KR102262645B1 - Monitoring system and method for resource manegement in cloud computing - Google Patents

Abstract

The present invention relates to a monitoring system and method for resource management in a cloud computing environment. The monitoring system for resource management in a cloud computing environment in accordance with the present invention comprises: a plurality of physical machines constituting a cloud environment; a plurality of virtual machines generated in the physical machines to provide a cloud computing service; a first monitoring unit for monitoring in real time overall computing resource information including CPU usage and memory usage for the overall virtual machines; a second monitoring unit for monitoring in real time real computing resource information including real CPU usage and memory usage for each of the virtual machines; and a migration manager unit for classifying the virtual machines into types and migrating one or more of the classified virtual machines to any one of the physical machines, based on a monitoring result obtained by the first monitoring unit and the second monitoring unit.

Description

Monitoring system and method for resource management in cloud computing environment

The present invention relates to a monitoring system and method for resource management in a cloud computing environment.

The cloud computing system may include a plurality of virtual machines in one physical machine to increase resource efficiency. However, the cloud computing system causes problems such as a bottleneck when an excessive load occurs on a virtual machine included in a physical machine, which results in performance and quality of service (QoS) of the entire cloud computing system. this may be lowered. Therefore, it is an important issue in a cloud computing system to efficiently distribute resources in consideration of the utilization rate of resources utilized by various virtual machines.

In particular, in a cloud computing system, since it takes too much cost to provide hardware itself to provide a computing environment to users, virtualization technology is used to reduce these costs. Virtualization refers to a form of installing a virtual machine monitor, which is system software that manages a virtualized operating system, on a single piece of hardware, placing a virtual machine to be provided to the user on it, and installing the operating system there.

However, cloud computing using operating system virtualization causes problems that have not occurred before. The first is that the utilization rate of hardware increases rapidly and thrashing occurs which frequently causes page faults. Second, all services are stopped when the hardware on which the virtual machine is loaded is overloaded and a failure occurs. situation will occur.

Because of these problems, the monitoring problem of resource management has emerged, and studies on a technique for migrating to another hardware resource so that the user does not feel inconvenience without stopping the service even if there is a hardware problem are continuously being conducted.

US 10-1814560 B1

Therefore, the present invention has been devised to solve the above problem, and an object of the present invention is to monitor the state of resources in a cloud computing environment to prevent overload or performance degradation, and to enable dynamic migration without service interruption. It is to provide a monitoring system and method for resource management in a computing environment.

Other objects and advantages of the present invention will be set forth below and will be learned by way of example of the present invention. Furthermore, the objects and advantages of the present invention may be realized by means and combinations indicated in the claims.

In order to achieve the above object, according to one aspect of the present invention, a monitoring system for resource management in a cloud computing environment, a plurality of physical machines constituting the cloud environment; a plurality of virtual machines created in the physical machine and providing cloud computing services; a first monitoring unit for monitoring the entire computing resource information including CPU usage and memory usage for the entire virtual machine in real time; a second monitoring unit for monitoring real computing resource information including actual CPU usage and memory usage for each of the virtual machines in real time; and a migration manager unit that classifies the virtual machines by type based on the monitoring results obtained from the first monitoring unit and the second monitoring unit, and migrates one or more of the classified virtual machines to any one of the plurality of physical machines. characterized by including.

According to another aspect of the present invention, there is provided a monitoring method for resource management in a cloud computing environment including a plurality of physical machines constituting a cloud environment, and a plurality of virtual machines generated in the physical machine and providing cloud computing services. , a step of monitoring, by the first monitoring unit, total computing resource information including CPU usage and memory usage for the entire virtual machine in real time; monitoring, by a second monitoring unit, actual computing resource information including actual CPU usage and memory usage for each of the virtual machines in real time; and the migration manager unit classifies the virtual machines by type based on the monitoring results obtained from the first monitoring unit and the second monitoring unit, and migrates one or more of the classified virtual machines to any one of the plurality of physical machines. It is characterized in that it comprises the step of making.

In addition, the first monitoring unit, a first monitoring manager for monitoring computing resource information on the CPU, memory, storage and network for the entire virtual machine; a second monitoring manager for monitoring status information including at least one of creation, execution, stop, deletion, and movement of the virtual machine; and a user interface unit for transmitting computing resource information and state information of the virtual machine to a service user.

