KR20160025926A - Apparatus and method for balancing load to virtual application server - Google Patents

Abstract

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus and a method for distributing load among application servers operating in a network-connected computer-based service environment.
A load balancing method for a load balancing apparatus according to the present invention is a method for balancing load on virtual application servers included in physical servers in a load balancing system, Determining whether the number of virtual application servers operating in the load balancing system is appropriate according to the information and generating and changing or deleting one or more virtual application servers according to a result of the determination; The state information indicates the size or amount of load the load distribution system is receiving and processing.

Description

[0001] APPARATUS AND METHOD FOR BALANCING LOAD TO VIRTUAL APPLICATION SERVER [0002]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus and method for distributing a load to virtual application servers, and more particularly, to an apparatus and a method for distributing a load to virtual application servers operating in a network-

A variety of services based on client-server architecture are being provided for various network environments, such as the Internet, corporate networks or home networks. In such a client-server-based service, when many computing and network resources are required for providing a service to a large number of clients or providing a service, the size of the server is dynamically changed to maintain the good quality and performance of the service Can be used.

In this dynamic control scheme, there are typically a scale-up scheme for increasing the resources of the application servers participating in the service and a scale-out scheme for increasing the number of application servers participating in the service .

In the scale-out method, load balancing is performed through a load balancer composed of hardware or software based on Layer-4 or Layer-7 information of OSI (Open Systems Interconnection) model. At this time, the load balancing apparatus appropriately distributes the client's service request to the application servers in the service cluster according to the preset load distribution algorithm, thereby preventing the load from being concentrated on the specific application server and improving the overall service quality.

1 is a block diagram illustrating such a general load balancing system 10. 1, the load balancing system 10 includes clients 11, a network 14, a load balancer 12, and a service cluster 13. When a service request directed to a specific Internet Protocol (VIP) is received from a plurality of clients 11_1, 11_2, and 11_n via a network, the load distributing apparatus 12 transmits the received service request To the application servers 13_1, 13_2, and 13_n in the service cluster 13. [ At this time, when the load of the application servers 13_1, 13_2, 13_n in the service cluster 13 explosively increases or exceeds a predetermined reference value due to the increase of the received service request, the load distributing apparatus 12 performs the scale- The processing capacity of the service cluster 13 can be increased.

However, in this conventional scale-out method, an application server to process a service request of the client 11 must be set and registered in the service cluster 13 in advance. Therefore, if the pre-prediction of the overall service demand is inaccurate or the service requests are instantaneously concentrated, a problem arises that proper service provision can not be provided only by the application servers previously set and registered in the service cluster 13. [ That is, if a number of application servers are registered in the service cluster 13 more than necessary, unnecessary resource waste and cost may be caused due to an excessive configuration.

It is an object of the present invention to provide a load balancer and a method for appropriately controlling the processing capacity of a service cluster according to the state of the load balancing service without external intervention.

It is another object of the present invention to provide a load balancer and a method for minimizing resource waste and cost due to over-configuration of a service cluster.

It is still another object of the present invention to provide a load balancer and a method for continuously providing a service without deteriorating service quality in a service environment with a large fluctuation of load amount over time.

A load balancing method of a load balancing apparatus that distributes a load to virtual application servers included in physical servers in a load balancing system according to embodiments of the present invention includes a state information of the load balancing system, Determining whether the number of virtual application servers operating in the load balancing system is appropriate according to state information of the virtual application servers; And generating and deleting one or more virtual application servers according to the determination result, wherein the status information of the load distribution system includes at least one of a size and a quantity of the load that the load distribution system is receiving and processing .

As an embodiment, the one or more virtual application servers may be software entities operating in a virtualization environment provided by the hypervisor of the physical server, processing service requests provided by the clients, and in the virtualization environment of the physical server And the one or more virtual application servers and the other virtual application servers may operate using different operating systems.

In an embodiment, the step of determining whether the number of virtual application servers to be operated is appropriate may include comparing a value indicating a size or an amount of a load that the load distribution system is receiving and processing with a predetermined threshold, May include determining that the load being received and processed by the load balancing system is excessive if the value indicating the load balancing system exceeds the predetermined threshold.

In an embodiment, the step of determining whether the number of virtual application servers to be operated is appropriate may include comparing a value indicating a size or an amount of a load that the load distribution system is receiving and processing with a predetermined threshold, Determining that the load being received and processed by the load balancing system is underestimated if the value indicating the load balancing system is less than the predetermined threshold.

In an embodiment, the value indicating the size or amount of the load being received and processed by the load balancing system may be a value obtained by dividing the number of service sessions in process by the load balancing system by the number of virtual application servers that are processing the service session.

In an embodiment, the threshold refers to load distribution service status information of the load distribution system, monitoring information of a virtual application server and a physical server included in the load distribution system, and / or system load information collected by the load distribution system Can be determined.

In an embodiment, the one or more virtual application servers or the physical servers may be included in one of an active group, a passive group, and a potential group.

As an embodiment, the step of creating, changing, or deleting one or more virtual application servers may include: creating the one or more virtual application servers in at least one physical server among the physical servers belonging to the potential group; And changing the state of the generated one or more virtual application servers and physical servers into the active group.

In an embodiment, the generated one or more virtual application servers are generated using a virtual machine image generated in advance, and the virtual machine image is a software image file including an operating system for driving the generated one or more virtual application servers, , And may be configured to include the application programs and preferences necessary to process the service request.

As an embodiment, the virtual machine template may be referred to in order to select the pre-created virtual machine image from the virtual machine image pool including a plurality of virtual machine images.

As an embodiment, creating, changing, or deleting the one or more virtual application servers may include changing the state of the one or more virtual application servers from the non-active group to the active group.

As an embodiment, the step of creating, changing, or deleting the one or more virtual application servers may include changing the state of the one or more virtual application servers from the active group to the passive group.

As an embodiment, the step of creating, changing, or deleting one or more virtual application servers may include: checking whether the session being processed is completed for the one or more virtual application servers that have been changed to the passive group; And stopping or deleting the one or more virtual application servers according to the confirmation result.

