WO2021174731A1

WO2021174731A1 - Disk performance optimization method, apparatus and device, and computer readable storage medium

Info

Publication number: WO2021174731A1
Application number: PCT/CN2020/099512
Authority: WO
Inventors: 刘仁仕
Original assignee: 平安科技（深圳）有限公司
Priority date: 2020-03-05
Filing date: 2020-06-30
Publication date: 2021-09-10
Also published as: CN111459405A

Abstract

The present application relates to the technical field of clouds and is applied to artificial intelligence, and discloses a disk performance optimization method comprising the following steps: acquiring a cluster offset, and allocating a cache item on the basis of the cluster offset, the cluster offset being the difference between the initial size of a cluster and the actual size of the cluster; configuring a flash parameter within the cache item, the flash parameter comprising dirty bit granularity and dirty bit map size; determining a target dirty bit map according to the dirty bit map size and a preset formula and initializing the target dirty bit map; determining whether a disk has received a request for flashing data; and if the disk has received a request for flashing data, then identifying corresponding dirty data by means of the target dirty bit map, and flashing the dirty data. Further disclosed by the present application are a disk performance optimization apparatus and device, and a computer readable storage medium. The disk performance optimization method provided by the present application improves the performance of a disk.

Description

Disk performance optimization method, device, equipment and computer readable storage medium

This application claims the priority of a Chinese patent application filed with the Chinese Patent Office on March 5, 2020, the application number is 202010146235.7, and the invention title is "Disk performance optimization methods, devices, equipment, and computer-readable storage media", and its entire content Incorporated in the application by reference.

Technical field

This application relates to the technical field of cache data distribution, and is applied to artificial intelligence, and in particular to a method, device, device, and computer-readable storage medium for optimizing disk performance.

Background technique

In the prior art, due to the difference in the data sequence written by the host, hot and cold data will be generated. The frequency of cold data being updated is very low, the block occupied by it is rarely erased, and the block occupied by hot data is erased many times. The inventor realizes that if special wear leveling is not done, blocks occupied by hot data are frequently used and will reach their lifespan earlier. Therefore, it is necessary to continuously move cold data to blocks with more erasing times, and write hot data to blocks with fewer erasing times to achieve the effect of wear leveling. The result of this will inevitably increase the number of data moved. The number of times, resulting in the phenomenon of write amplification, resulting in the performance of the virtual disk degraded.

At the same time, the inventor found that as long as there is an item update in the cache item, the file will be written in a whole block during the cache. Taking the cluster size of 4M as an example, assuming that the user writes 4K randomly, when the random write 4K area needs to be allocated In the case of a new cluster, these write operations will cause the flushing of 4M secondary index table cache entries and the flushing of 4M reference counting table cache entries, and an additional 8M metadata write amplification. As a result, the actual used space of the virtual disk decreases and performance decreases.

Summary of the invention

The main purpose of this application is to provide a method, device, device, and computer-readable storage medium for optimizing disk performance, aiming to solve the technical problem of low disk performance.

To achieve the above objective, the present application provides a method for optimizing disk performance. The method for optimizing disk performance includes the following steps:

Acquiring a cluster offset, and allocating a cache entry based on the cluster offset, where the cluster offset is the difference between the initial size of the cluster and the actual size of the cluster;

Set flushing parameters in the cache entry, the flushing parameters include dirty bit granularity and dirty bitmap size, where the dirty bitmap size is equal to the ratio of the actual size of the cluster to the dirty bit granularity, the The dirty bitmap size is used to indicate the total number of bits of the dirty bitmap storing dirty data, and the dirty bit granularity is used to indicate the amount of data represented by each bit;

Determine a target dirty bitmap according to the size of the dirty bitmap and a preset formula, and initialize the target dirty bitmap;

Determine whether the disk has received a request for flashing data;

If the disk receives the request for flushing data, the corresponding dirty data is identified through the target dirty bitmap, and the dirty data is flushed.

