bottleneck VM 201 as a result of the determination process, the VNF determination unit 100 instructs the auto-scaling execution unit 101 to perform auto-scaling of VNF # 1. As a result, as shown in FIG. 13, a new firewall FW # 2 is added to the VNF # 1.

  FIG. 19 is a flowchart showing processing of the management server 1 of the fifth embodiment. In FIG. 19, the same processing as in FIG. 15 is denoted by the same reference numeral, and the description thereof is omitted.

  After the VNF determination unit 100 acquires the storage amount Lin of the input queues 80 and 84 of the VM [k] 201 from the queue monitoring unit 220 (step St 25), the VNF determination unit 100 outputs the output queue 81 of the VM [k] 201 from the queue monitoring unit 220, The storage amount Lout of 85 is acquired (step St26a). When Lin> THa is satisfied (Yes in step St27), the VNF determination unit 100 compares the storage amount Lout of the output queues 81 and 85 with a predetermined threshold THd (step St28a).

  When Lout> THd is satisfied (Yes in step St28a), the VNF determination unit 100 receives a lot of traffic (packets) in the VM [k] 201, but the transfer processing is performed by the VM [k] 201 Since there is also a lot of traffic, each process after step St31 is executed without determining VM [k] 201 as a bottleneck. When Lout ≦ THd is satisfied (No in step St28a), the VNF determination unit 100 determines that VM [k] 201 is a bottleneck because the traffic transferred by VM [k] 201 is small. The scaling execution unit 101 is instructed to add the VM 201 to the VNF (step St29).

  Thus, the processing of the management server 1 is executed. The present process is an example of a management method executed by the management server 1.

  As described above, among the plurality of VNFs on the common SC, the VNF determination unit 100 determines that the storage amount Lin of the input queues 80 and 84 is larger than the predetermined threshold THa, and the storage amount Lout of the output queues 81 and 85 is predetermined. The addition of the VM 201 to the VNF including the VM 201 which is equal to or less than the threshold THd is instructed. Thus, not only the above-described effect can be obtained, but the VNF determination unit 100 can easily identify the VM 201 of the bottleneck for each SC.

  The same effect can be obtained when the storage amount Lin of the input queues 80 and 84 is replaced by the increase amount ΔLin per unit time of the storage amount Lin. Also, the same effect can be obtained when the increase amount ΔLout per unit time of the storage amount Lout is used instead of the storage amount Lout of the output queues 81 and 85.

Sixth Embodiment
So far, the VNF determination unit 100 instructs to execute auto scaling based on the result of specifying the VM 201 of the bottleneck, but when the virtual switches 210 and 211 are bottlenecks, even if the auto scaling of VNF is performed Traffic transfer volume is not improved. Therefore, the VNF determination unit 100 identifies not only the VM 201 but also the virtual switches 210 and 211 as bottlenecks, and when the VM 201 is a bottleneck, instructs to execute auto scaling, and the virtual switches 210 and 211 are bottlenecks. In some cases, migration may be instructed. An execution example of migration will be described below.

  FIG. 20 and FIG. 21 are diagrams showing the operation of the management server 1 of the sixth embodiment. FIG. 20 shows a state before execution of migration, and FIG. 21 shows a state after execution of migration. In FIGS. 20 and 21, the same components as those in FIGS. 5 and 6 are denoted by the same reference numerals, and the description thereof will be omitted.

  Referring to FIG. 20, in the general-purpose physical server 2, a firewall FW # 1 is formed as the VM 201. The firewall FW # 1 exchanges traffic with the virtual switch (# 1) 210 of the hypervisor 200.

  The SC of this example passes through the firewall FW # 1 and the virtual switch 210. Traffic is input from router 4 to input queue 80 via virtual switch 210.