As an embodiment, the step of creating, changing, or deleting the one or more virtual application servers may include: if the virtual application server running on the physical server including the one or more virtual application servers no longer exists, And changing the state of the server into a potential group.

The load balancer for distributing the service requests received from the client to the virtual application servers included in the physical server in the load balancing system according to the embodiments of the present invention may further include a state information of the load balancing system, A transmission control unit for determining whether the number of virtual application servers operated in the load balancing system is appropriate according to the information or status information of the virtual application servers; And a server control unit for controlling one or more virtual application servers to be created, changed, or deleted according to the determined result, wherein the status information of the load distribution system includes information about a size of a load Or amount.

According to the embodiments of the present invention, the virtual application server can be actively managed according to the status information of the load balancing system, the status information of the virtual application server, or the status information of the physical server without any intervention from the outside such as a service manager or an external application program By adding, changing state or removing, the processing capacity of the service cluster can be appropriately controlled.

Accordingly, it is possible to minimize the resource waste and the cost incurred due to the excessive configuration of the service cluster, and to continuously provide the service without deteriorating the service quality in the service environment where the fluctuation of the load amount with time is large.

1 is a block diagram showing a general load distribution system.
2 is a block diagram illustrating a typical load-balancing system based on a virtual environment.
3 is a block diagram illustrating a load balancing system, in accordance with one embodiment of the present invention.
4 is a state diagram for explaining a state change of a virtual application server or a physical server in detail according to an embodiment of the present invention.
5 is a flow diagram illustrating a load balancing method, in accordance with an embodiment of the present invention.

The following detailed description of the invention refers to the accompanying drawings, which illustrate, by way of illustration, specific embodiments in which the invention may be practiced. It should be understood that the various embodiments of the present invention are different, but need not be mutually exclusive. For example, certain features, structures, and characteristics described herein may be implemented in other embodiments without departing from the spirit and scope of the invention in connection with one embodiment.

It should also be understood that the position or arrangement of the individual components within each disclosed embodiment can be variously modified without departing from the spirit and scope of the present invention. The following detailed description is, therefore, not to be taken in a limiting sense, and the scope of the present invention is principally defined by the appended claims, encompassed by the appended claims and their equivalents. Where similar reference numerals are used in the figures, like reference numerals refer to the same or similar functions for various embodiments.

Hereinafter, the contents and ideas of the present invention will be described with reference to the accompanying drawings.

2 is a block diagram illustrating a typical load-balancing system based on a virtual environment. 2, the load balancing system 20 includes a service cluster (not shown) including one or more clients 21, a network 24, a load balancer 22, and a plurality of physical servers 23_1, 23_2, 23).

The client 21 is connected to the load balancer 22 via the network 24 and transmits a service request directed to a specific Internet protocol (VIP) 25. [ The client 21 includes one or more client devices 21_1, 21_2 and 21_n and each of the client devices 21_1, 21_2 and 21_n may be a homogeneous or heterogeneous computer device, a mobile terminal or other network 24, Lt; RTI ID = 0.0 > data communication < / RTI >

Network 24 is a network of hardware or software that mediates communication connections between client 21 and load balancer 22. The network 24 may be a LAN (Local Area Network), a Metropolitan Area Network (MAN), a Global System for Mobile Network (GSM), an Enhanced Data GSM Environment (EDGE), a High Speed Downlink Packet Access (HSDPA) Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Bluetooth, Zigbee, Wi-Fi, Voice over Internet Protocol (VoIP) Advanced, IEEE 802.16m, WirelessMAN-Advanced, HSPA +, 3GPP Long Term Evolution (LTE), Mobile WiMAX (IEEE 802.16e), UMB (formerly EV- DO Rev. C), Flash-OFDM, iBurst and MBWA systems, HIPERMAN, Beam-Division Multiple Access (BDMA), and ultrasound-enabled communication networks.

The service cluster 23 includes one or more physical servers 23_1, 23_2, and 23_3. Each of the physical servers 23_1, 23_2, and 23_3 includes hypervisors 23_1a, 23_2a, and 23_3a and virtual machine controllers 23_1b, 23_2b, and 23_3b. In addition, each of the physical servers 23_1, 23_2, and 23_3 may optionally include one or more virtual application servers 23_1c, 23_1d, 23_2c, and 23_3c.

The hypervisors 23_1a, 23_2a, and 23_3a provide the virtualization environment to the corresponding physical servers 23_1, 23_2, and 23_3. The hypervisors 23_1a, 23_2a, and 23_3a are thin layer software that controls various operating system access methods for various computing resources included in the physical servers 23_1, 23_2, and 23_3 such as a processor and a memory. The hypervisors 23_1a, , 23_2a, and 23_3a, different operating systems in the physical servers 23_1, 23_2, and 23_3 may be operated simultaneously. The details of the virtualization environment provided by the hypervisors 23_1a, 23_2a, and 23_3a are well known in the art, and a description thereof will be omitted here.

The virtual machine control units 23_1b, 23_2b, and 23_3b are external control interfaces of the hypervisors 23_1a, 23_2a, and 23_3a, or an upper software library linked to the external control interfaces. The virtual machine control units 23_1b, 23_2 and 23_3b provide control functions for the virtual application servers 23_1c, 23_1d, 23_2c and 23_3c in the respective physical servers 23_1, 23_2 and 23_3. For example, each of the physical servers 23_1, 23_2, and 23_3 can create, delete, start, stop, or pause a virtual application server through the virtual machine controllers 23_1b, 23_2b, and 23_3b.

The virtual application servers 23_1c, 23_1d, 23_2c and 23_3c operate in the virtualization environment provided by the hypervisors 23_1a, 23_2a and 23_3a and are included in the corresponding physical servers 23_1, 23_2 and 23_3, Lt; RTI ID = 0.0 > (22). ≪ / RTI > Each of the virtual application servers 23_1c, 23_1d, 23_2c, and 23_3c can operate using a different operating system from the other virtual application servers in one physical server, and the service requests allocated to the load distributing apparatuses 22 . The virtual application servers 23_1c, 23_1d, 23_2c, and 23_3c may be newly created, deleted, suspended, or paused under the control of the load distributing apparatus 22.