In addition, the first monitoring unit further includes a third monitoring manager for monitoring the availability, throughput, and response time indicators of the virtual machine, based on whether the indicators reach each threshold and stay for how long. It is characterized in that it determines whether to perform migration of the virtual machine.

In addition, the migration manager unit, a type classification unit for classifying the types for each of the virtual machines; a migration target determination unit that determines a target virtual machine to be migrated among a plurality of virtual machines based on the type of the virtual machine classified by the type classification unit; and a migration execution unit for performing migration for the virtual machine whose migration is determined by the migration target determining unit.

In addition, the type classification unit extracts actual CPU usage and memory usage for a unit time for each of the virtual machines, and performs normalization based on the average CPU usage and memory usage for the entire virtual machine, and the virtual machine Classifies the virtual machine into CPU or memory type through the ratio of actual CPU usage and memory usage during each unit time of the virtual machine to the overall average CPU usage and memory usage, and the migration target determining unit is a CPU type virtual machine Alternatively, the memory-type virtual machine is made to exist in the physical machine according to a certain ratio, but it is characterized in that the CPU-type virtual machine or the memory-type virtual machine exceeding a certain ratio is determined as the target virtual machine to be migrated.

In addition, the user interface unit is characterized in that it transmits the computing resource information and state information of the virtual machine to the user in the format of a Really Simple Syndication (RSS) document.

As described above, according to the monitoring system and method for resource management in a cloud computing environment according to the present invention, it is possible to prevent overload or performance degradation by monitoring the status of resources in the cloud computing environment, and to perform dynamic migration without service interruption. It has the effect of making it possible.

1 is a schematic configuration diagram of a monitoring system for resource management in a cloud computing environment according to the present invention;
Figure 2 is a schematic configuration diagram of the first monitoring unit of Figure 1,
3 is a schematic configuration diagram of the migration manager unit of FIG. 1;
4 is a diagram for explaining a migration process of a virtual machine according to the present invention.

Since the present invention can have various changes and can have various forms, specific embodiments are illustrated in the drawings and described in detail in the text.

However, this is not intended to limit the present invention to the specific disclosed form, it should be understood to include all modifications, equivalents and substitutes included in the spirit and scope of the present invention. Terms such as first, second, etc. may be used to describe various elements, but the elements should not be limited by the terms. The above terms are used only for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, a first component may be referred to as a second component, and similarly, the second component may also be referred to as a first component.

The terminology used in the present application is only used to describe specific embodiments, and is not intended to limit the present invention. The singular expression includes the plural expression unless the context clearly dictates otherwise. In the present application, terms such as “comprise” or “have” are intended to designate that a feature, number, step, operation, component, part, or combination thereof described in the specification is present, and includes one or more other features or It should be understood that the existence or addition of numbers, steps, operations, components, parts, or combinations thereof does not preclude the possibility of addition.

Hereinafter, with reference to the accompanying drawings, preferred embodiments of the present invention will be described in more detail.

1 is a schematic configuration diagram of a monitoring system for resource management in a cloud computing environment according to the present invention, FIG. 2 is a schematic configuration diagram of the first monitoring unit of FIG. 1, and FIG. 3 is a migration manager unit of FIG. It is a schematic configuration diagram, and FIG. 4 is a diagram for explaining the migration process of a virtual machine according to the present invention.

First, referring to FIG. 1 , a monitoring system for resource management in a cloud computing environment according to the present invention includes a plurality of physical machines (PM1, PM2) and a plurality of virtual machines (VM1, VM2), the first monitoring unit 200 , the second monitoring unit 300 , and the migration manager unit 400 .

The plurality of physical machines PM1 and PM2 constitute a cloud environment, and the plurality of virtual machines VM1 and VM2 are created in the physical machines PM1 and PM2 and provide cloud computing services to users.

The physical machines PM1 and PM2 are computing devices equipped with real hardware including a CPU (or processor), memory, storage, and network. For example, the physical machines PM1 and PM2 may include general computing devices such as server computers, workstations, desktop computers, and notebook computers.