  The firewall FW # 1 reads traffic from the input queue 80 on a packet basis and processes it. The firewall FW # 1 writes the processed traffic to the output queue 81 on a packet basis. The virtual switch 210 reads the traffic from the output queue 81 in units of packets and transfers the traffic to the router 4. In this way, traffic is forwarded along the SC.

  The VNF determination unit 100 periodically acquires, from the queue monitoring unit 220, the elements included in the SC, that is, the storage amounts Lin of the input queues of the firewall FW # 1 and the virtual switch 210. More specifically, the VNF determination unit 100 acquires the storage amount Lin of the input queue 80 of the firewall FW # 1 and the storage amount Lin (corresponding to Lout) of the output queue 81 corresponding to the input queue of the virtual switch 210. . In the following description, the output queue 81 is described as the input queue 81a of the virtual switch 210.

  Further, the VNF determination unit 100 acquires, from the traffic monitoring unit 40, an increase amount ΔTsc per unit time of input traffic. The VNF determination unit 100 executes autoscaling based on the increase amount ΔTsc per unit time of input traffic, and the comparison result of the storage amount Lin of the input queue 80 or 81a of each element of SC and a predetermined threshold THe. Instruct to execute migration.

  More specifically, when the increase amount ΔTsc of input traffic per unit time is equal to or less than a predetermined threshold value THb, the VNF determination unit 100 determines that the storage amount Lin of the input queues 80 and 81a is larger than the predetermined threshold value THe. Identify the element as a bottleneck. The VNF determination unit 100 instructs the auto scaling execution unit 101 to execute auto scaling when the bottleneck element is the VM 201, and migrates to the migration execution unit 102 when the bottleneck element is the virtual switches 210 and 211. Direct the execution of

  In this example, the increase amount ΔTsc of input traffic per unit time is equal to or less than a predetermined threshold THb, the storage amount Lin of the input queue 80 is equal to or less than a predetermined threshold THe, and the storage amount Lin of the input queue 81a is a predetermined threshold. Assume that it is more than THe. Therefore, the VNF determination unit 100 identifies the virtual switch 210 as a bottleneck and migrates the firewall FW # 1 from the virtual switch (# 1) 210 to another virtual switch (# 2) 211 as the migration execution unit 102. Instruct

  Thus, the firewall FW # 1 is transferred from the virtual switch 210 to another virtual switch 211 as shown in FIG. At this time, the migration execution unit 102 searches the virtual switch management table 132 for the virtual switch 211 having a load equal to or less than a predetermined value.

  After migrating the firewall FW # 1, the migration execution unit 102 requests the path setting unit 103 to set the SC. The path setting unit 103 sets the SC in the virtual switch 211 and the firewall FW # 1.

  The path setting unit 103 updates the VM management table 131 in accordance with the setting of the SC. In the VM management table 131, an identifier of the virtual switch 210 is registered as management information of the VM 201.

  FIG. 22 is a diagram showing an example of the VM management table 131 before and after migration. In the VM management table 131, as management information of the VM 201, a virtual SW-ID for identifying the virtual switch 210 of the output destination of the traffic of the VM 201 is registered. Before migration, the virtual SW-ID of the VM # 1 (firewall FW # 1) is “SWa” (see FIG. 4) indicating the virtual switch (# 1) 210.

  Since the firewall FW # 1 is connected to the virtual switch (# 2) 211 after migration, the virtual SW-ID is “SWb” indicating the virtual switch (# 2) 211 as shown by the dotted line frame (FIG. Is updated). Thus, the VM management table 131 is updated.

  As described above, the VNF determining unit 100 executes or migrates autoscaling based on the increase amount ΔTsc of input traffic per unit time, and the comparison result of the storage amount Lin of the input queues 80 and 81a and the predetermined threshold THe. Direct the execution of As a result, bottlenecks in the traffic forwarding process are eliminated, and the SC traffic forwarding volume increases.