3 is a block diagram illustrating a load balancing system 100, in accordance with one embodiment of the present invention. 3, the load balancing system 100 includes a network 140, a load balancer 120, a service cluster 130 including a plurality of physical servers 131, 132, and 133, 170).

3, the load balancer 120 distributes a service request of a client (not shown) received via the network 140 to each physical server 131, 132, 133 or each virtual application Distributed to the servers 131c and 132c. For example, when a service request for a specific Internet protocol (not shown) is received from the client, the load balancer 120 transmits the received service request according to a predetermined load balancing algorithm to the physical servers (not shown) in the service cluster 130 131, 132, and 133 or the virtual application servers 131c and 132c. At this time, if the received service request is excessive as compared with the number of operated virtual application servers, the load distributing apparatus 120 creates a new virtual application server in the physical servers 131, 132, and 133 or changes state (for example, Active group). On the other hand, if the received service request is inferior to the number of operated virtual application servers, the load balancer 120 deletes or interrupts at least some of the existing virtual application servers in the physical servers 131, 132, State change (e.g., in a passive group).

In this way, controlling the number of virtual application servers in the service cluster 130 according to the service request amount can be similar to a general scale-out method. However, Since the load balancing system dynamically controls the number of virtual application servers according to the load of the service based on the state information and the state information of the physical server, the service request from the client is more dynamically flexible than the general scale- Can be processed.

Furthermore, the present invention can be applied to a virtual application server or a physical server as a possible state, in addition to an active state (a state in which a service resource is allocated to service processing) and a latent state (a state in which a service resource is returned and waiting) 3, the creation, modification, deletion, start, pause, stop and restart of the virtual application servers 131c and 132c can be performed more flexibly and promptly.

Here, the status information of the load balancing service 100 may be information indicating the current status of the load balancing system 100, for example, information indicating the amount or size of the service session being processed by the load balancing system 100. [

The status information of the virtual application server is information indicating the current status of the virtual application servers 131c and 132c and is information indicating the degree of load applied to the virtual application servers 131c and 132c What percentage of the available resources are being used for current session processing, etc.).

The status information of the physical server is information indicating the current status of the physical application servers 131, 132, and 133. For example, it is determined how many virtual application servers are operating in the physical server, Whether it is allocated to the virtual application server, what percentage of the system resources of the physical server are additionally usable, or the current temperature of the physical server (i.e., the current temperature), and the like.

The status information of the load balancing service, the status information of the virtual application server, and the status information of the physical server may additionally include other information in addition to or in addition to the information described above.

Hereinafter, the specific configuration and operation method of the load distributing apparatus 120 will be described in more detail below.

The load distributing apparatus 120 includes a transmission processing unit 121, a transmission control unit 122, a session management unit 123, an application server monitoring unit 124 (hereinafter referred to as a monitoring unit), a server control unit 125 and a service management unit 125 do. Each component of the load balancer 120 may be implemented as a hardware component or a software component.

In FIG. 3, the load distributing apparatus 120 enlarges or reduces the sizes of the virtual application servers 131c and 132c, based on the current virtual application server status and service session information.

The transmission processing unit 121 receives a service request having a specific VIP (for example, 15 in FIG. 2) as a destination address through the network 140 and transmits the service request to the virtual application servers 131c and 132c, (A). ≪ / RTI > Specifically, the transmission processing unit 121 generates a flow table (not shown) based on the packet information (for example, the sender IP address, the sender port number, the receiver IP address, the receiver port number, ) To determine the virtual application server to which the received data packet will be forwarded. The flow table is a table containing information associating a specific packet flow (or service request) with a specific application server. For example, the flow table may include information specifying a data packet belonging to the first packet flow to be delivered to the first virtual application server .

At this time, if the received packet belongs to a new service request (or packet flow) newly delivered to the load distributing apparatus 120, packet flow information for the received packet will not exist in the flow table. In this case, the transmission processing unit 121 requests the transmission control unit 122 for the processing policy for the newly received service request. Conversely, if the received packet is a part of a service request received by the load distributing apparatus 120, the flow table of the transmission processing unit 121 will have packet flow information for the received packet. Accordingly, in this case, the transmission processing unit 121 refers to the corresponding packet flow information and immediately performs the network address translation of the received packet.

Here, the network address translation is performed by, for example, changing the IP address and the port number of the client (for example, 11 in FIG. 2) received as the sender IP address and port number to the IP address and port number of the load balancer 120 May include a recipient address translation to change the IP address and port number of the load balancer 120 received as the sender address translation or recipient IP address and port number to the IP address and port number of the virtual application server to process the service request have. At this time, the transmission processing unit 121 may perform both the sender address translation and the receiver address translation as the network address translation, and may perform only the recipient address translation without omitting the sender address translation. Alternatively, the network address translation may further include a method of designating an overlay network such as a tunneling method between the load balancer 120 and a virtual application server to process a service request to specify a transmission path of the received packet . After performing the network address translation, the transmission processing unit 121 transmits the received packet to the corresponding virtual application server (i.e., the received packet or a virtual application server designated to process the service request corresponding to the received packet).

The transmission control unit 122 determines a processing policy for the inquired service request from the transmission processing unit 121 (for example, a policy as to which path or destination the data packet belonging to the specific service request is to be transmitted). For this, the transmission control unit 122 determines which virtual application server among the virtual application servers 131c and 132c on the service cluster 130 existing in relation to the specific VIP is to be allocated the service request.

Specifically, the transmission control unit 122 determines a virtual application server to process the inquired service request according to a predetermined load distribution algorithm (for example, an algorithm for determining a virtual application server or an algorithm for scheduling a service request process) do. Here, the load balancing algorithm used is an individualized algorithm according to the VIP, and the virtual load balancing algorithm used when the VIP of the received service request is different may also be changed.

In the process of determining a virtual application server to process a service request, the transfer control unit 122 receives a service processing state of the virtual application servers 131c and 132c from the session management unit 123 (for example, (E.g., how much of each service is being processed by each of the terminals 131c and 132c).