The virtual machines VM1 and VM2 are virtualized computing devices including a virtualized CPU (or processor), virtualized memory, virtualized storage, and virtualized network. The virtual machines VM1 and VM2 are included in the physical machines PM1 and PM2, and a virtualized computing environment is implemented as software. Therefore, the virtual machines VM1 and VM2 may use the resources of the physical machines PM1 and PM2 to install and execute an operating system or an application program. Virtual machines (VM1, VM2) directly access and execute the resources of the physical machines (PM1, PM2) using the operating system installed on the virtual machines (VM1, VM2), or run the host operating system installed on the physical machines (PM1, PM2). can be used to access resources

When explaining the relationship between the physical machines (PM1, PM2) and the virtual machines (VM1, VM2), the physical machines (PM1, PM2) are included in the physical machines (PM1, PM2) using a hypervisor (not shown). One or more virtual machines (VM1, VM2) can be managed. The hypervisor is a logical platform for simultaneously executing the operating systems of a plurality of virtual machines (VM1, VM2) on the physical machines (PM1, PM2). The virtual machines VM1 and VM2 may use the hypervisor to virtualize and use hardware devices (not shown) of the physical machines PM1 and PM2.

The first monitoring unit 200 performs a function of monitoring the entire computing resource information including CPU usage and memory usage for the entire virtual machines VM1 and VM2 in real time.

The second monitoring unit 300 performs a function of monitoring real computing resource information including actual CPU usage and memory usage for each of the virtual machines VM1 and VM2 in real time.

In this regard, in the present invention, three experiments were performed to confirm the performance degradation due to the resource concentration phenomenon of the virtual machine in the experimental environment shown in [Table 1]. The first measure performance by executing the same CPU-intensive task on six virtual machines, and the second measure performance by executing the same memory-intensive task on six virtual machines. In the third, CPU-intensive tasks were performed on three virtual machines and memory-intensive tasks were performed on the other three virtual machines. This experiment is to measure the performance of virtual machines in an environment where the resources used by virtual machines are concentrated in CPU and memory, and in a state where the concentration is relaxed by mixing the two tasks, and to check the performance degradation of the virtual machine caused by the concentration of resources. purpose Sysbench was used for performance evaluation. For CPU-intensive tasks, the task of detecting prime numbers in natural numbers was used, and for memory-intensive tasks, the task of performing memory I/O tasks was used. The experimental results are shown in [Table 2].

Classification physics machine virtual machine CPU Intel i5 4570 3.4 Ghz 2 VCPUs Memory 16 GB 1 GB Storage (HDD) 1 TB 50 GB

CPU intensive
If you only have a virtual machine If you have only memory-intensive virtual machines Mixed virtual machine types Number
line
city
liver
(second) VM1 324.5031 392.8933 202.7013
(CPU intensive) VM2 325.7753 393.7775 202.6237
(CPU intensive) VM3 325.8942 393.4798 202.7391
(CPU intensive) VM4 325.5934 393.8911 369.5391
(memory intensive) VM5 325.2640 394.3149 368.5391
(memory intensive) VM6 325.1203 339.5632 368.2444
(memory intensive)

As can be seen in [Table 2], when virtual machines perform tasks, if the virtual machine is concentrated on a specific resource such as CPU or memory, competition for resource use intensifies and the performance of virtual machines deteriorates. could check However, when virtual machines performing CPU- and memory-intensive tasks were mixed, resource concentration was alleviated, and it was confirmed that the performance of tasks performed in the virtual machine was improved. As shown in the experimental results, if virtual machines are focused on a specific resource when performing a task, the performance of the virtual machine will deteriorate. In order to alleviate this, virtual machines can be used to minimize resource concentration based on the resource usage of the virtual machine. You have to manage the machine.

For reference, in the prior art, because the type of virtual machine is predefined, input from the user is used, or only a specific type of virtual machine is considered, resource usage, the number of virtual machines, and the type of work performed on the virtual machine are constantly changing. There are many restrictions to apply to the cloud environment.