  FIG. 23 is a flowchart showing processing of the management server 1 of the sixth embodiment. The VNF determination unit 100 selects an SC to be determined from the SC management table 130 (step St41). At this time, the VNF determination unit 100 selects, for example, the SC designated by the input device 15 by the user.

  Next, the VNF determination unit 100 acquires the number of elements H (a positive integer) included in the SC from the SC management table 130 (Step St42). At this time, the VNF determination unit 100 sums the number of elements H by summing the number of VMs m of each of the VNFs # 1 to #n in the SC management table 130 and the number of virtual SW-IDs of the corresponding VM 201 in the VM management table 131. get.

  Next, the VNF determination unit 100 sets 1 to a variable j (positive integer) (step St43). The variable j is an identifier that identifies an element (VM 201, virtual switches 210 and 211) included in the SC being selected. The VNF determination unit 100 assigns the variable j to each VM 201 in the order of the priorities of the VMs 201 and the virtual switches 210 and 211, thereby executing the determination process of each VM 201 and the virtual switches 210 and 211 in the order of the priorities.

  Next, the VNF determination unit 100 selects the element [j] (step St44). The element [j] is the VM 201 or the virtual switches 210 and 211 to which the variable j is assigned.

  Next, the VNF determination unit 100 acquires the storage amount Lin of the input queue 80 or 81a of the element [j] from the queue monitoring unit 220 (Step St45). Next, the VNF determination unit 100 acquires an increase amount ΔTsc per unit time of input traffic from the traffic monitoring unit 40 (Step St46).

  Next, the VNF determination unit 100 compares the storage amount Lin of the input queue 80 or 81a of the element [j] with a predetermined threshold THe (step St47). For example, when the variable j = 1 is assigned to the firewall FW # 1 and the variable j = 2 is assigned to the virtual switch 210, the VNF determination unit 100 determines the storage amount Lin of the input queue 80 in the process of the first step St47. Is compared with a predetermined threshold THe, and the storage amount Lin of the input queue 81a is compared with the predetermined threshold THe in the process of the next step St47.

  If Lin ≦ THe is satisfied (No in operation St47), the VNF determining unit 100 adds 1 to the variable j (operation St54). Next, the VNF determination unit 100 compares the variable j with the number of elements H (step St55). If j> H is satisfied (Yes in step St55), the VNF determining unit 100 determines that the process of determining all the elements included in the SC is completed, and ends the process. In addition, when j ≦ H is satisfied (No in Operation St55), the VNF determining unit 100 executes the processing after Operation St44 again for the other elements.

  When Lin> THe is satisfied (Yes in step St47), the VNF determination unit 100 compares the increase amount ΔTsc per unit time of input traffic with a predetermined threshold THb (step St48). If ΔTsc> THb is satisfied (Yes in step St48), the VNF determination unit 100 ends the process.

  When Tsc ≦ THb is satisfied (No in operation St48), the VNF determining unit 100 determines whether the element [j] is the VM 201 (operation St49). When the element [j] is the VM 201 (Yes in Operation St49), the VNF determining unit 100 instructs the auto-scaling execution unit 101 to add the VM 201 to the VNF (Operation St50). Next, the auto-scaling execution unit 101 requests the path setting unit 103 to set SC (step St51).

  When the element [j] is the virtual switch 210 or 211 (No in operation St49), the VNF determining unit 100 instructs the migration execution unit 102 to transfer the corresponding VNF (operation St52). Next, the migration execution unit 102 requests the path setting unit 103 to set the SC (Step St53). Thus, the processing of the management server 1 is executed. The present process is an example of a management method executed by the management server 1.

  As described above, the VNF determination unit 100 determines that the increase amount ΔTsc per unit time of input traffic is larger than a predetermined amount and the storage amount Lin of the input queues 80 and 81a is the predetermined threshold THe among the plurality of elements of the common SC. Instructs VNF auto-scaling or migration of larger elements to be performed. Therefore, not only when the bottleneck is the VM 201 but also when the bottleneck is the virtual switches 210 and 211, it is possible to increase the transfer amount of SC traffic.

