KR102240996B1 - Methods and apparatuses for processing input/output completion in virtualization system - Google Patents

Abstract

The present invention relates to a method and an apparatus for processing input/output completion in a virtualization system. According to one embodiment of the present invention, the method for processing input/output completion in the virtualization system comprises the steps of: transmitting, to a second thread, size information of an input/output request inputted to a first thread in a virtualization system; transmitting the inputted input/output request to a host kernel; setting the second thread in an idle state in accordance with an idle time preset for each size information of the input/output request; and confirming whether or not the input/output request is completed by operating the second thread in a polling method when the set idle time is expired. Therefore, fast response time of high-performance storage devices is utilized while maintaining low CPU utilization.

Description

I/O completion processing method and device in the virtualization system {METHODS AND APPARATUSES FOR PROCESSING INPUT/OUTPUT COMPLETION IN VIRTUALIZATION SYSTEM}

The present invention relates to a method and apparatus for processing input/output completion in a virtualization system.

The Linux kernel completes and processes I/O (Input/Output) requests in a polling method or an interrupt method. Because the polling method periodically checks the completion of I/O of the storage device, you can quickly check the completion of I/O, but it consumes a lot of CPU. On the other hand, in the interrupt method, the process is converted to the dormant state until the storage device sends an interrupt to the process that sent the I/O request. Therefore, the CPU usage is small, but I/O completion processing is delayed due to additional scheduling overhead. Since version 4.10, the Linux kernel provides a hybrid polling method to take advantage of both the polling method and the interrupt method. In the hybrid polling method, the process that sent the I/O request is put into a dormant state, and after a certain period of time, the process is scheduled to periodically check for I/O completion. Therefore, it uses a smaller amount of CPU than the polling method and shows a similar level of response time as the polling method. However, since the hybrid polling method is operated only when the RWF_HIPRI flag is set among I/O requests performed by the preadv2 and pwritev2 functions, it does not operate for general I/O requests.

Meanwhile, in a virtualized environment, information on I/O requests from a virtual machine is not transmitted to the host due to a semantic gap between the guest and the host. Therefore, since the virtual machine cannot use the I/O completion processing method provided by the host kernel, it uses its own I/O completion processing method to check the I/O completion of the host kernel. Since version 2.9, QEMU-KVM provides an adaptive polling method to take advantage of the fast response time of high-performance storage devices. The adaptive polling method predicts the response time of the next I/O request based on the response time of the current I/O request, and appropriately applies the polling method and the interrupt method according to the expected time. However, since it operates in a polling method for I/O requests of less than 64KB, it uses almost 100% of the CPU, and the CPU consumption is very high. Therefore, there is a need for a new I/O completion processing method with low CPU consumption while maintaining fast response time in a virtualized environment.

Embodiments of the present invention are to provide a method and apparatus for processing input/output completion in a virtualization system capable of maintaining a low CPU usage rate while utilizing a fast response time of a high-performance storage device in a virtualization system based on a high-performance storage device considering CPU usage. .

Embodiments of the present invention introduce a new input/output (I/O) completion processing method to the hypervisor layer in order to utilize the fast response time of a high-performance storage device in a virtualization system to provide a fast response time to users and efficiently save CPU resources. It is intended to provide a method and apparatus for processing input/output completion in a virtualization system that can be utilized.

Embodiments of the present invention may be referred to as HyPIV (Hybrid Polling and Interrupt in Virtualized Systems), and maintain a balance between CPU usage and response time by applying an appropriate I/O completion processing method according to the size of the I/O request. It is intended to provide a method and apparatus for processing input/output completion in a virtualization system.

In addition, unlike the conventional polling method that has a high response time but has a high CPU usage rate, or the interrupt method that has a low CPU use rate but has a slow response time, the embodiments of the present invention have two embodiments. It is intended to provide a method and apparatus for processing input/output completion in a virtualization system that can collect the advantages of the two methods and reduce the overhead by using the method according to the conditions.