Therefore, in the present invention, the actual CPU usage and memory for each of the virtual machines (VM1, VM2) by the second monitoring unit 300 to support the minimization of the resource concentration phenomenon when performing virtual machine migration as shown in the experiment. Actual computing resource information including usage is monitored in real time. At this time, the total size of the CPU or memory allocated to all virtual machines (VM1, VM2) can be extracted, but the CPU or memory usage information that is actually being used among the CPUs or memory allocated to the virtual machines (VM1, VM2) can be extracted. it is impossible Therefore, a separate usage monitoring module (not shown) that can monitor the actual usage of CPU or memory for extracting the usage of CPU or memory inside the virtual machines (VM1, VM2) and sending it to the migration manager unit 400 is provided. can be added

The migration manager unit 400 classifies the virtual machines VM1 and VM2 by type based on the monitoring results obtained from the first monitoring unit 200 and the second monitoring unit 300, and the classified virtual machines ( It performs a function of migrating one or more of VM1 and VM2 to any one of the plurality of physical machines PM1 and PM2.

Referring to FIG. 2 , the first monitoring unit 200 includes a first monitoring manager 210 , a second monitoring manager 220 , a third monitoring manager 230 , and a user interface unit 240 . is composed

The first monitoring manager 210 performs a function of monitoring computing resource information on the CPU, memory, storage, and network for the entire virtual machines VM1 and VM2.

The second monitoring manager 220 performs a function of monitoring status information including at least one of creation, execution, stopping, deletion, and movement of the virtual machines VM1 and VM2.

Therefore, through the results monitored by the first monitoring manager 210 and the second monitoring manager 220, the user can grasp both the computing resource information of the virtual machines (VM1, VM2) and the state information including the state change. there is an advantage

The third monitoring manager 230 performs a function of monitoring the availability, throughput, and response time indicators of the virtual machines VM1 and VM2.

Here, the migration manager unit 400 determines whether the availability, throughput, and response time indicators of the virtual machines (VM1, VM2) reach the respective thresholds and stay for how long, based on the virtual machines (VM1, VM2). can decide whether to carry out the migration of

Specifically, the third monitoring manager 230 monitors whether the values of the indicators approach a threshold while monitoring the availability, throughput, and response time of the virtual machines VM1 and VM2. And, when even one of the values of the indicators reaches a threshold, and the indicator stays at the threshold for more than a certain time, it is regarded as a failure situation and the failure information is transmitted to the management server (not shown) to perform recovery.

For example, it is determined whether any one of the index values reaches a threshold, and if all three do not reach the threshold, monitoring may be continued. And, when even one of the index values reaches a threshold, it is determined whether the index stays at the threshold for a predetermined time (threshold time) or longer, and if it does not stay for a predetermined time (threshold time), monitoring can be continued. However, if it stays at the threshold for more than a certain time, a failure information message is delivered to the management server (not shown), and the management server can decide to perform migration of virtual machines (VM1, VM2).

In the above, the threshold means an average value of each of the three indicators, and the threshold time means the maximum time that the threshold can be reached after reaching the threshold. The threshold value and the threshold time may be set as reference values according to a service level agreement with the user.

Before performing the final migration above, compare the LCPU (Local CPU) of the virtual machines (VM1, VM2) to be migrated with the ACPU (Automatic CPU) of the corresponding virtual machines (VM1, VM2) and compare the corresponding virtual machine ( Migration can be performed only when the ACPU of VM1, VM2) is smaller than the LCPU.

Here, ACPU means the remaining CPU resources of the corresponding virtual machines (VM1, VM2), and LCPU means the CPU resources requested by the user.

The user interface unit 240 is characterized in that it includes a user interface unit for transmitting the computing resource information, state information, and indicators of the virtual machines (VM1, VM2) to the service user.

Preferably, the user interface unit 240 may transmit a monitoring result including computing resource information and state information of the virtual machines VM1 and VM2 to the user in the format of a Really Simple Syndication (RSS) document.

In general, there is a risk of personal information being leaked in the process of transmitting the monitoring result to the user. Therefore, in the present invention, monitoring information is transmitted to the user in the form of a security-enhanced RSS document. In addition to the monitoring information collected by the first monitoring unit 200, the RSS document may additionally include data transmission rate, service time, and application information being used. In addition, user information including user ID and history information may be included. It is displayed to the user through an RSS reader.

Referring to FIG. 3 , the migration manager unit 400 includes a type classification unit 410 , a migration target determining unit 420 , and a migration execution unit 430 .

