WO2019136967A1

WO2019136967A1 - Task scheduling optimization method applied to storage system

Info

Publication number: WO2019136967A1
Application number: PCT/CN2018/099743
Authority: WO
Inventors: 黄中柱; 李庭育; 魏智汎; 齐元辅
Original assignee: 江苏华存电子科技有限公司
Priority date: 2018-01-12
Filing date: 2018-08-09
Publication date: 2019-07-18
Also published as: CN108182158A

Abstract

A task scheduling optimization method applied to a storage system, comprising a host (1), a flash translation layer (2), a virtual drive (3) and a flash array (4), wherein the host (1) is connected to the flash translation layer (2) by means of a bus, the virtual drive (3) is connected to the flash array (4) by means of a bus, and a plurality of task queues (8, 9, 10, 11), a task scheduling unit (5), a physical layer driver (6), an execution scheduling unit (7) and a plurality of execution queues (12, 13, 14, 15) are provided in the virtual drive (3). By means of a method for assigning task priorities on a storage system, the efficiency of the storage system is improved.

Description

Task scheduling optimization method applied in storage system

Technical field

The invention relates to the technical field of scheduling optimization, in particular to a task scheduling optimization method applied in a storage system.

Background technique

The conventional storage system includes a flash controller and a flash array; the flash controller communicates with the flash through the flash interface and communicates with the host through the host interface; the flash controller includes a flash translation layer (FTL) and a physical layer driver; Architecture, when the host device generates instructions, the Flash Translation Layer (FTL) receives the instructions and then assigns the tasks and executes; when the physical layer drives are in an executable state; the arbitration unit drives the flash array with the physical layer drivers; in other words, when the physical layer When the drive is busy, the task to be executed must wait until the previous task is executed before it can be executed; therefore, the execution order of the task will affect the performance of the storage system, so an improved method is proposed for the previous technology.

Summary of the invention

It is an object of the present invention to provide a task scheduling optimization method for use in a storage system to solve the problems raised in the above background art.

To achieve the above objective, the present invention provides the following technical solution: a task scheduling optimization method applied in a storage system, including a host, a flash translation layer, a virtual drive, and a flash array, wherein the host connects to a flash translation layer through a bus, The virtual drive is connected to the flash array through a bus. The virtual drive is provided with a plurality of task queues, a task scheduling unit, a physical layer driver, an execution scheduling unit, and a plurality of execution queues, and the plurality of task queues include a first task queue and a second task. a queue, a third task queue, and a fourth task queue, the plurality of execution queues including a first execution queue, a second execution queue, a third execution queue, and a fourth execution queue; the user data access unit is disposed in the flash translation layer a garbage collection unit, an average wipe control unit, and a metadata management unit; the user data access unit is connected to the first task queue; the garbage collection unit is connected to the second task queue; and the average erase control unit is connected to the third a task queue; the metadata management unit is connected to a fourth task queue; The task queue, the second task queue, the third task queue, and the fourth task queue are respectively connected to the task scheduling unit; the task scheduling unit is connected to the physical layer driver, and the physical layer driver is respectively connected to the first execution queue, the second execution queue, The third execution queue and the fourth execution queue.

Preferably, the steps are as follows:

A. Classify the task type of each task in a large number of tasks;

B. The task is placed in the task queue according to the task type, and each task queue stores a task type;

C. assign a priority to each type of task;

D. Converting the first task stored in the task queue; the task generates at least one command into the execution queue, and at least one command is executed in the execution queue according to the priority level.

Compared with the prior art, the beneficial effects of the present invention are: in the prior art, without prior use, the tasks are processed in order; if the physical layer driver is in a busy state, the task needs to remain in a waiting state until the previous task has been completed; Therefore, the physical layer driver must first end the previous task before starting to execute the next task, resulting in the task that needs to be executed first needs to wait for the previous task to be executed before executing; in contrast, the present invention can be prioritized by assigning tasks. Right to improve the performance of the storage system.

DRAWINGS

Figure 1 is a schematic structural view of the present invention;

Figure 2 is a flow chart of the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention are clearly and completely described in the following with reference to the accompanying drawings in the embodiments of the present invention. It is obvious that the described embodiments are only a part of the embodiments of the present invention, but not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts are within the scope of the present invention.

