KR102846276B1 - Methdo and apparatus for operating virtual machine - Google Patents

Abstract

A virtual machine operating method performed by a virtual machine operating device according to one embodiment includes a step of providing a hypervisor with information about a virtual CPU on which a process requiring QoS guarantee related to a service response time is running, by a guest operating system of a plurality of virtual machines sharing the same physical resource, and a step of the hypervisor dynamically adjusting, based on the information about the virtual CPU, the allocation time of a physical CPU for the virtual CPU on which the process requiring QoS guarantee is running while maintaining the total allocation time of the physical CPUs for each of the plurality of virtual machines the same.

Description

Method and Apparatus for Operating Virtual Machines

The present invention relates to a method for operating a virtual machine and a device capable of performing the method.

In virtualized systems, limited physical resources (e.g., CPUs, central processing units) are shared among multiple virtual machines. Typically, a single physical CPU is shared by multiple virtual CPUs for a set period of time. This means that the computing power of each virtual CPU decreases as the number of virtual machines increases.

Typically, non-virtualized systems prioritize response-critical services by increasing the priority of the relevant processes. However, in virtualized systems, the actual physical CPU occupancy time allocated to a virtual CPU is inversely proportional to the number of virtual machines, resulting in frequent service response time delays. This is a fundamental problem that cannot be solved within the guest operating system within the virtual machine.

In particular, simply increasing the priority of response-time-critical services within the guest operating system has limitations. Even if a response-time-critical process is prioritized within the guest operating system, the CPU time it can occupy is determined by the virtual CPU's execution time.

However, because the hypervisor lacks knowledge of the internal operating processes of the virtual machine, it has limitations in prioritizing response-time-critical services. While techniques for detecting response-time-critical services in the hypervisor have been proposed, they only detect specific types of behavior and suffer from low accuracy. In particular, accuracy deteriorates significantly as the number of response-time-critical services increases.

Japanese Patent No. 5284809, registered on June 7, 2013.

According to one embodiment, a method and device for operating a virtual machine are provided, in which a hypervisor dynamically adjusts the allocation time of a physical CPU for a virtual CPU on which a process requiring quality of service (QoS) guarantee related to a service response time is running, while maintaining the total allocation time of the physical CPU for each virtual machine the same, based on information about the virtual CPU on which the process requiring quality of service (QoS) guarantee is running.

The problems to be solved by the present invention are not limited to those mentioned above, and other problems to be solved that are not mentioned will be clearly understood by a person having ordinary skill in the art to which the present invention pertains from the description below.

As a first aspect, a virtual machine operating method performed by a virtual machine operating device includes a step of providing a hypervisor with information about a virtual CPU on which a process requiring QoS guarantee related to a service response time is running, by a guest operating system of a plurality of virtual machines sharing the same physical resource, and a step of the hypervisor dynamically adjusting, based on the information about the virtual CPU, the allocation time of the physical CPU for the virtual CPU on which the process requiring QoS guarantee is running while maintaining the total allocation time of the physical CPUs for each of the plurality of virtual machines to be the same.

As a second aspect, a virtual machine operating device includes a memory unit storing at least one command, and a processor unit, and by executing the at least one command by the processor unit, a plurality of virtual machines and a hypervisor sharing the same physical resource are created, and a guest operating system of the plurality of virtual machines provides the hypervisor with information about a virtual CPU on which a process requiring QoS guarantee related to a service response time is running, and the hypervisor dynamically adjusts the allocation time of the physical CPU for the virtual CPU on which the process requiring QoS guarantee is running, based on the information about the virtual CPU, while maintaining the total allocation time of the physical CPU for each of the plurality of virtual machines to be the same.

As a third aspect, a computer-readable recording medium storing a computer program includes instructions for causing the processor to perform the virtual machine operating method when the computer program is executed by the processor.

As a fourth aspect, a computer program stored in a computer-readable recording medium includes instructions for causing the processor to perform the virtual machine operating method when the computer program is executed by the processor.

In one embodiment, based on information about the virtual CPUs on which processes requiring QoS guarantees related to service response times are running, the hypervisor dynamically adjusts the physical CPU allocation time for virtual CPUs on which processes requiring QoS guarantees are running, while maintaining the same total allocated physical CPU time for each virtual machine. This provides performance benefits in cloud environments where virtualization systems are heavily utilized and microservices with critical response times are frequently used.