The type classification unit 410 performs a function of classifying the types of each of the virtual machines VM1 and VM2.

The migration target determining unit 420 determines a target virtual machine to be migrated among a plurality of virtual machines (VM1, VM2) based on the types of the virtual machines (VM1, VM2) classified by the type classification unit 410 . perform the function

The migration performing unit 430 performs a function of performing migration for the virtual machine whose migration is determined by the migration target determining unit 420 .

The type classification unit 410 extracts the actual CPU usage and memory usage for a unit time for each of the virtual machines (VM1, VM2), and calculates the average CPU usage and memory usage for the entire virtual machine (VM1, VM2). Based on the normalization, the operation is performed.

In addition, the type classification unit 410 is a ratio of the actual CPU usage and memory usage during each unit time of the virtual machines (VM1, VM2) to the average CPU usage and memory usage of the entire virtual machines (VM1, VM2). classifies the virtual machines (VM1, VM2) by CPU or memory type.

For reference, since the task for classifying the virtual machine type is executed for a very small amount of time in milliseconds, it incurs negligible overhead to the entire system, and as the number of concurrently executed virtual machines increases, the type classification Even if the number of subjects for classification increases, the work time for classification does not occur significantly.

Next, the migration target determining unit 420 causes the virtual machine of the CPU type or the virtual machine of the memory type to exist in the physical machines PM1 and PM2 according to a predetermined ratio, but exceeding a predetermined ratio of the CPU type virtual machine Alternatively, you can determine the virtual machine of type memory as the target virtual machine to execute the migration.

Preferably, the migration target determining unit 420 may allow the CPU type virtual machine or the memory type virtual machine to exist in the physical machines PM1 and PM2 according to a 1:1 ratio.

If CPU-type virtual machines or memory-type virtual machines exist in the physical machines (PM1, PM2) according to a ratio of 1: 2, the virtual machines of memory type exceeding a certain ratio are used as target virtual machines to be migrated. can decide

Meanwhile, the migration process of the virtual machine (the virtual machine is migrated from the currently loaded physical server to another physical server and disposed) according to the present invention will be described with reference to FIG. 4 .

In the present invention, a migration method for transferring a virtual machine page using a physical memory space called the shadow storage 500 is used. Therefore, when migration is to be performed from the virtual machine being serviced, the changed page information is stored in the shadow storage. It is possible to prevent service failure by temporarily saving it to the newly allocated space and migrating it to the newly allocated space.

For example, assuming that the virtual machine (VM1) of the physical machine (PM1) is determined as the target virtual machine of the type of memory to be migrated to the physical machine (PM2), change the memory page for the virtual machine (VM1) Temporarily store in the shadow storage 500 (1), repeat the transfer of the stored contents to the physical machine (PM2) (2), and then stop the virtual machine (VM1) running on the physical machine (PM1) Changes in the page for the virtual machine (VM1) are not stored in the shadow storage 500 and the remaining pages are transmitted (3).

Those of ordinary skill in the art to which the present invention pertains will understand that the present invention may be embodied in other specific forms without changing the technical spirit or essential features thereof. Therefore, it should be understood that the embodiments described above are illustrative in all respects and not restrictive. The scope of the present invention is indicated by the claims described below rather than the above detailed description, and all changes or modifications derived from the meaning and scope of the claims and their equivalent concepts are included in the scope of the present invention. should be interpreted

100: cloud computing system
PM1, PM2 : Physical machine
VM1, VM2: virtual machine
200: first monitoring unit
210: first monitoring manager
220: second monitoring manager
230: third monitoring manager
240: user interface unit
300: second monitoring unit
400: migration manager unit
410: type classification part
420: Relocation target decision unit
430: migration entourage