According to an embodiment of the present invention, there is provided a method of processing input/output completion in a virtualization system, the method comprising: transmitting, in the virtualization system, size information of an input/output request input to a first thread to a second thread; Transmitting the input input/output request to a host kernel; Setting the second thread to a dormant state according to a preset dormant time for each size information of the input/output request; And checking whether the input/output request is completed by operating the second thread in a polling manner when the set dormancy time expires. A method of processing input/output completion in a virtualization system may be provided.

The method may further include stopping the polling method operation for the second thread when it is confirmed that the input/output request has been completed.

In the setting of the second thread to the dormant state, when the size of the input/output request is less than or equal to a preset request size value, the second thread may be set to the dormant state according to a preset dormancy time.

The method may further include operating the second thread in an interrupt manner when the size of the input/output request exceeds a preset request size value.

Checking whether the input/output request has been completed may include periodically checking an I/O ring buffer in which completion information from the host kernel is updated to check whether the input/output request has been completed.

When processing of the input/output request is completed in the storage device connected to the host kernel, completion information of the input/output request may be updated in the input/output ring buffer through the host kernel.

The input/output request may be a preset general synchronous input/output request or an asynchronous input/output request.

The first thread and the second thread may operate in a hypervisor layer in the virtualization system, and the host kernel may operate in a host layer in the virtualization system.

The first thread and the second thread may be operated by a QEMU-KVM (Quick Emulator Kernel-Based Virtual Machine) hypervisor operating in the virtualization system.

Meanwhile, according to another embodiment of the present invention, there is provided a storage device for processing an input/output request; A memory for storing at least one program; And a processor connected to the storage device and the memory, wherein the processor executes the at least one program to transmit size information of the input/output request input to the first thread in the virtualization system to the second thread, and the host To process the input I/O request in the storage device connected to the kernel, the input I/O request is transmitted to the host kernel, and the second thread is put in a dormant state according to a preset sleep time for each size information of the input/output request. A device for processing input/output completion in a virtualization system may be provided for setting and checking whether the input/output request is completed by operating the second thread in a polling method when the set dormancy time expires.

When it is determined that the input/output request has been completed, the processor may stop the polling method operation for the second thread.

When the size of the input/output request is less than or equal to a preset request size value, the processor may set the second thread to a dormant state according to a preset dormancy time.

When the size of the input/output request exceeds a preset request size value, the processor may operate the second thread in an interrupt manner.

The processor may periodically check an I/O ring buffer in which completion information from the host kernel is updated to determine whether the input/output request has been completed.

When the processing of the input/output request is completed in the storage device connected to the host kernel, the processor may update completion information of the input/output request in the input/output ring buffer through the host kernel.

The input/output request may be a preset general synchronous input/output request or an asynchronous input/output request.

The first thread and the second thread may operate in a hypervisor layer of the virtualization system, and the host kernel may operate in a host layer of the virtualization system.

The first thread and the second thread may be operated by a QEMU-KVM (Quick Emulator Kernel-Based Virtual Machine) hypervisor operating in the virtualization system.

On the other hand, according to another embodiment of the present invention, a host operating by a host operating system and including a host kernel; A storage device that processes an input/output request requested through the host kernel; A kernel-based virtual machine operated by a guest operating system running on the host operating system; And a hypervisor located between the virtual machine and the host and monitoring the kernel-based virtual machine, wherein the hypervisor transmits size information of the input/output request input to the first thread in the virtualization system to the second thread. Transmits, transmits the input input/output request to the host kernel, sets the second thread to a sleep state according to a preset sleep time for each size information of the input/output request, and when the set sleep time expires, the second thread A virtualization system for processing input/output completion may be provided to check whether the input/output request is completed by operating in a polling method.

Meanwhile, according to another embodiment of the present invention, as a non-transitory computer-readable storage medium including at least one program executable by a processor, when the at least one program is executed by the processor, the processor causes: In the virtualization system, the size information of the input/output request input to the first thread is transmitted to the second thread, the input input/output request is transmitted to the host kernel, and the second idle time is set according to the size information of the input/output request. A non-transitory computer-readable storage medium including instructions for setting a thread to a dormant state and checking whether the input/output request is completed by operating the second thread in a polling method when the set dormant time expires is provided. I can.