Referring to FIG. 1, the present invention provides a technical solution: a task scheduling optimization method applied in a storage system, including a host 1, a flash translation layer 2, a virtual drive 3, and a flash array 4. The host 1 is connected to the flash translation layer 2 via a bus. The virtual drive 3 is connected to the flash array 4 via a bus. The virtual drive 3 is provided with a plurality of task queues, a task scheduling unit 5, a physical layer driver 6, and an execution scheduling unit. 7 and a plurality of execution queues, the plurality of task queues include a first task queue 8, a second task queue 9, a third task queue 10, and a fourth task queue 11, the plurality of execution queues including the first execution queue 12, and the second execution a queue 13, a third execution queue 14 and a fourth execution queue 15; the flash data conversion layer 2 is provided with a user data access unit 16, a garbage collection unit 17, an average wipe control unit 18, and a metadata management unit 19; The user data access unit 16 is connected to the first task queue 8; the garbage collection unit 17 is connected to the second task queue 9; the average erase control unit 18 is connected to the third task queue 10; the metadata management unit 19 is connected a fourth task queue 15; the first task queue 8, the second task queue 9, the third task queue 10, and the fourth task queue 11 are respectively connected to the task scheduling unit 5; the task scheduling unit 5 The physical layer 6 then drives the physical layer driver 6 are connected to a first execution queue 12, the second execution queue 13, the third 14 and the fourth execution queue execution queue 15. The task scheduling unit is configured to determine the priority of the task and the conversion task becomes a command; the execution scheduling unit allocates the order of the commands in the execution queue according to the task priority; the priority may be adjusted according to the state of the storage system, for example, if the flash array The available space is insufficient or less than a critical value. The garbage collection task has the highest priority. After the garbage collection task is completed, the physical layer driver can drive the flash to erase and reclaim the available space for the next operation, if the storage system The performance is important, the user data access task will have the highest priority, thereby maintaining a large amount of data throughput; in practical applications, priority arrangements can be made according to different conditions to optimize task execution.

The steps of the present invention are as follows:

A. Classify the task type of each task in a large number of tasks;

C. assign a priority to each type of task;

In the prior art, without priority use, tasks are processed in order; if the physical layer driver is busy, the task needs to remain in a wait state until the previous task has been completed; therefore, the physical layer driver must first end the previous task before The execution of the next task begins, and the task that needs to be executed first needs to wait for the previous task to be executed before execution; in contrast, the present invention can improve the performance of the storage system by assigning task priorities.

While the embodiments of the present invention have been shown and described, it will be understood by those skilled in the art The scope of the invention is defined by the appended claims and their equivalents.

Claims

A task scheduling optimization method applied in a storage system, comprising a host (1), a flash translation layer (2), a virtual drive (3) and a flash array (4), characterized in that: the host (1) passes through a bus Connecting a flash translation layer (2), the virtual drive (3) is connected to the flash array (4) through a bus, and the virtual drive (3) is provided with a plurality of task queues, a task scheduling unit (5), and a physical layer driver ( 6) executing a scheduling unit (7) and a plurality of execution queues, the plurality of task queues including a first task queue (8), a second task queue (9), a third task queue (10), and a fourth task queue (11) a plurality of execution queues including a first execution queue (12), a second execution queue (13), a third execution queue (14), and a fourth execution queue (15); the flash conversion layer (2) is provided a user data access unit (16), a garbage collection unit (17), an average wipe control unit (18), and a metadata management unit (19); the user data access unit (16) is connected to the first task queue (8) The garbage collection unit (17) is connected to the second task queue (9); the average erase control unit (18) is connected to the third task queue (10); The metadata management unit (19) is connected to the fourth task queue (15); the first task queue (8), the second task queue (9), the third task queue (10), and the fourth task queue (11) Connected to the task scheduling unit (5); the task scheduling unit (5) is connected to the physical layer driver (6), and the physical layer driver (6) is connected to the first execution queue (12) and the second execution queue (13), respectively. The third execution queue (14) and the fourth execution queue (15).
A task scheduling optimization method applied in a storage system according to claim 1, wherein the steps are as follows:

A. Classify the task type of each task in a large number of tasks;

B. The task is placed in the task queue according to the task type, and each task queue stores a task type;

C. assign a priority to each type of task;

D. Converting the first task stored in the task queue; the task generates at least one command into the execution queue, and at least one command is executed in the execution queue according to the priority level.