FIG. 1 is a configuration diagram of a computer device that can function as a virtual machine operating device according to one embodiment of the present invention.
FIG. 2 is a schematic diagram of an operation when the computer device of FIG. 1 functions as a virtual machine operating device and performs a virtual machine operating method.
Figure 3 is a graph comparing the response times of the conventional technology and the present invention in moses, a service where response time is important.
FIG. 4 is a graph showing the CDF results of response time according to changes in priority weights in a virtual machine operation method according to one embodiment of the present invention.
Figure 5 is a graph comparing the execution time of the freqmine benchmark running on different virtual machines. VM1 is a virtual machine running with moses, and VM2 is a virtual machine running only the ferqmine benchmark.

The advantages and features of the present invention, and the methods for achieving them, will become clearer with reference to the embodiments described in detail below together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below and may be implemented in various different forms. These embodiments are provided only to ensure that the disclosure of the present invention is complete and to fully inform those skilled in the art of the scope of the invention, and the present invention is defined only by the scope of the claims. Like reference numerals designate like elements throughout the specification.

When describing embodiments of the present invention, detailed descriptions of known functions or configurations will be omitted if they are deemed to unnecessarily obscure the gist of the invention. Furthermore, the terms described below are defined in light of their functions in the embodiments of the present invention and may vary depending on the intent or custom of the user or operator. Therefore, their definitions should be based on the overall content of this specification.

FIG. 1 is a configuration diagram of a computer device (100) that can function as a virtual machine operating device according to one embodiment of the present invention, and FIG. 2 is a schematic diagram of an operation when the computer device (100) of FIG. 1 functions as a virtual machine operating device and thus performs a virtual machine operating method.

Referring to FIGS. 1 and 2, a computer device (100) capable of functioning as a virtual machine operating device includes a memory unit (110) storing at least one command and a processor unit (120), and at least one command of the memory unit (110) is executed by the processor unit (120), thereby creating a plurality of virtual machines (210) and hypervisors (220) that share the same physical resources.

The guest operating systems (211) of the plurality of virtual machines (210) can provide the hypervisor (220) with information about the virtual CPUs on which processes requiring QoS guarantees related to service response time are running, and the hypervisor (220) can dynamically adjust the allocation time of the physical CPUs for the virtual CPUs on which processes requiring QoS guarantees are running while maintaining the total allocation time of the physical CPUs for each of the plurality of virtual machines the same based on the information about the virtual CPUs.

An application (212) running on multiple virtual machines (210) can provide a hint about a service response time to a guest operating system (211) through a system call, and a guest process scheduler (213) of the guest operating system (211) can transmit information about a virtual CPU on which a process requiring QoS guarantee runs to a hypervisor (220) through a hyper call based on the system call, and a virtual CPU scheduler (221) of the hypervisor (220) can dynamically adjust the allocation time of a physical CPU to a virtual CPU according to a hyper call handler corresponding to the hyper call.

The guest operating system (211) can set a separate flag for a process requiring QoS guarantee, and the guest process scheduler (213) can transmit a priority increase request and corresponding virtual CPU information through a hyper call when a process with a separate flag set is scheduled.

The guest process scheduler (213) can transmit the information via a hyper call whenever a process requiring QoS guarantee related to service response time is scheduled in and out of a virtual CPU.

FIG. 3 is a graph comparing the response times of the prior art and the present invention in moses, a service where response time is important; FIG. 4 is a graph showing the CDF results of the response time according to changes in priority weights in a virtual machine operating method according to an embodiment of the present invention; and FIG. 5 is a graph comparing the execution times of freqmine benchmarks running on different virtual machines, where VM1 is a virtual machine running together with moses, and VM2 is a virtual machine running only the ferqmine benchmark.

Hereinafter, with reference to FIGS. 1 and 5, a virtual machine operation method performed by a virtual machine operation device according to one embodiment of the present invention will be examined in detail.

At least one command that can be executed by the processor unit (120) is stored in the memory unit (110) of the virtual machine operating device (100), and at least one command of the memory unit (110) is executed by the processor unit (120), thereby creating a plurality of virtual machines (210) and hypervisors (220) that share the same physical resources.

The guest operating system (211) of the multiple virtual machines (210) created in this way can provide the hypervisor (220) with information about the virtual CPU on which the process requiring QoS guarantee related to service response time is running.