Embodiments of the present invention can maintain a low CPU usage rate while utilizing a fast response time of a high-performance storage device in a virtualization system based on a high-performance storage device in consideration of CPU usage.

Embodiments of the present invention introduce a new input/output (I/O) completion processing method to the hypervisor layer in order to utilize the fast response time of a high-performance storage device in a virtualization system to provide a fast response time to users and efficiently save CPU resources. Can be utilized.

Embodiments of the present invention may be referred to as HyPIV (Hybrid Polling and Interrupt in Virtualized Systems), and maintain a balance between CPU usage and response time by applying an appropriate I/O completion processing method according to the size of the I/O request. I can.

In addition, unlike the conventional polling method that has a high response time but has a high CPU usage rate, or the interrupt method that has a low CPU use rate but has a slow response time, the embodiments of the present invention have two embodiments. By using the method according to the conditions, the advantages of both methods can be gathered and the overhead can be reduced.

Embodiments of the present invention are delayed by an average of 4.3% compared to the conventional adaptive polling method, but can reduce CPU usage by up to 38.7%. Accordingly, it can be seen that the embodiments of the present invention are more efficient than the conventional adaptive polling method, because the CPU usage is high compared to the reduction of the response time. In addition, in the embodiments of the present invention, the response time was delayed by an average of 5.9% even when the two sizes of I/O requests of 4 KB and 64 KB were tested at different ratios from 99:1 to 1:99, but the CPU usage was reduced by an average of 28.9%. Can be reduced. That is, even though the I/O request of the smallest size and the I/O request of the largest size are mixed and executed in the implementation range, the embodiments of the present invention show higher efficiency compared to the conventional adaptive polling method.

1 is a block diagram showing the configuration of a virtualization system according to an embodiment of the present invention.
2 is a diagram illustrating an operation of processing completion of an input/output request in a virtualization system according to an embodiment of the present invention.
3 is a block diagram illustrating a configuration of an input/output completion processing apparatus in a virtualization system according to an embodiment of the present invention.
4 and 5 are flowcharts illustrating a method of processing input/output completion in a virtualization system according to an embodiment of the present invention.
6 to 9 are diagrams showing measurement results of response time and CPU usage according to an embodiment of the present invention and the prior art.

Since the present invention can make various changes and have various embodiments, specific embodiments will be illustrated in the drawings and described in detail.

However, this is not intended to limit the present invention to a specific embodiment, it should be understood to include all changes, equivalents, or substitutes included in the spirit and scope of the present invention.

Terms such as first and second 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 component. For example, without departing from the scope of the present invention, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element. The term and/or includes a combination of a plurality of related listed items or any of a plurality of related listed items.

When a component is referred to as being "connected" or "connected" to another component, it is understood that it may be directly connected or connected to the other component, but other components may exist in the middle. It should be. On the other hand, when a component is referred to as being "directly connected" or "directly connected" to another component, it should be understood that there is no other component in the middle.

The terms used in the present application are only used to describe specific embodiments, and are not intended to limit the present invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In the present application, terms such as "comprise" or "have" are intended to designate the presence of features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, but one or more other features. It is to be understood that the presence or addition of elements or numbers, steps, actions, components, parts, or combinations thereof does not preclude in advance.

Unless otherwise defined, all terms used herein including technical or scientific terms have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning in the context of the related technology, and should not be interpreted as an ideal or excessively formal meaning unless explicitly defined in the present application. Does not.

Hereinafter, preferred embodiments of the present invention will be described in more detail with reference to the accompanying drawings. In describing the present invention, in order to facilitate an overall understanding, the same reference numerals are used for the same elements in the drawings, and duplicate descriptions for the same elements are omitted.

1 is a block diagram showing the configuration of a virtualization system according to an embodiment of the present invention.

As shown in FIG. 1, a virtualization system 100 according to an embodiment of the present invention includes a plurality of virtual machines 140, a hypervisor 130, a host 120, and a storage device 110. However, not all of the illustrated components are essential components. The virtualization system 100 may be implemented by more components than the illustrated components, and the virtualization system 100 can be implemented by fewer components than the illustrated components.