Claims (12)

a plurality of physical machines constituting a cloud environment;
a plurality of virtual machines created in the physical machine and providing cloud computing services;
A first monitoring unit for monitoring the entire computing resource information including CPU usage and memory usage for the entire virtual machine in real time;
a second monitoring unit for monitoring real computing resource information including actual CPU usage and memory usage for each of the virtual machines in real time; and
A migration manager unit for classifying the virtual machines by type based on the monitoring results obtained from the first monitoring unit and the second monitoring unit and migrating one or more of the classified virtual machines to any one of the plurality of physical machines; and,
The first monitoring unit,
a first monitoring manager for monitoring computing resource information on CPU, memory, storage and network for the entire virtual machine;
a second monitoring manager for monitoring status information including at least one of creation, execution, stop, deletion, and movement of the virtual machine; and
Comprising a user interface unit for transmitting the computing resource information and state information of the virtual machine to a service user,
The first monitoring unit further comprises a third monitoring manager for monitoring the availability, throughput, and response time indicators of the virtual machine,
Determining whether to perform migration of the virtual machine based on whether the indicators reach each threshold and stay for how long,
The migration manager unit,
a type classification unit for classifying a type for each of the virtual machines;
a migration target determination unit that determines a target virtual machine to be migrated among a plurality of virtual machines based on the type of the virtual machine classified by the type classification unit; and
and a migration execution unit that performs migration for the virtual machine whose migration is determined by the migration target determining unit,
The type classification unit extracts actual CPU usage and memory usage for a unit time for each of the virtual machines, and performs normalization based on the average CPU usage and memory usage for the entire virtual machine, Classifying the virtual machine into CPU or memory type through the ratio of actual CPU usage and memory usage during each unit time of the virtual machine to the average CPU usage and memory usage,
The migration target determining unit causes the virtual machine of the CPU type or the virtual machine of the memory type to exist in the physical machine according to a predetermined ratio, but the target virtual machine to which the virtual machine of the CPU type or the virtual machine of the memory type exceeding the predetermined ratio is executed A monitoring system for resource management in a cloud computing environment, characterized in that it is determined by a machine. delete delete delete delete According to claim 1,
The user interface unit is a monitoring system for resource management in a cloud computing environment, characterized in that for transmitting the computing resource information and state information of the virtual machine to the user in the format of a RSS (Really Simple Syndication) document. A monitoring method for resource management in a cloud computing environment comprising a plurality of physical machines constituting a cloud environment, and a plurality of virtual machines generated in the physical machine and providing cloud computing services,
Monitoring, by the first monitoring unit, the entire computing resource information including CPU usage and memory usage for the entire virtual machine in real time;
monitoring, by a second monitoring unit, actual computing resource information including actual CPU usage and memory usage for each of the virtual machines in real time; and
The migration manager unit classifies the virtual machines by type based on the monitoring results obtained from the first monitoring unit and the second monitoring unit, and migrates one or more of the classified virtual machines to any one of the plurality of physical machines. comprising steps,
The first monitoring unit,
a first monitoring manager for monitoring computing resource information on CPU, memory, storage and network for the entire virtual machine;
a second monitoring manager for monitoring status information including at least one of creation, execution, stop, deletion, and movement of the virtual machine; and
Comprising a user interface unit for transmitting the computing resource information and state information of the virtual machine to a service user,
The first monitoring unit further comprises a third monitoring manager for monitoring the availability, throughput, and response time indicators of the virtual machine,
Determining whether to perform migration of the virtual machine based on whether the indicators reach each threshold and stay for how long,
The migration manager unit,
a type classification unit for classifying a type for each of the virtual machines;
a migration target determination unit that determines a target virtual machine to be migrated among a plurality of virtual machines based on the type of the virtual machine classified by the type classification unit; and
and a migration execution unit performing migration for the virtual machine whose migration is determined by the migration target determining unit,
The type classification unit extracts actual CPU usage and memory usage for a unit time for each of the virtual machines, and performs normalization based on the average CPU usage and memory usage for the entire virtual machine, Classifying the virtual machine into CPU or memory type through the ratio of actual CPU usage and memory usage during each unit time of the virtual machine to the average CPU usage and memory usage,
The migration target determining unit causes the virtual machine of the CPU type or the virtual machine of the memory type to exist in the physical machine according to a predetermined ratio, but the target virtual machine to which the virtual machine of the CPU type or the virtual machine of the memory type exceeding the predetermined ratio is executed A monitoring method for resource management in a cloud computing environment, characterized in that it is determined by a machine. delete delete delete delete 8. The method of claim 7,
The user interface unit is a monitoring method for resource management in a cloud computing environment, characterized in that for transmitting the computing resource information and state information of the virtual machine to the user in the format of a RSS (Really Simple Syndication) document.

Images