Further, in order to achieve the above object, the present application also provides a disk performance optimization device, the disk performance optimization device includes the following modules:

An obtaining module, configured to obtain a cluster offset, and allocate a cache entry based on the cluster offset, where the cluster offset is the difference between the initial size of the cluster and the actual size of the cluster;

The setting module is used to set flashing parameters in the cache item, the flashing parameters including dirty bit granularity and dirty bitmap size, wherein the dirty bitmap size is equal to the actual size of the cluster and the dirty bit granularity The dirty bitmap size is used to indicate the total number of bits of the dirty bitmap storing dirty data, and the dirty bit granularity is used to indicate the amount of data represented by each bit;

A dirty bitmap allocation module, configured to determine a target dirty bitmap according to the size of the dirty bitmap and a preset formula, and initialize the target dirty bitmap;

The flashing data judgment module is used to judge whether the disk has received a request for flashing data;

The dirty data identification module is configured to, if the disk receives the request for flushing data, identify the corresponding dirty data through the target dirty bitmap, and flush the dirty data.

Further, in order to achieve the above object, the present application also provides a disk performance optimization device, the disk performance optimization device includes a memory, a processor, and a disk performance optimization device that is stored in the memory and can run on the processor. A program that, when the disk performance optimization program is executed by the processor, implements the steps of the disk performance optimization method as described below, including:

Determine whether the disk has received a request for flashing data;

Further, in order to achieve the above-mentioned object, the present application also provides a computer-readable storage medium having a disk performance optimization program stored on the computer-readable storage medium, and the disk performance optimization program is executed by a processor to achieve the following The steps of the disk performance optimization method include:

Determine whether the disk has received a request for flashing data;

This application uses a preset marking program to mark all bits of the dirty bitmap in the cache item, and flush the dirty data in the dirty bitmap in the cache item to the file, avoiding repeated data flushing, reading and writing The smallest unit of data is page page, 1page is generally 4KB, and the smallest unit of erased data is block, 1block is 64 pages, that is, 256KB, ideally, how much data is written to the main memory and how much to write to the flash memory Data, but in the actual scenario, the amount of dirty data must be written out in calculations, that is, the data consumes disk performance when writing, and it consumes disk performance when writing, thus causing write amplification and write amplification. Is the ratio of the amount of data actually written to the amount of data that should be written, for example, 4KB that should be written, 8KB actually written (4KB when writing, if it is dirty data, 4KB of dirty data Erase, so it is 8KB, but the smallest unit of erase is 256KB). This application recognizes dirty data, only flushes dirty data, avoids flushing all data, reduces write amplification, and improves disk performance.

Description of the drawings

FIG. 1 is a schematic structural diagram of an operating environment of a disk performance optimization device related to a solution of an embodiment of the application;

FIG. 2 is a schematic flowchart of a first embodiment of a method for optimizing disk performance of this application;

FIG. 3 is a detailed flowchart of an embodiment of step S50 in FIG. 2;

4 is a schematic flowchart of a second embodiment of a method for optimizing disk performance according to this application;

FIG. 5 is a schematic diagram of functional modules of an embodiment of an apparatus for optimizing disk performance of the present application.

Detailed ways

It should be understood that the specific embodiments described here are only used to explain the application, and are not used to limit the application.

The disk performance optimization method involved in the embodiments of the present application is mainly applied to a disk performance optimization device, and the disk performance optimization device may be a device with display and processing functions such as a PC, a portable computer, and a mobile terminal.

Referring to FIG. 1, FIG. 1 is a schematic diagram of the hardware structure of the disk performance optimization device involved in the solution of the embodiment of the application. In the embodiment of the present application, the disk performance optimization device may include a processor 1001 (for example, a CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components; the user interface 1003 may include a display (Display), an input unit such as a keyboard (Keyboard); the network interface 1004 may optionally include a standard wired interface, a wireless interface (Such as WI-FI interface); the memory 1005 can be a high-speed RAM memory, or a non-volatile memory (non-volatile memory), such as a disk memory. The memory 1005 can optionally be a storage device independent of the aforementioned processor 1001 .

Those skilled in the art can understand that the hardware structure shown in FIG. 1 does not constitute a limitation on the disk performance optimization device, and may include more or less components than shown in the figure, or a combination of certain components, or different component arrangements. .

Continuing to refer to FIG. 1, the memory 1005 as a computer-readable storage medium in FIG. 1 may include an operating system, a network communication module, and a disk performance optimization program, where the disk performance optimization program here implements any of the following embodiments Method of optimizing disk performance.

In FIG. 1, the network communication module is mainly used to connect to the server and perform data communication with the server; and the processor 1001 can call the disk performance optimization program stored in the memory 1005, and execute the disk performance optimization method provided in the embodiment of the present application.