Hereinafter, a detailed configuration and operation of each component of the virtualization system 100 of FIG. 1 will be described.

The host 120 is operated by a host operating system and includes a host kernel.

The storage device 110 processes I/O requests requested through the host kernel.

The virtual machine 140 operates based on a kernel and is operated by a guest operating system running on a host operating system.

The hypervisor 130 is located between the virtual machine 140 and the host 120 and monitors the kernel-based virtual machine 140. The hypervisor 130 transmits the size information of the input/output request input to the first thread from the virtualization system 100 to the second thread, transmits the input input/output request to the host kernel, and advances each input/output request size information. According to the set sleep time, the second thread is set to sleep, and when the set sleep time expires, the second thread is operated in a polling method to check whether the input/output request is completed.

2 is a diagram illustrating an operation of processing completion of an input/output request in a virtualization system according to an embodiment of the present invention.

In one embodiment of the present invention, a CPU thread and an I/O thread are executed by the hypervisor 130. 2 shows a process of confirming completion of I/O in the QEMU-KVM hypervisor. QEMU-KVM hypervisor is generally composed of CPU threads, I/0 threads, and several threads that play the same role as Virtual Network Computing (VNCC). Among them, the threads that perform I/O tasks of the virtualization system are CPU threads. And I/O threads. The CPU thread sends I/O requests from a virtual CPU-only queue (virt-queue) to the host kernel. When the CPU thread sends an I/O request, the I/O thread periodically checks the aio_ring buffer and starts to check whether the I/O is complete. And when the host kernel renews the aio_ring buffer, the I/O thread checks that the I/O request has been completed, and performs additional work necessary for processing the I/O completion.

Hereinafter, specific operations of a CPU thread and an I/O thread according to an embodiment of the present invention will be described.

First, in step S101, according to an embodiment of the present invention, a CPU thread receives an input/output (I/O) request through a virtual queue (virt_queue).

In step S102, the virtual CPU thread transfers the size information of the corresponding I/O request to the I/O thread before transferring the input/output (I/O) request to the host kernel. Here, in an embodiment of the present invention, the virtual CPU thread may be referred to as a first thread.

In step S103, the virtual CPU thread transmits an input/output (I/O) request to the host kernel.

In step S104, the I/O thread switches the I/O thread to the dormant state according to the dormant time preset according to the size of the received I/O request. Here, in an embodiment of the present invention, the virtual CPU thread may be referred to as a first thread, and the I/O thread may be referred to as a second thread.

In step S105, when the set sleep time expires, the I/O thread periodically checks the asynchronous input/output ring buffer (aio_ring_buffer) while operating in a polling manner.

In step S106, when processing of the I/O request is completed in the storage device 110, the host kernel updates the completion information of the corresponding I/O request in the asynchronous input/output ring buffer (aio_ring_buffer).

In step S107, when the information of the asynchronous input/output ring buffer (aio_ring_buffer) is updated, the I/O thread checks the information and stops the polling method.

In step S108, the I/O thread checks the completion of the I/O request, and then performs additional I/O completion processing required to complete the processing of the I/O request.

As such, FIG. 2 shows a complete processing operation for an input/output request according to an embodiment of the present invention. In an embodiment of the present invention, unlike the adaptive polling method, an I/O request is first sent to an I/O thread and then the I/O thread is switched to a dormant state. And I/O threads are scheduled after a certain period of time to periodically check for I/O completion. Similar to the hybrid polling method provided by the Linox kernel, it is converted into a dormant state for a certain period of time, so CPU usage is lower than that of the existing adaptive polling method. In addition, in an embodiment of the present invention, unlike maintaining the dormant state for 50% of the average response time in the hybrid pollung method, the optimal dormancy time may be experimentally found and the time may be designated for each I/0 size. Therefore, unlike the hybrid polling method, it is possible to provide the lowest CPU usage while keeping the response time to a minimum. The sleep time is set so that the response delay time of an embodiment of the present invention is 5% or less compared to the conventional polling method, of which the sleep time showing the lowest CPU usage may be used. In addition, the conventional hybrid polling method operates only for synchronous I/O requests in the host operating system, but since an embodiment of the present invention is implemented in the hypervisor 130 like the conventional adaptive polling method, synchronous I/0 It can operate on asynchronous I/0 requests as well as requests.