The session management unit 123 is a module for storing or managing the load processing information of currently available virtual application servers 131c and 132c. The details of the session management unit 123 will be described later.

When the transmission control unit 122 uses a uniform load balancing algorithm as the load distribution algorithm, the transmission control unit 122 notifies the session management unit 123 of the order of the virtual application servers 131c and 132c (That is, if a service request has been recently assigned to a certain virtual application server 131c), the next virtual application server 132c ) Will have the appropriate sequence number. Assign the user request to the virtual application server. On the other hand, when the transmission control unit 122 uses the least connection algorithm as the load distribution algorithm, the transmission control unit 122 notifies the session management unit 123 of the fact that each of the virtual application servers 131c and 132c is in process After receiving the load processing information indicating the number of service sessions, the virtual application server having the smallest number of service sessions currently being processed (or in service) is determined as a virtual application server to process the service request.

Unlike the conventional load balancing systems 10 and 20, in the present invention, an upper threshold value and a lower threshold value are set in order to prevent an overload of a virtual application server and to stably guarantee a service quality thereof, And performs a provisioning step of adjusting the size of the virtual application servers in the service cluster 130 each time a new service request is received based on the threshold of the number of service requests. In accordance with a predetermined load distribution algorithm, a virtual application server to process a new service request among the prepared virtual application servers through the provisioning step is determined.

Specifically, the transfer control unit 122 receives the status information of the virtual application server, the status information of the physical server, or the status information of the load balancing system (e.g., (E.g., the number or the throughput of the service sessions being processed by the load balancing system 100), it is determined whether the number of service sessions currently being processed is appropriate for the capacity of the service cluster 130. That is, the transmission control unit 122 determines whether the number of the virtual application servers operated in the load distribution system 100 is appropriate based on the status information of the load distribution system 100, the status information of the physical server, or the status information of the virtual application servers do.

For example, the transmission control unit 122 determines whether the average number of service sessions currently being processed is greater than a predetermined upper threshold. If the transmission control unit 122 determines that the average number of service sessions is greater than the upper threshold, Out request to add a new virtual application server to the service cluster 130 in order to prevent the virtual application server 130 from performing a scale-out operation.

Conversely, the transmission control unit 122 determines whether the average number of service sessions currently being processed is smaller than a predetermined lower threshold. At this time, if it is determined that the size of the virtual application servers 131c and 132c is smaller than the lower threshold value, the transmission control unit 122 may allocate at least some of the virtual application servers 131c and 132c The server control unit 125 requests the server control unit 125 to perform a scale-in operation to exclude or delete a service request from the service cluster 130. As an example, the upper threshold or the lower threshold may be the status information of the virtual application server, Can be determined by referring to the status information of the physical server. For example, as a result of referring to the status information of the virtual application server, if it is determined that the virtual application server occupies a relatively small amount of system resources (or if the spare amount of available system resources is small) The threshold value can be adjusted downward. As a similar example, if it is determined that each physical server reserves a relatively small amount of system resources (that is, there is no room to create a new virtual application server) as a result of referring to the status information of the physical server, The threshold value can be adjusted upward.

As an example, in order to determine an upper threshold value or a lower threshold value, the monitoring information of the virtual application servers 131c and 132c may be further referred to. More detailed contents of the monitoring information will be described later in the description related to the monitoring unit 124. [

The session management unit 123 may determine the UUID (Universally Unique Identification) of the virtual application servers 131c and 132c that are processing the received service requests with respect to the specific VIP 150, the number of service sessions in process, And state information indicating the activation state of the application server. At this time, the activation status of the virtual application server means, for example, whether the virtual application server belongs to the active group 162, the passive group 161 and the potential group 163.

The session management unit 123 refers to the monitoring information of the virtual application servers 131c and 132c periodically checked by the application server monitoring unit 124 and updates the failure of the virtual application servers 131c and 132c can do.

The monitoring unit 124 collects monitoring information from the virtual application servers 131c and 132c. For example, the monitoring unit 124 can collect monitoring information from the virtual application server 132c by interfacing with any of the plurality of virtual application servers 132c. The collected monitoring information may be provided to the session management unit 123 as described above.

The monitoring information of the virtual application servers 131c and 132c may include not only the information indicating whether the virtual application servers 131c and 132c are faulty, the CPU performance of the virtual application servers 131c and 132c, the CPU utilization rate, And may further include system resource information of virtual application servers 131c and 132c such as utilization rate or network performance.

In response to a request from the transfer control unit 122, the server control unit 125 transmits a new virtual application server to a service cluster (not shown) through a scale-out or a scale-in process considering the availability of system resources of the physical servers 131, 130 or excludes or removes at least some of the existing virtual application servers 131c, 132c from the service cluster 130. [

The server control unit 125 is configured to actively add or remove a virtual application server without the assistance of an external system administrator or other external application programs. The server control unit 125 may refer to the physical server information or the virtual machine template stored in the service management unit 126 to add or remove the virtual application server. The physical server information or the virtual machine template is predefined information or data in the service management unit 126 for each VIP, and the physical server information includes UUID, IP address, manager account information, system specification Etc., and the virtual machine template may include the number of virtual CPUs of the virtual machine, memory capacity, disk capacity, network interface information, virtual machine image information, and the like.

A concrete method of adding or removing the virtual application server by the server control unit 125 is as follows.

When the server control unit 125 adds a new virtual application server, the server control unit 125 first determines a physical server to create a new virtual application server. At this time, the server control unit 125 gives selection priority to each of the available physical servers 131, 132, and 133 according to the status information of the physical server, and selects an optimal physical server accordingly. The status information of the physical server is information indicating the activation status of the physical server, as described above with respect to the virtual application server, and the physical server is any one of the active group 162, the passive group 161 and the potential group 163 Lt; / RTI >

The status information of the physical server may change based on the status information of the virtual application server included in the physical server, or may be replaced or replaced with the status information of the virtual application server.

