KR20230071381A - Scheduling method and apparatus of i/o process - Google Patents

Abstract

Disclosed are a scheduling method and a device of an I/O process. According to one embodiment of the present invention, a method of scheduling an I/O (input/output) process in a computer system includes the steps of: requesting to read a page stored in a storage device through an I/O process; transferring the I/O process to a waiting queue of the page with the lowest index among the pages requested to be read; receiving a page transmitted from the storage device in response to a read request; and determining a status of the I/O process by comparing the received page with the page requested to be read. The present invention provides an I/O process scheduling technique which minimizes a waking-up action of the I/O processes to reduce an overall context switch frequency.

Description

I/O process scheduling method and apparatus

The disclosure below relates to a method and apparatus for scheduling an I/O process.

The size of files processed by computer systems for ultra-high-performance computing, artificial intelligence, and big data is constantly increasing, and it is also common for multiple application processes to access the files simultaneously. For example, scientific experiment data is being shared between various projects and parallelized in ways appropriate to each discipline. In addition, there are cases in which several artificial intelligence models are simultaneously trained on the same files.

Linux is an operating system that allows ordinary users to easily use Unix, which is mainly used in servers and medium-large computers. Linux can change the source according to various peripheral devices or the characteristics of the system in use, so various variants are appearing.

According to the embodiment, it is possible to provide an I/O process scheduling technique that reduces the total number of context switches by minimizing an operation in which an I/O process sleeps and wakes up.

However, the technical challenges are not limited to the above-described technical challenges, and other technical challenges may exist.

A method of scheduling an input/output (I/O) process in a computer system according to an embodiment includes an operation of requesting a read of a page stored in a storage device through the I/O process and a page having the lowest index among the requested pages. The operation of transitioning the I/O process to the waiting queue for the page, the operation of receiving the page transmitted by the storage device in response to the read request, and the status of the I/O process by comparing the received page with the page requested to read. It may include an operation to determine.

The operation of determining the state of the I/O process may include, when the received page is not the same as the read-requested page, the I/O process moves the I/O process to the waiting queue of a page having the lowest index among pages not transmitted by the storage device. It may include an operation of transitioning.

The operation of determining the state of the I/O process may include switching the I/O process to a ready state when the received page is the same as the read-requested page.

The determining of the state of the I/O process may include switching the I/O process to a ready state when the number of received pages is greater than or equal to a threshold value.

The received page may be a page transmitted by the storage device in response to a fragmented read request.

An I/O process scheduling apparatus according to an embodiment includes a memory for storing one or more instructions and a processor for executing the instruction, and when the instruction is executed, the processor performs the storage through an I/O process. A read request for a page stored in a device is requested, the I/O process is transferred to a wait queue for a page having the lowest index among pages requested for reading, and a page transmitted by the storage device is received in response to the read request. The state of the I/O process may be determined by comparing the page with the page requested to be read.

If the received page is not the same as the read-requested page, the processor may transfer the I/O process to a waiting queue for a page having the lowest index among pages not transmitted by the storage device.

The processor may switch the I/O process to a ready state when the received page is the same as the read-requested page.

The processor may switch the I/O process to a ready state when the number of received pages is greater than or equal to a threshold value.

The received page may be a page transmitted by the storage device in response to a fragmented read request.

1 shows a simplified block diagram of an I/O process scheduling device according to one embodiment.
2 is a diagram for explaining an I/O process of Linux.
3 is a diagram for explaining an example of an I/O process scheduling method.
4 is a diagram for explaining a fragmented read request.
5 is a diagram for explaining an I/O process scheduling method according to an exemplary embodiment.
6 is a diagram for explaining an effect of an I/O process scheduling method according to an embodiment.

Specific structural or functional descriptions of the embodiments are disclosed for illustrative purposes only, and may be changed and implemented in various forms. Therefore, the form actually implemented is not limited only to the specific embodiments disclosed, and the scope of the present specification includes changes, equivalents, or substitutes included in the technical idea described in the embodiments.

Although terms such as first or second may be used to describe various components, such terms should only be construed for the purpose of distinguishing one component from another. For example, a first element may be termed a second element, and similarly, a second element may be termed a first element.

It should be understood that when an element is referred to as being “connected” to another element, it may be directly connected or connected to the other element, but other elements may exist in the middle.

Singular expressions include plural expressions unless the context clearly dictates otherwise. In this specification, terms such as "comprise" or "have" are intended to designate that the described feature, number, step, operation, component, part, or combination thereof exists, but one or more other features or numbers, It should be understood that the presence or addition of steps, operations, components, parts, or combinations thereof is not precluded.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art. 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 art, and unless explicitly defined in this specification, it should not be interpreted in an ideal or excessively formal meaning. don't

Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. In the description with reference to the accompanying drawings, the same reference numerals are given to the same components regardless of reference numerals, and overlapping descriptions thereof will be omitted.