Seventh Embodiment
In the sixth embodiment, the VNF determination unit 100 uses the storage amount Lin of the input queue 80, 81a of the element of SC for the bottleneck determination process, but instead, the storage amount Lin is increased per unit time The quantity ΔLin may be used. In this case, the VNF determination unit 100 uses the increase amount per unit time of the storage amount Lout of the input queue 81a, that is, the output queue 81, as a parameter related to the traffic queue. In this case, as in the second embodiment, the management server 1 holds the storage amount Lin ′ of the previous input queue 80, 81a.

  FIG. 24 is a diagram showing an example of the VM management table 131 and the virtual SW management table 132 of the seventh embodiment. In the VM management table 131, the storage amount Lin 'of the input queue 80 of VM # 1 is registered as management information of VM # 1. Further, in the virtual SW management table 132, the storage amount Lin 'of the input queue 81a is registered for each of the virtual switches 210 and 211.

  The VNF determination unit 100 sets the storage amount Lin ′ registered in the VM management table 131 and the virtual SW management table 132 as the storage amount Lin ′ of the previous input queue 80 or 81 a, and increases the storage amount Lin per unit time. Used to calculate ΔLin.

  FIG. 25 is a flowchart showing processing of the management server 1 of the seventh embodiment. In FIG. 25, the same processing as that in FIG. 23 is denoted by the same reference numeral, and the description thereof is omitted.

  After acquiring the storage amount Lin of the input queue 80, 81a of the element [j] from the queue monitoring unit 220 (step St45), the VNF determination unit 100 acquires the previous input queue 80 from the VM management table 131 or the virtual SW management table 132. , 81a are acquired (step St45a). At this time, when the element [j] being selected is the VM 201, the VNF determination unit 100 acquires the storage amount Lin ′ from the VM management table 131, and the element [j] being selected is the virtual switch 210, 211. In the case, the storage amount Lin ′ is acquired from the virtual SW management table 132.

  Next, the VNF determination unit 100 calculates an increase amount ΔLin (= Lin−Lin ′) per time of the storage amount Lin of the input queue 80, 81a (Step St45b). After the VNF determination unit 100 acquires the increase amount ΔTsc of input traffic per unit time from the traffic monitoring unit 40 (Step St46), the increase amount ΔLin per time of the storage amount Lin of the input queues 80 and 81a is set to a predetermined threshold It compares with THf (step St47a).

  If ΔLin ≦ THf is satisfied (No in step St47a), the VNF determining unit 100 executes the processing in step St54 and subsequent steps to select the next element. Further, when Lin> THf is satisfied (Yes in step St47a), the VNF determining unit 100 executes each process after step St48 described above.

  In addition, after setting SC (steps St51 and St53), the VNF determination unit 100 updates the storage amount Lin ′ of the VM management table 131 or the virtual SW management table 132 with the storage amount Lin acquired from the queue monitoring unit 220 (step St 54). Thus, the processing of the management server 1 is executed. The present process is an example of a management method executed by the management server 1.

  As described above, the VNF determination unit 100 determines that the increase amount ΔTsc per unit time of input traffic is larger than a predetermined amount among the plurality of elements of the common SC, and the storage amount Lin of the input queue 80 or 81a per unit time. It instructs to execute auto-scaling or migration of VNF of an element whose increase amount ΔLin is larger than a predetermined threshold THf. For this reason, the same effect as that of the sixth embodiment can be obtained.

  More specifically, when it is determined that the virtual switch 210 is a bottleneck as in the examples of FIGS. 20 and 21, the VNF determination unit 100 instructs the execution of migration. That is, the VNF determination unit 100 instructs transfer of VNF based on the increase amount ΔTsc of input traffic per unit time and the comparison result of the storage amount Lout of the output queue 81 of the VM 201 and the predetermined threshold THf. Therefore, even when the VM 201 is not a bottleneck, the management server 1 can increase the SC transfer amount by transferring the VNF of the VM 201.