When a physical server (for example, 132) to create a new virtual application server is selected, the server control unit 125 instructs the virtual machine control unit 132b in the physical server 132 to access the virtual machine And a new virtual application server optimized for the VIP 150 is generated by transmitting the template. At this time, the physical server 132 refers to the virtual machine image information (for example, the virtual machine image UUID) defined in the virtual machine template and stores the virtual machine images 171, 172), and creates a new virtual application server using the selected virtual machine image. The virtual machine image used for creating the virtual application server is an image configured as software in which application program installation and environment setting necessary for processing service requests related to the VIP 150 are completed. Thus, the newly created virtual application server can be used for service request processing as soon as it is created.

Conversely, when the server control unit 125 removes at least a part of the existing virtual application servers 131c and 132c, the server control unit 125 controls the virtual application server or any virtual application server specified in a predetermined manner And 131c, and classifies it into a passive group 161. [ The virtual application server 131c belonging to the passive group 161 is set to process even the service session currently being processed and no further service session is allocated to the virtual application server 131c. If the state of the virtual application server 131c is not changed to the activation group 162 again, the virtual application server 131c belonging to the passive group 161 does not have any service session to be processed at any point in time. In this case, after the virtual application server 131c suspends or shutdown the operation, the system resource used by the virtual application server 131c is returned and deleted.

As an embodiment, a change in state (e.g., change from active group 162 to passive group 161) or deletion of virtual application server 131c may be provided to physical server 131, including virtual application server 131c, And the change in the state of. For example, when all the virtual application servers 131c included in the physical server 131 belong to the passive group 161 in relation to the VIP, the physical server 131 is also classified into the passive group for the VIP, When all the virtual application servers 131c included in the physical server 131 are deleted, the physical server 131 is classified as a potential group 163 for the corresponding VIP to prepare for the subsequent scale-out.

Setting the transient passive group 161 between the activation group 162 and the potential group 163 in this manner can dynamically create a virtual application server without dynamically generating a problem such as interruption of service processing or delay in service processing of the virtual application server. So that the scale of the servers can be expanded or reduced.

The service management unit 126 stores the UUIDs of the usable physical servers 131, 132 and 133, the status information of the physical servers 131, 132 and 133, the configuration information of the virtual application servers 131c and 132c The number of CPUs, the memory capacity, the disk capacity, the network interface information, the virtual machine image UUID, etc.), the default number of virtual application servers to perform initial service request processing, And the upper threshold value and the lower threshold value to be used for expanding or reducing the size of the virtual application server are stored and managed.

Meanwhile, the load balancing system 100 according to the present invention includes a virtual machine image pool 170 including one or more virtual machine images 171 and 172.

The virtual machine image pool 170 stores and manages virtual machine images 171 and 172, which are images used when a new virtual application server is created under the control of the load balancing apparatus 120. The virtual machine images 171 and 172 are stored and managed according to the VIP 150 related to the service request to be processed in a state where installation and environment setting of the application programs necessary for the virtual application server to be created are completed. When a new virtual application server is created, the virtual machine image pool 170 provides the virtual machine images 171 and 172 in a state in which all the installation and environment settings are completed, so that the created virtual application server processes the service request Help. Further, since the virtual application servers use the same virtual machine template and virtual machine images 171 and 172 for each VIP, all virtual application servers that process service requests related to the specific VIP 150 have the same system configuration. Accordingly, the dynamic virtual application server control using the upper threshold value and the lower threshold value proposed in the present invention can be applied more effectively.

Each component of the load distributing apparatus 120 has been described so far. In order to facilitate a thorough understanding of the load distributing apparatus 120 described above, an operation method of the load distributing apparatus 120 will be described in detail with specific examples. Meanwhile, in this example, it is assumed that the load distributing apparatus 120 controls the number of virtual application servers according to the average number of service sessions being processed by the respective virtual application servers 131c and 132c. In other words, the average number of service sessions means the total number of service sessions in process with respect to a particular VIP divided by the number of virtual application servers that are processing the service sessions. However, this is only an exemplary assumption, and the scope of the present invention is not limited thereto. For example, the load balancer 120 may control the number of virtual application servers based on the average traffic volume of data processed by the virtual application servers.

First, when the number of average service sessions exceeds a predetermined upper threshold value, the load distributing apparatus 120 determines that the load (or load size) processed by the load balancing service is excessively larger than the current size of the virtual application server . The load balancer 120 then adds a new virtual application server to distribute service requests related to the VIP. As a result, since the newly added virtual application server shares and processes the service session, the number of average service sessions for the VIP decreases below the upper threshold and the quality of the service provided to the client is also guaranteed.

Meanwhile, as an example, the upper threshold may be a value indicating the type of service request (or service session) being processed (e.g., a web server service request, a database server service request, an application server service request, Service request, etc.), it can be predetermined as a different value for each type of service request.

Conversely, if the number of average service sessions is less than a predetermined lower threshold, the load balancer 120 determines that the service load is less than the current virtual application server size. This means that system resources are overloaded compared to the service request being processed, so that the load balancer 120 excludes at least some of the virtual application servers 131c and 132c that are processing the service request from the service. To this end, the load distributing apparatus 120 first selects a virtual application server (for example, 131c) having the smallest number of service sessions to be processed out of the virtual application servers 131c and 132c that are processing a service request related to the service VIP And classifies the selected virtual application server 131c into a passive group 161. [ A new service session is not additionally allocated to the virtual application server 131c classified into the passive group 161 and the virtual application server 131c processes only the service session currently being processed. If the virtual application server 131c processes all the service sessions allocated to the virtual application server 131c after the lapse of time and no more service sessions are being processed, the virtual application server 131c in the passive group Suspend or shutdown after a period of time elapses or return system resources that have been removed and occupied.

If, by way of example, no virtual application server exists to process a service request associated with the VIP in the physical server (e.g., 133) by aborting or deleting the virtual application server, (163), and then a new virtual application server needs to be expanded.

The virtual machine controllers 131b, 132b, and 133b, the virtual machine controllers 131a, 132b, and 133b, the virtual machine controllers 131b, 132b, and 133b, The contents of the application servers 131c and 132c other than those described herein are the same as those of the client 21, the VIP 25, the network 24, the service cluster 23, the physical servers 23_1, 23_2 and 23_3 23-2b, and 23_3b and the virtual application servers 23_1c, 23_1d, 23_2c, and 23_3c of the hypervisors 23_1a, 23_2a, and 23_3a, respectively.