3 is a block diagram illustrating a configuration of an input/output completion processing apparatus in a virtualization system according to an embodiment of the present invention.

As shown in FIG. 3, the input/output completion processing apparatus 200 in the virtualization system 100 according to an embodiment of the present invention includes a storage device 110, a memory 220, and a processor 230. However, not all of the illustrated components are essential components. The input/output completion processing apparatus 200 may be implemented by more components than the illustrated components, and the input/output completion processing apparatus 200 may be implemented by fewer components. The input/output completion processing apparatus 200 according to an embodiment of the present invention may be included in the hypervisor 130 connected to the host 120 to operate, or may operate as a hypervisor.

Hereinafter, a detailed configuration and operation of each component of the input/output completion processing apparatus 200 in the virtualization system 100 of FIG. 3 will be described.

The storage device 110 processes the input/output request requested through the hypervisor 130.

The memory 220 stores at least one program.

The processor 230 is connected to the storage device 110 and the memory 220.

The processor 230 transmits the size information of the input/output request input to the first thread in the virtualization system 100 to the second thread by executing the at least one program, and transmits the size information of the input/output request input to the first thread to the second thread, and in the storage device 110 connected to the host kernel. In order to process the input I/O request, the input I/O request is transmitted to the host kernel, the second thread is set to sleep according to the preset sleep time according to the size information of the I/O request, and the set sleep time expires. If so, it checks whether the I/O request has been completed by operating the second thread in a polling method.

According to various embodiments, when it is determined that the input/output request has been completed, the processor 230 may stop the polling method operation for the second thread.

According to various embodiments, when the size of the input/output request is less than or equal to a preset request size value (eg, 64 KB), the processor 230 may set the second thread to a sleep state according to a preset sleep time.

According to various embodiments, the processor 230 may operate the second thread in an interrupt method when the size of the input/output request exceeds a preset request size value.

According to various embodiments, the processor 230 may check whether the input/output request is completed by periodically checking an I/O ring buffer in which completion information from the host kernel is updated.

According to various embodiments, when processing of the input/output request is completed in the storage device 110 connected to the host kernel, the processor 230 may update completion information of the input/output request to the ring buffer through the host kernel.

According to various embodiments, the input/output request may be a preset general synchronous input/output request or an asynchronous input/output request.

According to various embodiments, the first thread and the second thread may operate in the hypervisor layer of the virtualization system 100, and the host kernel may operate in the host layer of the virtualization system 100.

According to various embodiments, the first thread and the second thread may operate by a Quick Emulator Kernel-Based Virtual Machine (QEMU-KVM) hypervisor operating in the virtualization system 100.

4 and 5 are flowcharts illustrating a method of processing input/output completion in a virtualization system according to an embodiment of the present invention.

First, the overall flow in the input/output completion processing method according to an embodiment of the present invention will be described with reference to FIG. 4.

In step S201, the input/output completion processing apparatus 200 checks the size information of the input/output request input to the first thread.

In step S202, the input/output completion processing apparatus 200 checks whether the input/output request is equal to or less than a preset request value.

In step S203, if the input/output request is less than or equal to the preset request value, the input/output completion processing apparatus 200 checks whether the input/output request is completed by operating the second thread in a polling manner.

On the other hand, in step S204, when the input/output request exceeds a preset request value, the input/output completion processing apparatus 200 checks whether the input/output request is completed by operating the second thread in an interrupt manner.

Thereafter, in step S205, the input/output completion processing apparatus 200 checks whether the input/output request is completed.

In step S206, when the input/output request is completed, the input/output completion processing apparatus 200 may perform an additional operation necessary for processing the input/output completion through the second thread.

Meanwhile, with reference to FIG. 5, a detailed flow of step S203 of confirming whether the input/output request is completed by operating the second thread in a polling manner will be described.

In step S301, the input/output completion processing apparatus 200 transmits the size information of the input/output request input to the first thread to the second thread.

In step S302, the input/output completion processing device 200 transmits the input input/output request to the host kernel.