  More specifically, the VNF determination unit 100 transfers the VNF from the virtual switch 210 to another virtual switch 211 when the increase amount per unit time of the storage amount Lout of the output queue 81 of the VM 201 is larger than a predetermined threshold THf. The migration execution unit 102 is instructed to do this. Therefore, when the virtual switch 210 is a bottleneck of transfer processing, the management server 1 can increase the transfer amount of SC by relocating the VNF.

Eighth embodiment
In the sixth embodiment, the VNF determination unit 100 uses the increase amount ΔTsc of input traffic per unit time and the storage amount Lin of the input queue 80 for determination processing, but instead of the increase amount ΔTsc, storage of the output queue 81 The amount Lout may be used in the determination process. That is, the VNF determination unit 100 may instruct auto scaling or migration based on the parameters related to the input queue 80 and the parameters related to the output queue 81.

  FIG. 26 is a diagram showing the operation of the management server 1 of the eighth embodiment. In FIG. 26, the same components as in FIG. 20 will be assigned the same reference numerals and descriptions thereof will be omitted.

  The VNF determination unit 100 acquires the storage amount Lin of the input queue 80 and the storage amount Lout of the output queue 81 from the queue monitoring unit 220. The VNF determination unit 100 compares the storage amount Lin of the input queue 80 with a predetermined threshold THa, and compares the storage amount Lout of the output queue 81 with a predetermined threshold THd. The VNF determination unit 100 identifies a bottleneck in traffic transfer processing based on each comparison result. The identified bottleneck is the VM 201 and the virtual switches 210 and 211 as in the sixth embodiment.

  When the bottleneck is the VM 201, the VNF determination unit 100 instructs the auto scaling unit 101 to perform auto scaling of VNF. In this case, for example, as shown in FIG. 6, a new firewall FW # 2 is added, and traffic is equally distributed to the firewalls FW # 1 and FW # 2. This increases traffic forwarding volume.

  Further, when the bottleneck is the virtual switch 210, the VNF determination unit 100 instructs the migration execution unit 102 to migrate the VNF. In this case, for example, as illustrated in FIG. 21, the firewall FW # 1 is transferred from the virtual switch 210 to another virtual switch 211. This increases traffic forwarding volume.

  FIG. 27 is a flowchart showing processing of the management server 1 of the eighth embodiment. In FIG. 27, the processing common to FIG. 17 is assigned the same reference numeral, and the description thereof is omitted.

  When Lin ≦ THa is satisfied (No in step St4), the VNF determination unit 100 compares the storage amount Lout of the output queue 81 with a predetermined threshold THd (step St9). If Lout ≦ THd is satisfied (No in step St9), the VNF determination unit 100 ends the process.

  If Lout> THd is satisfied (Yes in step St9), the VNF determination unit 100 determines that the virtual switch 210 is a bottleneck, and instructs the migration execution unit 102 to transfer the corresponding VNF (step St10). Thus, the migration execution unit 102 transfers the VM 201 from the virtual switch 210 to another virtual switch 211.

  Next, the migration execution unit 102 requests the path setting unit 103 to set the SC (Step St11). Thus, the processing of the management server 1 is executed. The present process is an example of a management method executed by the management server 1.

  As described above, the VNF determination unit 100 instructs autoscaling or migration based on the parameters related to the input queue 80 and the parameters related to the output queue 81. More specifically, the VNF determination unit 100 instructs autoscaling or migration based on the storage amount Lin of the input queue 80 and the storage amount Lout of the output queue 81.

  Therefore, the VNF determination unit 100 can reliably identify the VM 201 or the virtual switch 210 or 211 of the bottleneck based on the states of the input queue 80 and the output queue 81. For this reason, addition of unnecessary resources is suppressed, and auto scaling or migration is appropriately performed.