For example, an application (212) running on multiple virtual machines (210) can provide a hint about a service response time to a guest operating system (211) through a system call. For example, processor identification information can be provided to the gate operating system (211) through a system call. In addition, a guest process scheduler (213) of the guest operating system (211) can transmit information about a virtual CPU on which a process requiring QoS guarantee is running to a hypervisor (220) through a hyper call based on the system call. For example, the guest process scheduler (213) can periodically call and transmit a virtual CPU number as information about the virtual CPU to the hypervisor (220).

Here, the guest operating system (211) can set a separate flag for a process requiring QoS guarantee, and the guest process scheduler (213) can transmit a priority increase request and corresponding virtual CPU information through a hyper call when a process with a separate flag set is scheduled in. For example, the guest process scheduler (213) can transmit the corresponding information through a hyper call whenever a process requiring QoS guarantee related to service response time is scheduled in and out of a virtual CPU. Here, in the case of being scheduled out, a priority decrease request can be transmitted together with the corresponding CPU information (S310).

Then, based on the information about the virtual CPU, the hypervisor (220) dynamically adjusts the allocation time of the physical CPU for the virtual CPU on which the process requiring QoS guarantee is running, while maintaining the total allocation time of the physical CPU for each of the multiple virtual machines to be the same. For example, the virtual CPU scheduler (221) of the hypervisor (220) can dynamically adjust the allocation time of the physical CPU for the virtual CPU according to the hyper call handler corresponding to the hyper call. For example, the hyper call handler of the hypervisor (220) can adjust the nice value indicating the priority of the virtual CPU according to the virtual CPU number and priority increase/decrease request received through step S310. Here, in order to maintain the total allocation time of the physical CPU for each of the multiple virtual machines to be the same, that is, in consideration of resource fairness between virtual machines, the CPU allocation time for each virtual machine can be guaranteed by utilizing the concept of a control group. That is, when the priority of one of the virtual CPUs of the virtual machine increases, the total CPU usage of the virtual machine can be kept the same by reducing the physical CPU occupancy time of other virtual CPUs (S320).

FIG. 3 is a graph comparing the response times of the prior art and the present invention in moses, a service where response time is important; FIG. 4 is a graph showing the CDF results of the response time according to changes in priority weights in a virtual machine operating method according to an embodiment of the present invention; and FIG. 5 is a graph comparing the execution times of freqmine benchmarks running on different virtual machines, where VM1 is a virtual machine running together with moses, and VM2 is a virtual machine running only the ferqmine benchmark.

A performance evaluation of a virtual machine operation method according to one embodiment of the present invention will be described with reference to FIGS. 3 to 5.

To evaluate performance, experiments were conducted on a two-socket server. Each socket was equipped with a 12-core Intel Xeon Silver 4116 CPU and 64GB of DDR4 DRAM. KVM-based virtual machines were used for the experiments, each allocated 12 virtual CPUs and 10GB of memory.

Two benchmarks were used in the experiment. One is moses, a microservice used in online translation systems and characterized by a high response time. The other is freqmine, a benchmark within the PARSEC benchmark suite, which is characterized by high CPU usage and a low response time priority.

This experiment involved two virtual machines running, configuring them to compete for CPU resources to simulate a real-world cloud environment. Each virtual CPU maps to a physical CPU, and each physical CPU is shared between two virtual CPUs belonging to different virtual machines.

In an actual cloud environment, response time-critical and non-critical services are run simultaneously to maximize resource utilization. To configure this environment, moses and freqmine are set to run simultaneously on one virtual machine, while only freqmine is set to run on another virtual machine.

<Experimental Results>

Figure 3 shows the standardized response time according to the number of moses threads. In general, the virtual machine operation method according to the embodiment of the present invention is more effective when the process of a service where response time is important has a high load. Therefore, the smaller the number of threads, the greater the performance improvement. The average response time was improved by up to 66.6%, and the tail response time, which is important for QoS, was improved by up to 60.7% at 95% and up to 60.6% at 99%.

The virtual machine operation method according to an embodiment of the present invention shows that the degree of actual performance improvement varies depending on how much the priority of the virtual CPU is adjusted. To verify this, a performance evaluation was performed while increasing the priority weight of the virtual CPU by 2x, 5x, and 10x. Figure 4 shows the CDF results for the response time of moses. As expected, as the weight increases, the reduction in response time increases, and in the case of a 10x weight, the 99% tail response time was reduced by 7.3% compared to a 5x weight, by 37.7% compared to a 2x weight, and by 51.3% compared to the existing system.