1 shows a simplified block diagram of an I/O process scheduling device according to one embodiment.

In the Linux system, when a large number of processes concurrently request large-capacity reads for overlapping file areas, even though the requested data is not ready, the process in the I/O waiting state (Blocked) is switched to the Ready state. contains problems. As a result, unnecessary context switching may occur in Linux systems, which may degrade the performance of CPU-intensive processes.

The I/O process scheduling apparatus 100 may perform scheduling to suppress excessive state transitions of the I/O process. The I/O process scheduling device 100 may restrict transitioning the I/O process to a ready state when all data requested for reading has not arrived.

The I/O process scheduling apparatus 100 can reduce the total number of context switches by minimizing an operation in which an I/O process sleeps and wakes up unnecessarily.

The I/O process scheduling apparatus 100 may determine the status of the I/O process by comparing a page requested by the process to read from the storage device 150 and a page received from the memory 110 . Specifically, if the read-requested page and the received page are not the same, the I/O process scheduling device 100 transfers the I/O process to the waiting queue of the page having the lowest index among the unreceived pages, and the read-requested page By switching the I/O process to the ready state when the received page and the received page are the same, unnecessary context switching can be reduced.

The I/O process scheduling apparatus 100 may prevent data copy from being delayed by switching the I/O process to a ready state when the number of received pages is greater than or equal to a threshold value.

The I/O process scheduling device 100 may allow more CPU resources to be allocated to other processes that are concurrently running.

The I/O process scheduling device 100 may include a memory 110 , a processor 130 , and a storage device 150 .

The memory 110 may be implemented as a volatile memory device or a nonvolatile memory device.

The volatile memory device may be implemented as dynamic random access memory (DRAM), static random access memory (SRAM), thyristor RAM (T-RAM), zero capacitor RAM (Z-RAM), or twin transistor RAM (TTRAM).

Non-volatile memory devices include electrically erasable programmable read-only memory (EEPROM), flash memory, magnetic RAM (MRAM), spin-transfer torque (STT)-MRAM (conductive bridging RAM), and conductive bridging RAM (CBRAM). , FeRAM (Ferroelectric RAM), PRAM (Phase change RAM), Resistive RAM (RRAM), Nanotube RRAM (Polymer RAM (PoRAM)), Nano Floating Gate Memory Memory (NFGM)), holographic memory, molecular electronic memory device (Molecular Electronic Memory Device), or Insulator Resistance Change Memory.

The memory 110 may include a page cache for receiving pages stored in the storage device 150 .

The processor 130 may execute computer readable code (eg, software) stored in the memory 110 and instructions triggered by the processor 130 .

The processor 130 may be a hardware-implemented data processing device having a circuit having a physical structure for executing desired operations. For example, desired operations may include codes or instructions included in a program.

For example, a data processing unit implemented in hardware includes a microprocessor, a central processing unit, a processor core, a multi-core processor, and a multiprocessor. , Application-Specific Integrated Circuit (ASIC), and Field Programmable Gate Array (FPGA).

The processor 130 may request reading of pages stored in the storage device (eg, block device) 150 through an I/O process.

The processor 130 may transfer the I/O process to the wait queue of the page having the lowest index among the read-requested pages.

The processor 130 may receive a page transmitted by the storage device (eg, a block device) 150 in response to a read request (or, the processor 130 may read a page from the storage device 150). .).

The processor 130 may compare the received page with the read-requested page to determine the state of the I/O process.

For example, if the received page is not the same as the read-requested page, the processor 130 has the lowest index among pages not transmitted by the storage device 150 (or not read from the storage device 150). You can transition the I/O process to the page's wait queue. At this time, the received page may be a page transmitted (read from the storage device 150) by the storage device 150 in response to the fragmented read request.

Also, when the received page is the same as the read-requested page, the processor 130 can reduce unnecessary context switching by switching the I/O process to a ready state.

When the number of received pages is greater than or equal to a threshold value, the processor 130 may prevent data copying from being delayed by switching the I/O process to a ready state.

The storage device 150 may be a data processing device implemented in hardware having a circuit having a physical structure for storing information. The storage device 150 may be implemented as a Solid State Drive (SSD), Hard Disk Drive (HDD), NVME, or the like.

The storage device 150 may store pages. The storage device 150 may transmit a page requested through an I/O process to a page cache existing in the memory 110 . The following describes the I/O process of the Linux system.

2 is a diagram for explaining an I/O process of Linux.

When performing an I/O process in a Linux system, data may be loaded into the page cache 200.

The Linux system can read in advance data expected to be requested in the future, in addition to the data requested by the user, to the page cache 200 through readahead.