The embodiment of the present application provides a method for optimizing disk performance.

Referring to FIG. 2, FIG. 2 is a schematic flowchart of the first embodiment of the disk performance optimization method of the present application. In this embodiment, the disk performance optimization method includes:

Step S10: Obtain a cluster offset, and allocate a cache entry based on the cluster offset, where the cluster offset is the difference between the initial cluster size and the actual size of the cluster;

In this embodiment, the cluster offset is the difference between the initial cluster size and the actual size of the cluster. After the cluster offset is obtained, the cache entry will be allocated according to the size of the cluster offset.

Step S20: Set flashing parameters in the cache item. The flashing parameters include dirty bit granularity and dirty bitmap size, where the dirty bitmap size is equal to the ratio of the actual size of the cluster to the dirty bit granularity, and the dirty bitmap size Used to indicate the total number of bits in the dirty bitmap for storing dirty data, and the dirty bit granularity is used to indicate the amount of data represented by each bit;

In this embodiment, the dirty bitmap size is the ratio of the cluster size to the dirty bit granularity. When the target dirty bitmap is determined according to the dirty bitmap size and a preset formula and the target dirty bitmap is initialized, the preset formula is : Target dirty bitmap = dirty bitmap size/8. The amount of data written in the block can be monitored as a basis for whether data needs to be flushed. For example, when the amount of data occupies 50% of the block space, it indicates that there is a request for flushing data.

The dirty bitmap, dirty bitmap size, and dirty bit granularity are preset in the Qcow2 virtual disk cache item. Among them, the dirty bitmap represents the dirty data area bitmap of the cache item, and the dirty bitmap size represents the dirty bitmap. The total number of digits, the dirty bit granularity represents the amount of data in the cache item represented by each dirty bitmap. Among them, the dirty bitmap represents the dirty data area bitmap of the cache item, the dirty bitmap size represents the total number of bits of the dirty bitmap, and the dirty bit granularity represents the amount of data in the cache item represented by each dirty bitmap. The dirty bit granularity parameter defaults to 4KB.

Set a new cache entry dirty bit granularity parameter for the qcow2 virtual disk, where the parameter defaults to 4KB.

Step S30: Determine the target dirty bitmap according to the size of the dirty bitmap and the preset formula, and initialize the target dirty bitmap;

In this embodiment, the preset formula is

Data storage disks are based on clusters. When clusters are allocated and released, the reference count of the cluster needs to be updated. According to the corresponding cluster index, the cluster index corresponds to the cluster, and the corresponding relationship can be established through mapping. The cluster index is pre-deployed in In the cache item, when the number of cluster references in the cache item is zero, the cluster is released; otherwise, the reference count item is updated, and the corresponding bit in the dirty bitmap is updated to 1 according to the cluster index in the cache item. The calculation method of several dirty bits is: index in cache item * total number of digits of reference count/8/granularity of dirty bits.

Step S40, it is judged whether the disk has received a request for flashing data;

In step S50, if the disk receives a request for flushing data, it identifies the corresponding dirty data through the target dirty bitmap, and flushes the dirty data.

In this embodiment, for the secondary index table cache entry after the cluster is written, the corresponding dirty_bit is calculated through the secondary index. The specific calculation method of the dirty bit is the secondary index*8/dirty bit granularity. When qcow2 is called During the virtual disk data flushing process, for the updated cache item, the dirty data area that needs to be flushed is determined according to the starting file cluster offset of the cache item, the dirty bitmap, and the dirty bit granularity, and flushed to the back-end file .

Mark all the bits of the dirty bitmap in the cache item with a bitmap, and flush the dirty data in the dirty bitmap of the cache item into the file to avoid repeated flushing of data. The smallest unit for reading and writing data is page. 1page is generally 4KB, and the smallest unit of erased data is a block, and 1block is 64 pages, that is, 256KB. In an ideal state, how much data is written to the main memory and how much data must be written to the flash memory, but in actual scenarios, The amount of dirty data must also be written out in calculations, that is, the data consumes disk performance when writing, and disk performance when writing, which causes write amplification, which refers to the amount of data actually written The ratio of the amount of data that should be written, for example, the 4KB that should be written, the actual write is 8KB (4KB when writing, if it is dirty data, the dirty data of 4KB must be erased, so it is 8KB, but The smallest unit of erasing is 256KB). This application recognizes dirty data, only flushes dirty data, avoids flushing all data, reduces write amplification, and improves disk performance.