According to the embodiments of the present invention, the virtual application server can be actively managed according to the status information of the load balancing system, the status information of the virtual application server, or the status information of the physical server without any intervention from the outside such as a service manager or an external application program By adding, changing state or removing, the processing capacity of the service cluster can be appropriately controlled.

Accordingly, it is possible to minimize the resource waste and the cost incurred due to the excessive configuration of the service cluster, and to continuously provide the service without deteriorating the service quality in the service environment where the fluctuation of the load amount with time is large.

4 is a state diagram for explaining a state change of a virtual application server or a physical server in detail according to an embodiment of the present invention. 4, a virtual application server or a physical server exists in one of an active state 210, a passive state 220, and a latent state 230. [

Referring to FIG. 4, the virtual application servers 131c and 132c or the physical servers 131, 132, and 133 of the load distribution system according to the present invention must be in the active state 210, the passive state 220, ). ≪ / RTI > All physical servers 131, 132, and 133 are in potential state 230 prior to providing a service in connection with a VIP and initiate service provision (or service request processing) associated with the VIP, (For example, stored in the service management unit 126) of the virtual application server to be executed, and processes the service request. At this time, the physical server 162 in which the virtual application server 132c and the virtual application server 132c are generated transits to the active state 210 (203). If the service provision related to the VIP is terminated by the external system administrator, the virtual application server 132c in the active state 210 is also terminated and the physical server 132 operating the virtual application server 132c is also in the potential state 230 (205).

On the other hand, if the load of the load balancing service received while the service request is being processed is increased and the average number of service sessions to the VIP exceeds the upper threshold value, the new virtual application server is further added to the service request processing. At this time, the load distributing apparatus 120 firstly checks whether the virtual application server 131c in the passive state 220 is present. If the virtual application server 131c exists in the passive state 220, the load distributing apparatus 120 transitions the virtual application server 131c to the active state 210 and puts the virtual application server 131c into the service request processing (204). Conversely, if there is no virtual application server, which is the passive state 220, the load balancer 120 selects the optimal physical server 133 among the physical servers in the potential state 230, And creates a new virtual application server in the service request process. At this time, the selected physical server 133 and the generated virtual application server transit to the active state 210 (203).

Conversely, if the load on the load balancing service received while processing the service request is reduced and the average number of service sessions for the VIP is less than the lower threshold, at least some of the running virtual application servers are excluded from service request processing. At this time, the load balancer 120 transmits a virtual application server (or a virtual application server selected by another selection method) having the smallest number of service sessions being processed among the virtual application servers in the active state 210 to the passive state 220 (201).

At this time, a new service session is no longer assigned to the virtual application server 131c transited to the passive state 220, and only the existing service session that is being processed by the virtual application server 131c transited to the passive state 220 is processed . If the virtual application server 131c in the passive state 220 processes all the service sessions in process, the virtual application server 131c suspends the operation after waiting for a certain time-out or shutdown) and returns all occupied system resources.

In this way, when none of the virtual application servers for processing the service requests related to the VIPs exist in the physical server 131, the physical server 131 transits to the latent state 230, - Prepare for out.

By dividing the activation state of the virtual application servers 131c and 132c and the physical servers 131, 132, and 133 into one of the active state, the passive state, and the latent state, the virtual application server and the physical server can be dynamically more flexible . Further, even when the size of the virtual application server is dynamically expanded or reduced, the service provided to the client 110 may not be interrupted or delayed, and an unexpected change in service load may be caused by an external system administrator or an external application program Without assistance, the load balancing system 100 can effectively scale the size of the virtual application server.

5 is a flow diagram illustrating a load balancing method, in accordance with an embodiment of the present invention. Referring to FIG. 5, the load balancing method of the load balancer 120 includes steps S111 to S123.

The load balancer 120 receives data packets received over a network 140 addressed to a specific VIP. If the received data packet belongs to an existing service session in process, the load balancer 120 refers to the flow table of the transport processor 121 to forward the packet received to the virtual application server that processes the existing service session do. On the other hand, if the received data packet corresponds to a new service session, the load balancing apparatus 120 analyzes the service processing status of the virtual application servers through the transmission control unit 122, The virtual application server to be processed is determined. Hereinafter, a load balancing method according to the present invention will be described with respect to starting receiving a new service session.

In step S111, the load balancer 120 receives a new service session associated with a particular VIP from the client over the network 140. [

In step S112, the load distributing apparatus 120 analyzes the service processing state of the active virtual application servers in order to allocate a new service session, so that the load distributing apparatus 110 calculates the number (or scale) Is appropriate. At this time, the load distributing apparatus 110 may refer to the status information of the load balancing service 100, the status information of the virtual application server, or the status information of the physical server.

For example, the load balancer 120 determines whether the number of average service sessions being processed by the virtual application servers (i.e., the number of total service sessions in process divided by the number of virtual application servers) exceeds an upper threshold. However, as an exemplary definition, other values (e.g., the total number of service sessions in process) may be used in place of the number of average service sessions.

If the average number of service sessions exceeds the upper threshold value, the load balancing method proceeds to step S113. Otherwise, the load balancing method proceeds directly to step S118.

In step S113, the load distributing apparatus 120 determines whether there is a virtual application server belonging to the passive group in the service cluster 130. [

Since the step S113 corresponds to a case where the virtual application server being operated is insufficient as compared with the load being processed by the load distribution service, the load distributing apparatus 120 may extend the scale of the virtual application server put into the service processing (for example, The number of application servers must be increased) and the average load of virtual application servers should be lowered. To this end, in step S113, the load balancing apparatus 120 detects a virtual application server in a passive state, which is generated in the current physical server but is not allocated a new service session.

If there is a passive group of virtual application servers, the load balancing method proceeds to step S114. Otherwise, the load balancing method proceeds to step S116.