The Linux system can adjust the amount of data to be read in advance through a read ahead window 210 and an asynchronous read ahead request (Async) 220 . The size of the asynchronous read-ahead request 220 is the size of data to be additionally read in addition to the data requested by the user, and the size of the read-ahead window 210 is the sum of the data requested by the user and the data to be additionally read. It can be any size you request.

When the user reads the first page of pages read ahead by readahead, the Linux system may determine that the user's file access pattern is sequential read and increase the size of the asynchronous readahead request 220 . The Linux system may set a flag on the first page 221 of the asynchronous read-ahead request 220 and increase the size of the asynchronous read-ahead request 220 when the first page 221 with the flag set is read.

In addition, the Linux system may reduce the size of the asynchronous read-ahead request 220 if it determines that the user's file access pattern is not sequential reading.

Hereinafter, a conventional I/O process scheduling method used in a Linux system will be described with reference to FIG. 3 .

3 is a diagram for explaining an example of an I/O process scheduling method.

The Linux system may perform I/O process scheduling based on a page cache 310, a received page 320, a run queue 330, and processes 351 to 355.

Referring to FIG. 3A, the Linux system makes a read request for six pages stored in the storage device 150 through I/O processes 351 to 353, and transfers I to the wait queue for the page having the lowest index among the requested pages. /O Processes 351 to 353 can be transitioned. The I/O processes 351 to 353 entering the wait queue may be in a wait state.

Referring to FIG. 3B, the Linux system may receive pages 321 and 322 in the page caches 311 and 312 to wake up I/O processes 351 to 353 in a standby state and transfer them to the run queue 330. there is. The I/O processes 351 to 353 transitioned to the run queue 330 may be in a ready state.

Referring to FIG. 3C , the Linux system may switch the I/O processes 351 to 353 to the running state by exchanging the order of the process 355 and the I/O processes 351 to 353 . The Linux system may copy the pages 321 and 322 arriving at the page caches 311 and 312 to the buffer based on the I/O processes 351 to 353 in the running state.

Referring to FIG. 3D, when a page that has not arrived among pages requested to be read through the I/O processes 351 to 353 exists, the Linux system transfers the page having the lowest index among the pages that have not arrived to the wait queue for the I/O process. (351 ~ 353) can be transferred. The I/O processes 351 to 353 that have entered the wait queue may be switched to a wait state.

Normally, an operation to wake up the I/O process occurs when all pages requested for reading have arrived. However, if read requests are fragmented, as described above, an operation to wake up the I/O process may occur even if only some pages arrive. Hereinafter, an operation in which a read request is fragmented will be described.

4 is a diagram for explaining a fragmented read request.

Referring to FIG. 4A , when the number of pages of a read request 410 exceeds 1024, the read request 410 includes a first read request 411, a second read request 412, and a third read request 413. ) can be fragmented. The number of pages of the first read request 411 , the second read request 412 , and the third read request 413 may not exceed 1024.

Referring to FIG. 4B , if there are pages 428 and 429 that have already been read among the pages to which the read request 420 has been made, the read request 420 starts with the already read pages 428 and 429 as the first read request. 421 , the second read request 422 , and the third read request 423 .

Referring to FIG. 4C , when a plurality of processes P1 and P2 request to read one page, the read request 440 includes a first read request 441, a second read request 442, and a third read request. (443), and a fourth read request (444).

The Linux system may perform an operation to wake up the I/O process even if only some pages arrive based on fragmented read requests, which may adversely affect CPU resource allocation. Hereinafter, a method of reducing the total number of context switches by minimizing an I/O process sleeping and waking operation will be described.

5 is a diagram for explaining an I/O process scheduling method according to an exemplary embodiment.

Referring to FIG. 5A, the I/O process scheduling device (the I/O process scheduling device 100 of FIG. 1) is configured for eight pages stored in the storage device 150 through the I/O processes 501 and 502. A read request may be made, and the I/O processes 501 and 502 may be transferred to the wait queue of the page having the lowest index among the read requested pages. The I/O processes 501 and 502 entering the wait queue may be in a wait state.

Referring to FIG. 5B , the I/O process scheduling apparatus 100 receives pages 521 and 522 from the page caches 511 and 512 and checks whether the page cache 510 has received all the pages requested to read. .

Referring to FIG. 5C , the I/O process scheduling apparatus 100 may transfer I/O processes 501 and 502 to the waiting queue of the page 513 having the lowest index among pages not received.

In addition, the I/O process scheduling apparatus 100 may prevent data copy from being delayed by making the I/O process transition to a ready state when the number of received pages is greater than or equal to a threshold value.

6 is a diagram for explaining an effect of an I/O process scheduling method according to an embodiment.