  Note that the VNF determination unit 100 may use the increase amount ΔLin per unit time of the storage amount Lin instead of the storage amount Lin of the input queue 80 for the determination processing, or store it instead of the storage amount Lout of the output queue 81. The increase amount ΔLout per unit time of the amount Lout may be used in the determination process. Also in this case, the same effect as described above can be obtained.

  The embodiments described above are examples of preferred implementations of the invention. However, the present invention is not limited to this, and various modifications can be made without departing from the scope of the present invention.

The following appendices will be further disclosed in connection with the above description.
(Supplementary Note 1) A function management unit that adds a virtual machine to a virtual network function unit formed in a physical server or relocates the virtual network function unit;
The increase amount per unit time of the traffic input to the transfer path passing through the virtual network function unit, the parameter regarding the queue of the traffic provided in the virtual machine configuring the virtual network function unit, and a predetermined threshold value A management apparatus comprising: an instruction unit that instructs the function management unit to add a virtual machine to the virtual network function unit or transfer the virtual network function unit based on a comparison result.
(Supplementary Note 2) The management apparatus according to Supplementary note 1, wherein the parameter is a storage amount of a queue storing the traffic input to a virtual machine configuring the virtual network function unit.
(Supplementary Note 3) The parameter is an increase amount per unit time of a storage amount of a queue storing the traffic input to a virtual machine configuring the virtual network function unit. Management device.
(Supplementary Note 4) In the physical server, a plurality of virtual network function units are formed on the transfer path,
Among the plurality of virtual network function units, the instruction unit adds a virtual machine to the virtual network function unit in which the increase amount of the traffic per unit time is larger than a predetermined amount and the parameter is larger than the predetermined threshold. The management apparatus according to any one of appendices 1 to 3, wherein the function management unit is instructed.
(Supplementary Note 5) The parameter is an increase amount per unit time of the storage amount of the queue storing the traffic output from the virtual machine configuring the virtual network function unit. Management device.
(Supplementary Note 6) The physical server is provided with a virtual switch that reads the traffic from the queue and transfers the traffic.
The instruction unit is configured to instruct the function management unit to transfer the virtual network function unit from the virtual switch to another virtual switch when the increase amount is larger than the predetermined threshold. Management device described in.
(Supplementary Note 7) A function management unit that adds a virtual machine to a virtual network function unit formed in a physical server or transfers the virtual network function unit.
The function managing unit may be configured to, based on a parameter related to an input queue storing traffic input to a virtual machine configuring the virtual network function unit and a parameter related to an output queue storing the traffic output from the virtual machine, A management apparatus comprising: an instruction unit instructing addition of a virtual machine to a virtual network function unit or transfer of the virtual network function unit.
(Supplementary Note 8) The management apparatus according to Supplementary note 7, wherein the parameter related to the input queue is the storage amount of the input queue, and the parameter related to the output queue is the storage amount of the output queue.
(Supplementary Note 9) An increase amount of traffic per unit time input to a transfer route via a virtual network function unit formed in a physical server, and the traffic provided in a virtual machine configuring the virtual network function unit A management method characterized by performing addition of a virtual machine to the virtual network function unit or relocation of the virtual network function unit based on a comparison result of a parameter related to a queue and a predetermined threshold value.
(Supplementary Note 10) The management method according to Supplementary note 9, wherein the parameter is a storage amount of a queue storing the traffic input to a virtual machine configuring the virtual network function unit.
(Supplementary Note 11) The parameter is an increase amount per unit time of the storage amount of the queue storing the traffic input to the virtual machine configuring the virtual network function unit. Management method.
(Supplementary Note 12) In the physical server, a plurality of virtual network function units are formed on the transfer path,
Among the plurality of virtual network function units, a virtual machine is added to the virtual network function unit in which the increase amount of the traffic per unit time is larger than a predetermined amount and the parameter is larger than the predetermined threshold. The management method according to any one of appendices 9 to 11.
(Supplementary Note 13) The parameter is an increase amount per unit time of the storage amount of the queue storing the traffic output from the virtual machine configuring the virtual network function unit. Management method.
(Supplementary Note 14) The physical server is provided with a virtual switch that reads the traffic from the queue and transfers the traffic.
The management method according to appendix 13, wherein the virtual network function unit is transferred from the virtual switch to another virtual switch when the increase amount is larger than the predetermined threshold.
(Supplementary Note 15) For a parameter related to an input queue storing traffic input to a virtual machine forming a virtual network function unit formed in a physical server and a parameter related to an output queue storing the traffic output from the virtual machine And adding the virtual machine to the virtual network function unit or transferring the virtual network function unit based on the management method.
(Supplementary note 16) The management method according to supplementary note 15, wherein the parameter related to the input queue is the storage amount of the input queue, and the parameter related to the output queue is the storage amount of the output queue.