In step S114, the load distributing apparatus 120 determines a virtual application server which is to be changed into an active group among the virtual application servers in the passive group.

As an embodiment, the stateful virtual application server may be determined according to a predetermined criterion. For example, the load balancer 120 may determine a virtual application server that is handling the largest number of service sessions among the passive group of virtual application servers as a virtual application server. In general, a virtual application server having a small number of service sessions is more likely to stop operation and return system resources at an earlier time than otherwise. Therefore, as in the above example, if a virtual application server having fewer service session sessions is left in a passive group, the virtual application server's resource recovery and conversion to a potential group of physical servers become faster, And can be more efficient in system resource management of the system 100.

However, it should be understood that the present invention is not limited thereto. For example, the load balancer 120 may determine any virtual application server in the passive group of virtual application servers as a virtual application server to change status.

If the virtual application server to be changed is determined, the load balancing method proceeds to step S115.

In step S115, the load distributing apparatus 120 controls the load distribution system 100 so that the status of the virtual application server determined in step S114 is changed to the active group.

When step S115 is completed, the load distribution method proceeds to step S121.

On the other hand, if it is determined in step S113 that there is no virtual application server belonging to the passive group in step S113, the load balancing method proceeds to step S116.

In step S116, the load distributing apparatus 120 creates a new virtual application server in at least one of the physical servers of the potential group. In step S116, it is necessary to expand the size of the virtual application server, but there is no virtual application server belonging to the passive group. In this case, it is necessary to create a new virtual application server and share the service load processing.

To this end, the load balancer 120 determines an optimal physical server to create a new virtual application server among the physical servers belonging to the potential group, and creates a new virtual application server in the determined physical server. The optimal physical server for creating a new virtual application server can be determined according to various criteria. For example, the physical server having the largest available system resources among the physical servers belonging to the potential group may be determined as a physical server to be created as a virtual application server, or a physical server having the fastest interface speed with the load balancing apparatus 120 may be virtual The application server may be determined as a physical server to be created. Alternatively, an arbitrary physical server among the physical servers belonging to the potential group may be determined as a physical server to be a virtual application server. Alternatively, a physical server having the largest surplus system resources among the physical servers belonging to the active group or having the least system load may be determined as a physical server to be created as a virtual application server.

Meanwhile, as an embodiment, the load balancer 120 can control the load balancing system 100 to create a new virtual application server using the virtual machine images 171 and 172 stored in the virtual machine image pool 170 have. The details of the virtual machine images 171 and 172 are the same as those described above.

At least one physical server is determined from the potential group in the above manner, and when a virtual application server is created in the determined physical server, the load balancing method proceeds to step S117.

In step S117, the load distributing apparatus 120 changes the state of the virtual application server created in step S116 to the active group. As an embodiment, the load distributing apparatus 120 may change the state of the created virtual application server and the state of the physical server including the generated virtual application server into the active group.

When step S117 is completed, the load distribution method proceeds to step S121.

Returning to step S112, if the average number of sessions being processed by the virtual application server does not exceed the upper threshold value, the load balancing method proceeds to step S118.

In step S118, the load balancer 120 determines whether the average number of service sessions being processed by the virtual application servers is less than a lower threshold. However, as an exemplary definition, other values (e.g., the total number of service sessions in process) may be used instead of the average number of service sessions.

If the average number of service sessions is less than the upper threshold value, the load balancing method proceeds to step S119. Otherwise, the load balancing method proceeds directly to step S121.

On the other hand, in the case of proceeding to the step S119, since the number of the virtual application servers is larger than that of the service session in process, the scale-in process of reducing the size of the virtual application server to increase the resource utilization rate of the load distribution system 100 is performed Needs to be.

In step S119, the load balancer 120 determines a virtual application server to change its status to a passive group among the virtual application servers of the active group.

As an embodiment, the load balancer 120 may determine a virtual application server having the smallest number of service sessions to be processed among virtual application servers in the active group as a virtual application server to change its status. However, this is an exemplary case, and the load balancer 120 may determine a virtual application server that will change state according to various other methods or criteria.

In step S120, the load distributing apparatus 120 changes the status of the virtual application server determined in step S137 to a passive group.

On the other hand, as described above, and as will be reiterated at the end of this discussion, a virtual application server that has changed state to a passive group will no longer be assigned a new service session, It returns the resource and stops or deletes the operation.

When step S120 is completed, the load balancing method proceeds to step S121.

The steps S111 to S120 may be defined as a provisioning step of preparing necessary virtual application servers on an appropriate scale before assigning a new service session. In steps S111 to S120, the load balancing method creates, states, or deletes a virtual application server according to the load balancing service status information, the virtual application server status information, or the physical server status information, Keep it properly.

In step S121, the load distributing apparatus 120 determines, through the transfer control unit 122, a virtual application server to process a new service session among the virtual application servers belonging to the active group according to a predetermined load distribution algorithm. As an example, the predetermined load balancing algorithm may be an equal load balancing algorithm or a least connecting algorithm. At this time, the load distributing apparatus 120 can determine a virtual application server to process a new service session by referring to the load processing information provided by the session managing unit 123. [ For example, when the load distribution algorithm is the minimum connection algorithm, the load distributing apparatus 120 reads information indicating the number of service sessions being processed by each of the virtual application servers among the load processing information stored in the session managing unit 123 , And determines a virtual application server having the smallest number of service sessions currently being processed (or in service) as a virtual application server to process a new service session according to the read information.

In step S122, the load distributing apparatus 120 updates the load processing information of the session managing unit 123 so as to reflect the allocation of a new service session to the previously determined virtual application server.

In step S123, the load distributing apparatus 120 transmits a new service session received via the transmission processing unit 121 to the previously determined virtual application server.

Steps S121 to S123 may be defined as a load balancing application step of actually distributing the load to the virtual application server prepared through the provisioning step.

On the other hand, when the processing method of the virtual application server is changed to the passive group, the load distributing apparatus 120 first determines whether there is a virtual application server belonging to the passive group. If there is a virtual application server belonging to the passive group, the load balancing apparatus 120 checks whether all of the sessions being processed by the virtual application server belonging to the passive group are completed, and suspends or deletes the virtual application server, do. At this time, the virtual application server, which is suspended or deleted, returns the system resources occupied by the virtual application server to the load distribution system 100.