The virtual machine operation method according to an embodiment of the present invention can be applied to an actual cloud system because it does not cause performance interference between virtual machines. Figure 5 shows the performance of the freqmine benchmark in each virtual machine. VM1 is a virtual machine in which freqmine and moses run simultaneously, and in this case, the performance of freqmine decreases. This is identical to the assumption of a typical cloud system that prioritizes response-time-critical services. VM2 is a virtual machine in which only freqmine runs, and it shows the same performance regardless of the fact that VM1's moses is prioritized in the virtual machine operation method according to an embodiment of the present invention. This means that there is no performance interference between virtual machines.

Meanwhile, a computer program may be implemented to include instructions for causing a processor to perform each step included in the virtual machine operation method performed by the virtual machine operation device (100) according to the above-described embodiment.

In addition, a computer program including commands for causing a processor to perform each step included in a virtual machine operation method performed by a virtual machine operation device (100) according to the above-described embodiment may be recorded on a computer-readable recording medium.

As described above, according to one embodiment of the present invention, based on information about the virtual CPUs on which processes requiring QoS guarantees related to service response times are running, the hypervisor dynamically adjusts the physical CPU allocation time for the virtual CPUs on which processes requiring QoS guarantees are running, while maintaining the same total physical CPU allocation time for each virtual machine. This provides performance advantages in cloud environments where virtualization systems are heavily utilized and microservices with critical response times are frequently used.

The combination of each block of the attached block diagram and each step of the flowchart may be performed by computer program instructions. These computer program instructions may be installed in a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing equipment, so that the instructions, when executed by the processor of the computer or other programmable data processing equipment, create a means for performing the functions described in each block of the block diagram or each step of the flowchart. These computer program instructions may also be stored in a computer-available or computer-readable memory that can direct a computer or other programmable data processing equipment to implement functions in a specific manner, so that the instructions stored in the computer-available or computer-readable memory can produce an article of manufacture that includes an instruction means for performing the functions described in each block of the block diagram or each step of the flowchart. Since the computer program instructions can also be installed on a computer or other programmable data processing device, a series of operational steps are performed on the computer or other programmable data processing device to create a computer-executable process, and the instructions that cause the computer or other programmable data processing device to perform the steps for executing the functions described in each block of the block diagram and each step of the flowchart can also provide steps for executing the functions described in each block of the block diagram and each step of the flowchart.

Additionally, each block or step may represent a module, segment, or portion of code that includes one or more executable instructions for performing a specific logical function(s). It should also be noted that in some alternative embodiments, the functions mentioned in the blocks or steps may occur out of order. For example, two blocks or steps depicted in succession may actually be performed substantially concurrently, or the blocks or steps may sometimes be performed in reverse order, depending on the functionality they perform.

The above description is merely an illustrative illustration of the technical idea of the present invention, and those skilled in the art will appreciate that various modifications and variations can be made without departing from the essential characteristics of the present invention. Therefore, the embodiments disclosed in the present invention are intended to illustrate, rather than limit, the technical idea of the present invention, and the scope of the technical idea of the present invention is not limited by these embodiments. The scope of protection of the present invention should be interpreted by the following claims, and all technical ideas within a scope equivalent thereto should be interpreted as being included in the scope of the rights of the present invention.

100: Virtual machine operating device
110: Memory section
120: Processor section

Claims (10)