In step S303, the input/output completion processing apparatus 200 sets the second thread to the dormant state according to a preset dormancy time for each size information of the input/output request.

In step S304, the input/output completion processing apparatus 200 checks whether the dormant time expires.

In step S305, when the dormant time expires, the input/output completion processing apparatus 200 checks whether the input/output request is completed by operating the second thread in a polling manner.

Thereafter, when it is determined that the input/output request has been completed, the input/output completion processing apparatus 200 may stop the polling method operation for the second thread.

6 to 9 are diagrams showing measurement results of response time and CPU usage according to an embodiment of the present invention and the prior art.

An experimental method of measuring response time and CPU usage according to an embodiment of the present invention and the prior art will be described. 6 to 9 are diagrams comparing a conventional I/O completion processing method according to the size of an I/O request through this experimental method, and a response time and CPU usage according to an embodiment of the present invention.

This experiment was performed on a system equipped with hardware equipment of Intel Core i7-8700, DRAM 16GB, and Intel Optane SSD 480GB. The host kernel version is 4.16.2 and the guest kernel version is 4.20.2. The operating system and file system used for the host and guest are Ubuntu 14.04LTS and Ext4 Filesystem. In addition, the virtual machine was created using QEMU-KVM (ver. 2.10.2), and the virtual machine has a 32GB virtual disk, 1 vCPU, and 1024MB of memory. In addition, the virtio-blk-pci device driver was used to provide an adaptive polling method. In order to measure I/O performance, the workload was performed using the FIO benchmark, which is typically used to measure storage performance. The workload used in the experiment uses the pvsync2 engine to perform random reads on the virtual disk. Since the pvsync2 engine performs synchronous I/O with the RWF_HIPRI option, hybrid polling works when the host executes the workload, but hybrid polling of the host does not work because this experiment is performed in a virtual machine.

The experiment was conducted in two ways. The first was to increase the size of the I/O request from 4KB to 64KB in a 4KB basis, and the adaptive polling method and interrupt method according to the size of each I/O request, and the response time and CPU usage of the proposed method. Was measured. The second is to change the ratio of 4KB and 64KB I/O requests from 99:1 to 1:99, and adjust the response time and CPU usage of the proposed method as well as the adaptive polling method and interrupt method according to the ratio of the I/O request size. It was measured.

Looking at the experimental results, as shown in FIGS. 6 and 7, in the case of the first experiment, the input/output completion processing method according to an embodiment of the present invention, that is, a Hypervisor Polling and Interrupt in Virtualized system (HPIV) is a conventional adaptation. Compared to the adaptive polling method, response time increased by up to 5.3%, but CPU usage decreased by up to 38.7%. That is, it can be seen that compared to the performance degradation due to the delay in response time, the CPU usage has been significantly reduced compared to the conventional method. The response time and CPU usage of the conventional I/O completion processing method according to the I/O size ratio and the input/output completion processing method according to an embodiment of the present invention are compared in FIGS. 6 and 7.

8 and 9, in the case of the second experiment, the HyPIV according to an embodiment of the present invention delayed the response time by an average of 5.9% compared to the conventional adaptive polling method, but the CPU usage decreased by an average of 28.9%. I did. In addition, even though the I/O request of the smallest size of 4KB and the I/O request of 64KB, which is the largest I/O size in the implementation range, were mixed and executed, it was confirmed that it showed higher efficiency compared to the adaptive polling method.

The input/output completion processing method in the virtualization system according to the embodiments of the present invention described above may be implemented as a computer-readable code on a computer-readable recording medium. The method of processing input/output completion in a virtualization system according to embodiments of the present invention may be implemented in the form of program commands that can be executed through various computer means and recorded in a computer-readable recording medium.

A non-transitory computer-readable storage medium including at least one program executable by a processor, wherein when the at least one program is executed by the processor, the processor causes: an input/output request input to a first thread in a virtualization system The size information of the input/output is transmitted to the second thread, and the input/output request is transmitted to the host kernel so that the input/output request is processed by the storage device connected to the host kernel, and preset for each size information of the input/output request. Non-transitory computer comprising instructions for setting the second thread to a dormant state according to a dormant time, and for checking whether the input/output request is completed by operating the second thread in a polling method when the set dormant time expires A readable storage medium may be provided.