Furthermore, the load balancer 120 determines whether there is another virtual application server in operation (i.e., processing a service session associated with a particular VIP) in the physical server that contained the suspended or deleted virtual application server. If there is no other virtual application server in operation (that is, there is no virtual application server in operation in the physical server), the load balancer 120 changes the state of the physical server to a potential group. The physical server placed in the potential group is reserved or prepared for subsequent scale-out.

The other details of the load distributing apparatus 120 and the load distributing system 100, which are not described in FIG. 5, are the same as those described in FIGS. 1 to 4 above.

According to the embodiments of the present invention, the virtual application server can be actively managed according to the status information of the load balancing system, the status information of the virtual application server, or the status information of the physical server without any intervention from the outside such as a service manager or an external application program It is possible to appropriately control the processing capacity of the service cluster by adding, changing, or removing the service cluster, thereby minimizing resource waste and cost incurred due to over-configuration of the service cluster, The service can be continuously provided in a large service environment without deteriorating the service quality.

In addition, when the size of the virtual application server is enlarged or reduced, the load distribution method of FIG. 5 changes the state of the virtual application server or creates a new virtual application server based on one state information of the passive group. According to such a configuration, the load balancing method of the present invention has an effect of preventing a delay or an interruption of a service provided to the client 110 when a new virtual application server is created or an existing virtual application server is deleted.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments.

Also, although specific terms are used herein, they are used for the purpose of describing the invention only and are not used to limit the scope of the present disclosure as defined in the claims or the claims. Therefore, the scope of the present specification should not be limited to the above-described embodiments, but should be defined by the appended claims and their equivalents.

10, 20, 100: load distribution system 11, 21: client
12, 22, 120: load distributing apparatus 13, 23, 130: cluster
14, 24, 140: network 15, 25: VIP
13_1, 13_2, 13_n: application server
23_1, 23_2, 23_3, 131, 132, 133: physical servers
23_1c, 23_1d, 23_2c, 23_3c, 131c, 131d, 132c, 133c:
23_1a, 23_2a, 23_3a, 131a, 132a, 133a:
23_1b, 23_2b, 23_3b, 131b, 132b, and 133b:
121: Transmission processing unit 122:
123: Session management unit 124: Application server monitoring unit
125: server control unit 126: service management unit
170: Virtual machine image pool 171, 172: Virtual machine image

Claims (14)

A load balancing method of a load balancing apparatus for balancing loads with virtual application servers included in physical servers in a load balancing system,
Determining whether the number of virtual application servers operating in the load balancing system is appropriate based on status information of the load balancing system, status information of the physical server, or status information of the virtual application servers; And
Generating, deleting or deleting one or more virtual application servers according to the determined result,
Wherein the status information of the load distribution system indicates a size or amount of a load that the load distribution system is receiving and processing.
The method according to claim 1,
Wherein the step of determining whether the number of the virtual application servers to be operated is appropriate,
Wherein the load balancing system compares a value indicative of the magnitude or amount of the load being received and processed by the load balancing system with a predetermined threshold and if the value indicative of the magnitude or magnitude of the load exceeds the predetermined threshold, And determining that the load under processing is excessive.
The method according to claim 1,
Wherein the step of determining whether the number of the virtual application servers to be operated is appropriate,
Wherein the load balancing system compares a value indicating a size or amount of a load that is being received and processed by the load balancing system with a predetermined threshold and if the value indicating the size or the size of the load is less than the predetermined threshold, And judging that the input device has been inserted.
The method according to claim 2 or 3,
Wherein a value indicating a size or amount of a load being received and processed by the load distribution system is a value obtained by dividing the number of service sessions being processed by the load distribution system by the number of virtual application servers processing the service session, Way.
The method according to claim 2 or 3,
Wherein the threshold value is determined by referring to monitoring information of a virtual application server included in the load distribution system or system load information collected by the load distribution system.
The method according to claim 1,
Wherein the one or more virtual application servers or the physical servers are included in one of an active group, a passive group, and a potential group.
The method according to claim 6,
The creating, changing, or deleting of the one or more virtual application servers comprises:
Creating at least one virtual application server in at least one physical server among the physical servers belonging to the potential group; And
And changing the state of the generated one or more virtual application servers to the active group.
8. The method of claim 7,
Wherein the generated one or more virtual application servers are created using a virtual machine image generated in advance,
Wherein the virtual machine image is a software image file including an operating system or the like for driving the generated one or more virtual application servers and configured to include an application program and an environment setting necessary for processing the service request, Load balancing method.
9. The method of claim 8,
And references the virtual machine template to select the pre-generated virtual machine image from the virtual machine image pool including a plurality of virtual machine images.
The method according to claim 6,
The creating, changing, or deleting of the one or more virtual application servers comprises:
And changing state of the one or more virtual application servers from the non-active group to the active group.
The method according to claim 6,
The creating, changing, or deleting of the one or more virtual application servers comprises:
And changing state of the one or more virtual application servers from the active group to the passive group.
12. The method of claim 11,
The creating, changing, or deleting of the one or more virtual application servers comprises:
Confirming that the session being processed has been completed for the one or more virtual application servers that have been changed to the passive group; And
And stopping or deleting the one or more virtual application servers according to the result of the checking.
13. The method of claim 12,
The creating, changing, or deleting of the one or more virtual application servers comprises:
Further comprising: changing a physical server including the one or more virtual application servers into a potential group.
A load balancer for distributing a service request received from a client to virtual application servers included in a physical server in a load balancing system,
A transmission control unit for determining whether the number of virtual application servers operated in the load distribution system is appropriate according to state information of the load distribution system, status information of the physical server, or status information of the virtual application servers; And
And a server control unit for controlling one or more virtual application servers to be created, changed, or deleted according to the determination result,
Wherein the status information of the load distribution system indicates a size or amount of a load being received and processed by the load distribution system, and distributes the service request received from the client to the virtual application servers.

Images