A virtual machine operation method performed by a virtual machine operation device,
A step in which the guest operating systems of multiple virtual machines sharing the same physical resources provide the hypervisor with information about the virtual CPU (central processing unit) on which a process requiring quality of service (QoS) guarantee related to service response time is running,
The hypervisor dynamically adjusts the allocation time of the physical CPU for the virtual CPU on which the process requiring QoS guarantee is running while maintaining the total allocation time of the physical CPU for each of the plurality of virtual machines to be the same based on information about the virtual CPU,
In order to keep the total allocation time of the physical CPU for each virtual machine the same, if the allocation time of the physical CPU for the virtual CPU on which the process requiring the QoS guarantee is running increases, the physical CPU occupancy time of other virtual CPUs in the same virtual machine is reduced.
How to operate a virtual machine. In the first paragraph,
An application running on the above multiple virtual machines provides a hint about the service response time to the guest operating system through a system call,
The guest process scheduler of the guest operating system transmits information about the virtual CPU on which the process requiring QoS guarantee is running to the hypervisor through a hyper call based on the system call.
The virtual CPU scheduler of the hypervisor dynamically adjusts the allocation time of the physical CPU to the virtual CPU according to the hyper call handler corresponding to the hyper call.
How to operate a virtual machine. In the second paragraph,
The guest operating system sets a separate flag for a process requiring QoS guarantee, and the guest process scheduler transmits a priority increase request and corresponding virtual CPU information through the hyper call when the process with the separate flag set is scheduled.
How to operate a virtual machine. In the third paragraph,
The guest process scheduler transmits the information through the hyper call whenever a process requiring QoS guarantee related to the service response time is scheduled in and out of the virtual CPU.
How to operate a virtual machine. A memory section that stores at least one instruction, and
Includes a processor unit,
By executing at least one instruction by the processor unit, a plurality of virtual machines and hypervisors sharing the same physical resources are created,
The guest operating systems of the above multiple virtual machines provide the hypervisor with information about the virtual CPUs on which processes requiring QoS guarantees related to service response time are running.
The hypervisor dynamically adjusts the allocation time of the physical CPU for the virtual CPU on which the process requiring QoS guarantee is running, based on information about the virtual CPU, while maintaining the total allocation time of the physical CPU for each of the plurality of virtual machines to be the same.
In order to keep the total allocation time of the physical CPU for each virtual machine the same, if the allocation time of the physical CPU for the virtual CPU on which the process requiring the QoS guarantee is running increases, the physical CPU occupancy time of other virtual CPUs in the same virtual machine is reduced.
Virtual machine operating device. In paragraph 5,
An application running on the plurality of virtual machines provides a hint about the service response time to the guest operating system through a system call, a guest process scheduler of the guest operating system transmits information about a virtual CPU on which a process requiring QoS guarantee runs to the hypervisor through a hyper call based on the system call, and a virtual CPU scheduler of the hypervisor dynamically adjusts the allocation time of a physical CPU to the virtual CPU according to a hyper call handler corresponding to the hyper call.
Virtual machine operating device. In paragraph 6,
The guest operating system sets a separate flag for a process requiring QoS guarantee, and the guest process scheduler transmits a priority increase request and corresponding virtual CPU information through the hyper call when the process with the separate flag set is scheduled.
Virtual machine operating device. In paragraph 7,
The guest process scheduler transmits the information through the hyper call whenever a process requiring QoS guarantee related to the service response time is scheduled in and out of the virtual CPU.
Virtual machine operating device. A computer-readable recording medium storing a computer program,
The above computer program, when executed by a processor,
A step of providing a hypervisor with information about a virtual CPU on which a process requiring QoS guarantee related to a service response time is running by a guest operating system of a plurality of virtual machines sharing the same physical resource, and a step of dynamically adjusting the allocation time of a physical CPU for a virtual CPU on which a process requiring QoS guarantee is running based on the information about the virtual CPU while maintaining the total allocation time of the physical CPU for each of the plurality of virtual machines to be the same,
In order to keep the total allocation time of the physical CPU for each virtual machine the same, if the allocation time of the physical CPU for the virtual CPU on which the process requiring the QoS guarantee is running increases, the physical CPU occupancy time of other virtual CPUs in the same virtual machine is reduced.
Contains instructions for causing the processor to perform a virtual machine operation method.
Computer-readable recording medium. A computer program stored in a computer-readable recording medium,
The above computer program, when executed by a processor,
A step of providing a hypervisor with information about a virtual CPU on which a process requiring QoS guarantee related to a service response time is running by a guest operating system of a plurality of virtual machines sharing the same physical resource, and a step of dynamically adjusting the allocation time of a physical CPU for a virtual CPU on which a process requiring QoS guarantee is running based on the information about the virtual CPU while maintaining the total allocation time of the physical CPU for each of the plurality of virtual machines to be the same,
In order to keep the total allocation time of the physical CPU for each virtual machine the same, if the allocation time of the physical CPU for the virtual CPU on which the process requiring the QoS guarantee is running increases, the physical CPU occupancy time of other virtual CPUs in the same virtual machine is reduced.
Contains instructions for causing the processor to perform a virtual machine operation method.
Computer program.