Referring to FIG. 6, while eight threads of an existing Linux system and an I/O process scheduling device (the I/O process scheduling device 100 of FIG. 1) complete a task of simultaneously reading a 4 GB file, You can check the number of context switches. The I/O process scheduling device 100 can reduce the number of context switches by up to 25%. Also, as the number of requested pages increases, the width of the number of context exchanges saved by the I/O process scheduling device 100 may increase.

The I/O process scheduling apparatus 100 can reduce the total number of context switches by minimizing an I/O process sleeping and waking operation.

The embodiments described above may be implemented as hardware components, software components, and/or a combination of hardware components and software components. For example, the devices, methods and components described in the embodiments may include, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate (FPGA). array), programmable logic units (PLUs), microprocessors, or any other device capable of executing and responding to instructions. The processing device may execute an operating system (OS) and software applications running on the operating system. A processing device may also access, store, manipulate, process, and generate data in response to execution of software. For convenience of understanding, there are cases in which one processing device is used, but those skilled in the art will understand that the processing device includes a plurality of processing elements and/or a plurality of types of processing elements. It can be seen that it can include. For example, a processing device may include a plurality of processors or a processor and a controller. Other processing configurations are also possible, such as parallel processors.

Software may include a computer program, code, instructions, or a combination of one or more of the foregoing, which configures a processing device to operate as desired or processes independently or collectively. You can command the device. Software and/or data may be any tangible machine, component, physical device, virtual equipment, computer storage medium or device, intended to be interpreted by or provide instructions or data to a processing device. , or may be permanently or temporarily embodied in a transmitted signal wave. Software may be distributed on networked computer systems and stored or executed in a distributed manner. Software and data may be stored on computer readable media.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer readable medium. A computer readable medium may store program instructions, data files, data structures, etc. alone or in combination, and program instructions recorded on the medium may be specially designed and configured for the embodiment or may be known and usable to those skilled in the art of computer software. there is. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks. - includes hardware devices specially configured to store and execute program instructions, such as magneto-optical media, and ROM, RAM, flash memory, and the like. Examples of program instructions include high-level language codes that can be executed by a computer using an interpreter, as well as machine language codes such as those produced by a compiler.

The hardware device described above may be configured to operate as one or a plurality of software modules to perform the operations of the embodiments, and vice versa.

As described above, although the embodiments have been described with limited drawings, those skilled in the art can apply various technical modifications and variations based on this. For example, the described techniques may be performed in an order different from the method described, and/or components of the described system, structure, device, circuit, etc. may be combined or combined in a different form than the method described, or other components may be used. Or even if it is replaced or substituted by equivalents, appropriate results can be achieved.

Therefore, other implementations, other embodiments, and equivalents of the claims are within the scope of the following claims.

Claims (10)

A method for scheduling an input/output (I/O) process in a computer system,
Requesting a read of a page stored in a storage device through an I/O process;
transitioning the I/O process to a wait queue of a page having the lowest index among read-requested pages;
receiving a page transmitted from the storage device in response to the read request; and
An operation of determining the status of the I/O process by comparing the received page with the read-requested page.
Including, I / O process scheduling method.
According to claim 1,
The operation of determining the state of the I / O process,
If the received page is not the same as the read-requested page, transitioning the I/O process to the waiting queue of a page having the lowest index among pages not transmitted by the storage device.
Including, I / O process scheduling method.
According to claim 1,
The operation of determining the state of the I / O process,
Switching the I/O process to a ready state when the received page is the same as the read-requested page
Including, I / O process scheduling method.
According to claim 1,
The operation of determining the state of the I / O process,
Transitioning the I/O process to a ready state when the number of received pages is greater than or equal to a threshold value
Including, I / O process scheduling method.
According to claim 2 or 4,
The received page is
A page transmitted by the storage device in response to a fragmented read request,
How to schedule I/O processes.
a memory that stores one or more instructions; and
A processor to execute the instruction
including,
When the instruction is executed, the processor:
Requests to read the pages stored in the storage device through the I/O process,
Transferring the I/O process to the wait queue of the page having the lowest index among the read-requested pages;
Receiving a page transmitted by the storage device in response to a read request;
determining the state of the I/O process by comparing the received page with the read-requested page;
I/O process scheduling device.
According to claim 6,
the processor,
Transitioning the I/O process to the wait queue of a page having the lowest index among pages not transmitted by the storage device when the received page is not the same as the read-requested page.
I/O process scheduling device.
According to claim 6,
the processor,
switching the I/O process to a ready state when the received page is the same as the read-requested page;
I/O process scheduling device.
According to claim 6,
the processor,
switching the I/O process to a ready state when the number of received pages is greater than or equal to a threshold value;
I/O process scheduling device.
The method of claim 7 or 9,
The received page is
A page transmitted by the storage device in response to a fragmented read request,
I/O process scheduling device.

Images