The method according to the present invention described above can be implemented as a computer-readable code on a computer-readable recording medium. The computer-readable recording medium includes all types of recording media in which data that can be decoded by a computer system is stored. For example, there may be read only memory (ROM), random access memory (RAM), magnetic tape, magnetic disk, flash memory, optical data storage device, and the like. In addition, the computer-readable recording medium can be distributed in a computer system connected through a computer communication network, and stored and executed as codes that can be read in a distributed manner.

Although the above has been described with reference to the drawings and examples, it does not mean that the scope of protection of the present invention is limited by the drawings or examples, and those skilled in the art It will be appreciated that various modifications and changes can be made to the present invention without departing from the spirit and scope.

Specifically, the described features may be implemented in digital electronic circuitry, or computer hardware, firmware, or combinations thereof. Features may be executed in a computer program product implemented in storage in a machine-readable storage device, for example, for execution by a programmable processor. And the features can be performed by a programmable processor executing a program of directives to perform the functions of the described embodiments by operating on input data and generating output. The described features include at least one programmable processor, at least one input device, and at least one output coupled to receive data and directives from the data storage system and to transmit data and directives to the data storage system. It can be executed within one or more computer programs that can be executed on a programmable system including the device. A computer program includes a set of directives that can be used directly or indirectly within a computer to perform a specific action on a given result. Computer programs are written in any form of programming language, including compiled or interpreted languages, and included as modules, elements, subroutines, or other units suitable for use in other computer environments, or as independently operable programs. It can be used in any form.

Suitable processors for execution of a program of directives include, for example, both general and special purpose microprocessors, and either a single processor or multiple processors of a different kind of computer. Storage devices suitable for implementing computer program directives and data that also implement the described features are, for example, semiconductor memory devices such as EPROM, EEPROM, and flash memory devices, magnetic devices such as internal hard disks and removable disks. Devices, magneto-optical disks and all types of non-volatile memory including CD-ROM and DVD-ROM disks. The processor and memory may be integrated within application-specific integrated circuits (ASICs) or added by ASICs.

The present invention described above is described on the basis of a series of functional blocks, but is not limited by the above-described embodiments and the accompanying drawings, and various substitutions, modifications and changes within the scope not departing from the technical spirit of the present invention It will be apparent to those of ordinary skill in the art to which this invention pertains.

Combinations of the above-described embodiments are not limited to the above-described embodiments, and combinations of various types as well as the above-described embodiments may be provided according to implementation and/or need.

In the above-described embodiments, the methods are described on the basis of a flow chart as a series of steps or blocks, but the present invention is not limited to the order of steps, and certain steps may occur in a different order or concurrently with other steps as described above. I can. In addition, those of ordinary skill in the art understand that the steps shown in the flowchart are not exclusive, other steps are included, or one or more steps in the flowchart may be deleted without affecting the scope of the present invention. You can understand.

The above-described embodiments include examples of various aspects. It is not possible to describe all possible combinations for representing the various aspects, but one of ordinary skill in the art will recognize that other combinations are possible. Accordingly, the present invention will be said to include all other replacements, modifications and changes falling within the scope of the following claims.

100: virtualization system
110; Storage device
120: host
130: hypervisor
140: virtual machine
200: I/O completion processing unit
210: storage device
220: memory
230: processor

Claims (20)