Referring to FIG. 3, FIG. 3 is a detailed flowchart of an embodiment of step S50 in FIG. 2. In this embodiment, step S50 includes:

Step S501, if the disk receives a request for flushing data, it is determined whether any reference count item in the reference count block has been updated, where the reference count block is a module used to record the reference count item in the disk, and the reference count item is used Bitmap with data in the record buffer;

In this embodiment, if the virtual disk currently has a request for flushing data, it is determined whether the number of initial cache item indexes is less than the number of cache items in the cache table.

In step S502, if any reference count item in the reference count block is updated, the corresponding dirty data is identified through the target dirty bitmap, and the dirty data is flushed.

In this embodiment, if any reference count item in the reference count block is updated, it means that there is data in the dirty data state. If there is data in the dirty data state, it can be judged by the dirty bitmap in the cache item Which areas in the cache item are dirty, and flush the dirty areas to the file.

All bits of the dirty bitmap in the cache item are marked by the preset marking program, and the dirty data in the dirty bitmap in the cache item is flushed into the file, avoiding repeated flushing of data, reducing write amplification, and improving Disk performance.

Referring to FIG. 4, FIG. 4 is a schematic flowchart of a second embodiment of a method for optimizing disk performance of the present application. In this embodiment, before step S10, the method further includes:

Step S60: It is judged whether the primary index entry is indexed to the historical secondary index table, the primary index entry has a preset mapping relationship with the historical secondary index table, and the primary index entry is used to index the historical secondary index table;

In this embodiment, the purpose of judging whether the primary index item is directly usable is to determine whether the secondary index table can be directly loaded from the file through the cluster offset of the secondary index table indicated in the primary index item, because for the second There are several situations for the level index table to consider: (1) Active primary index entry (ie, primary index table used for real-time reading and writing), where the secondary index table is directly available, that is, the secondary index table has been allocated and It is not indexed by the first-level index entries of other snapshots; (2) Allocated but also indexed by entries in the first-level index tables of other snapshots; (3) Not allocated, the first-level index entry is 0 at this time.

In step S70, if the primary index entry is indexed into the historical secondary index table, the historical secondary index table is loaded into the cache entry.

In this embodiment, each primary index item is 8 bytes, that is, 64 bits, the highest bit indicates whether the index item is directly usable, and the 0th to 55th bits, that is, the lower 56 bits, indicate the cluster offset of the secondary index table. The highest bit indicates that the index item is directly available, and the secondary index table is loaded from the file into the cache.

Writing all 0s to the cache entries means that the clusters indexed by the secondary index table have not been allocated. Because qcow2 supports allocation on writing, that is, clusters are allocated only when they are written to a specific location. Up to this step, the cluster to be written has not been allocated, but has reached the stage of allocating the secondary index table cluster, so the secondary index table must be written to all 0 first. Pre-set the rules for marking all the bits of the dirty bitmap in the cache item through the preset marking program, and flushing the dirty data in the dirty bitmap of the cache item into the file, and then the corresponding dirty bitmap in the cache item , Marked as 1 is achieved through automatic program identification, because this scheme has changed the cache item flushing process to flush according to the dirty bitmap of the cache item. All bits of the dirty bitmap in the cache item are marked by the preset marking program, and the dirty data in the dirty bitmap in the cache item is flushed into the file, avoiding repeated flushing of data, reducing write amplification, and improving Disk performance.

The following is the third embodiment of the method for optimizing disk performance of this application. In this embodiment, before step S10, it further includes:

Step S90, index the target secondary index table through the primary index item;

Step S100: Allocate clusters for the target secondary index table.

In this embodiment, the secondary index mechanism is adopted, so it is necessary to index to the target secondary index table according to the preset mapping relationship between the primary index item and the target secondary index table, which is the target secondary index table To allocate clusters, each secondary index item is 64 bits, and the highest bit is 1 to indicate that it is directly available. At this time, there is no need to allocate a new cluster, otherwise, it means that a new cluster needs to be allocated. The cluster needs to allocate a new cluster. There are also two situations. One is that the cluster is not allocated, and the secondary index entries are all 0s. The other is that the cluster is indexed by other snapshots. In both cases, write copy is required, and the cluster is not allocated. When the qcow2 virtual disk has a parent volume, data needs to be read from the parent volume, and data needs to be copied from the original cluster when the cluster is indexed by other snapshots.