1 management server 2 general-purpose physical server 4 router 10, 20 CPU
80, 81a, 82, 84 input queue 81, 83, 85 output queue 91 to 93 VNF
100 VNF determination unit 101 auto scaling execution unit 102 migration execution unit 201 VM
Lin, Lout storage amount

Claims (9)

A function management unit that adds a virtual machine to a virtual network function unit formed in a physical server or relocates the virtual network function unit;
The increase amount per unit time of the traffic input to the transfer path passing through the virtual network function unit, the parameter regarding the queue of the traffic provided in the virtual machine configuring the virtual network function unit, and a predetermined threshold value A management apparatus comprising: an instruction unit that instructs the function management unit to add a virtual machine to the virtual network function unit or transfer the virtual network function unit based on a comparison result.   The management apparatus according to claim 1, wherein the parameter is a storage amount of a queue storing the traffic input to a virtual machine configuring the virtual network function unit.   The management apparatus according to claim 1, wherein the parameter is an increase amount per unit time of a storage amount of a queue storing the traffic input to a virtual machine configuring the virtual network function unit. In the physical server, a plurality of the virtual network function units are formed on the transfer path,
Among the plurality of virtual network function units, the instruction unit adds a virtual machine to the virtual network function unit in which the increase amount of the traffic per unit time is larger than a predetermined amount and the parameter is larger than the predetermined threshold. The management apparatus according to any one of claims 1 to 3, wherein the function management unit is instructed.   The management apparatus according to claim 1, wherein the parameter is an increase amount per unit time of a storage amount of a queue storing the traffic output from a virtual machine configuring the virtual network function unit. The physical server is formed with a virtual switch that reads the traffic from the queue and transfers it.
The instruction unit is configured to instruct the function management unit to transfer the virtual network function unit from the virtual switch to another virtual switch when the increase amount is larger than the predetermined threshold value. The management device according to 5. A function management unit that adds a virtual machine to a virtual network function unit formed in a physical server or relocates the virtual network function unit;
The function managing unit may be configured to, based on a parameter related to an input queue storing traffic input to a virtual machine configuring the virtual network function unit and a parameter related to an output queue storing the traffic output from the virtual machine, A management apparatus comprising: an instruction unit instructing addition of a virtual machine to a virtual network function unit or transfer of the virtual network function unit.   An increase amount per unit time of traffic input to a transfer path via a virtual network function unit formed in a physical server, and a parameter related to a queue of the traffic provided in a virtual machine configuring the virtual network function unit And adding a virtual machine to the virtual network function unit or relocating the virtual network function unit based on the comparison result of the threshold value and the predetermined threshold value.   The virtual, based on parameters relating to an input queue storing traffic input to a virtual machine forming a virtual network function unit formed in a physical server, and parameters relating to an output queue storing the traffic output from the virtual machine A management method comprising: adding a virtual machine to a network function unit or transferring the virtual network function unit.

Images