In the method of processing input/output completion in a virtualization system,
Transmitting size information of the input/output request input to the first thread in the virtualization system to the second thread;
Transmitting the input input/output request to a host kernel;
Setting the second thread to a dormant state according to a preset dormant time for each size information of the input/output request; And
When the set dormant time expires, operating the second thread in a polling method to check whether the input/output request has been completed, and
The setting of the second thread to the dormant state includes setting the second thread to the dormant state according to a preset dormancy time when the size of the input/output request is less than or equal to a preset request size value. Processing method. The method of claim 1,
When it is confirmed that the input/output request has been completed, the method of processing input/output completion in a virtualization system further comprising: stopping a polling method operation for the second thread. delete The method of claim 1,
The method of processing input/output completion in a virtualization system, further comprising operating the second thread in an interrupt mode when the size of the input/output request exceeds a preset request size value. The method of claim 1,
Checking whether the input/output request is completed,
A method for processing input/output completion in a virtualization system, wherein the I/O ring buffer for which completion information from the host kernel is updated is periodically checked to check whether the input/output request is completed. The method of claim 5,
Checking whether the input/output request is completed,
When the processing of the input/output request is completed in the storage device connected to the host kernel, the input/output completion information of the input/output request is updated in the input/output ring buffer through the host kernel. The method of claim 1,
The input/output request,
A method of processing I/O completion in a virtualization system, which is a preset general synchronous I/O request or asynchronous I/O request. The method of claim 1,
The first thread and the second thread operate in a hypervisor layer in the virtualization system, and the host kernel operates in a host layer in the virtualization system. The method of claim 1,
The first thread and the second thread are operated by a QEMU-KVM (Quick Emulator Kernel-Based Virtual Machine) hypervisor operating in the virtualization system. A storage device that processes input/output requests;
A memory for storing at least one program; And
A processor connected to the storage device and the memory,
The processor, by executing the at least one program,
In the virtualization system, the size information of the input/output request input to the first thread is transmitted to the second thread,
Transmitting the input I/O request to the host kernel so that the input I/O request is processed by the storage device connected to the host kernel,
Set the second thread to sleep according to a preset sleep time according to the size information of the input/output request, but sleep the second thread according to a preset sleep time when the size of the input/output request is less than or equal to a preset request size value. Set to the state,
When the set dormant time expires, the second thread is operated in a polling method to check whether the input/output request is completed. The method of claim 10,
The processor,
When it is confirmed that the input/output request has been completed, the polling method operation for the second thread is stopped. delete The method of claim 10,
The processor,
When the size of the input/output request exceeds a preset request size value, the second thread is operated in an interrupt manner. The method of claim 10,
The processor,
An I/O completion processing device in a virtualization system that periodically checks an I/O Ring Buffer in which completion information from the host kernel is updated to check whether the I/O request has been completed. The method of claim 14,
The processor,
When processing of the input/output request is completed in the storage device connected to the host kernel, the input/output completion processing device in a virtualization system updates completion information of the input/output request to the input/output ring buffer through the host kernel. The method of claim 10,
The input/output request,
I/O completion processing device in a virtualization system, which is a preset general synchronous I/O request or asynchronous I/O request. The method of claim 10,
The first thread and the second thread operate in a hypervisor layer of the virtualization system, and the host kernel operates in a host layer of the virtualization system. The method of claim 10,
The first thread and the second thread are operated by a QEMU-KVM (Quick Emulator Kernel-Based Virtual Machine) hypervisor operating in the virtualization system. A host operated by a host operating system and including a host kernel;
A storage device that processes an input/output request requested through the host kernel;
A kernel-based virtual machine operated by a guest operating system running on the host operating system; And
And a hypervisor located between the virtual machine and the host and monitoring the kernel-based virtual machine,
The hypervisor transmits the size information of the input/output request input to the first thread in the virtualization system to the second thread,
Transmit the input I/O request to the host kernel,
Set the second thread to sleep according to a preset sleep time according to the size information of the input/output request, but sleep the second thread according to a preset sleep time when the size of the input/output request is less than or equal to a preset request size value. Set to the state,
When the set dormant time expires, the second thread is operated in a polling manner to check whether the input/output request is completed or not. A non-transitory computer-readable storage medium comprising at least one program executable by a processor, wherein the at least one program, when executed by the processor, causes the processor to:
In the virtualization system, the size information of the input/output request input to the first thread is transmitted to the second thread
Transmit the input I/O request to the host kernel,
Set the second thread to sleep according to a preset sleep time according to the size information of the input/output request, but sleep the second thread according to a preset sleep time when the size of the input/output request is less than or equal to a preset request size value. Set to the state,
And instructions for checking whether the input/output request is completed by operating the second thread in a polling manner when the set dormancy time expires.

Images