In this embodiment, if the offset of the secondary index table cluster indicated by the primary index entry is all 0, it means that the secondary index table has not been allocated, that is, there is no historical secondary index table. If it is not all 0, it means that there is a historical secondary Level index table; when the primary index table entry indicates that the offset of the secondary index table cluster is not all 0, the old secondary index table needs to be copied.

The following is the fourth embodiment of the disk performance optimization method of this application. In this embodiment, after step S20, the method further includes:

Step S110, judging whether the cluster is directly available;

Step S120, if the clusters are directly available, calculate the number of available clusters,

In this embodiment, the calculation method is:

(1) The requested secondary index is the starting secondary index;

(2) Find the end point secondary index, if (cluster offset + length)/(cluster size*number of secondary index items in cluster) is equal to (cluster offset/(cluster size*number of secondary index items in cluster) ), which means that the clusters in the entire area to be written this time can be indexed through the same secondary index table, the end secondary index = ((cluster offset + length)/cluster size)% (secondary index item), otherwise the end 2 The level index entry is the number of secondary index entries in the cluster minus one;

(3) Let the index be the initial secondary index plus 1, the number of consecutive clusters is 1, and the maximum value of the index is the desired end index. If the desired secondary index item is set with the directly available flag and the table cluster offset is the previous The cluster offset of an index entry plus the cluster size, then the consecutive clusters plus 1, otherwise, the writable length is MIN ((consecutive cluster number * cluster size-cluster offset within cluster offset), length), where cluster Offset cluster offset within cluster = cluster offset% cluster size, the cluster offset in the file is the cluster file cluster offset in the starting secondary index entry plus the cluster offset cluster offset within the cluster.

The following is the fifth embodiment of the method for optimizing disk performance of this application. In this embodiment, step S110 includes:

Step S1101: Obtain a starting secondary index entry according to the target secondary index table, where the target secondary index table and the starting secondary index entry have a preset mapping relationship;

Step S1102, judging whether the highest bit of the cluster is a preset value according to the initial secondary index entry;

In step S1103, if the highest bit of the initial secondary index entry is a preset value, it is determined that the cluster is directly available.

In this embodiment, each secondary index item is 64 bits. If the highest bit is 1, it means that it is directly available. At this time, there is no need to allocate a new cluster. If the highest bit is 0, it means that a new cluster needs to be allocated. There are also two situations when a new cluster needs to be allocated. One is that the cluster is not allocated, and the secondary index entries are all 0s. The other is that the cluster is indexed by other snapshots. In both cases, write copy is required. When the cluster is not allocated When the qcow2 virtual disk has a parent volume, data needs to be read from the parent volume, and data needs to be copied from the original cluster when the cluster is indexed by other snapshots.

The following is a detailed flow of step S120. In this embodiment, step S120 includes:

Step S1201, calculate separately

and

The value of to determine the

The value of the

Whether the values of are equal, where P is the cluster offset, L is the preset length of the cluster, B is the actual size of the cluster, and N is the number of secondary index entries in the cluster;

Step S1202, if

And the value of

If the value of is not equal, the number of secondary index entries in the cluster is subtracted by 1 to obtain the number of directly available clusters.

The preset length of the cluster is preset. For example, the preset length of the cluster can be preset to 64KB. If the actual size of the cluster is 64KB, the number of clusters with a length of 64KB is 1. According to the cluster bias of the secondary index table Shift, the starting cluster offset of the allocated clusters and the number of consecutively allocated clusters, update the secondary index entries in the secondary index cache entries and record the historical secondary index entries, and update the cache entries according to the updated secondary index cache entries Dirty bit, release the cluster represented by the historical secondary index entry (ie dereference), that is, decrement the reference count of the listed cluster by one.

When Qemu-kvm writes a qcow2 virtual disk, there are two main situations for flushing the cache: 1. The qcow2 virtual disk disables the cache, then after each write operation, the secondary index table cache will be flushed (only the updated ones will be flushed). Secondary index cache items, that is, new clusters are allocated) and reference count block cache (only refreshed reference count block cache items, that is, new clusters are allocated); 2. The guest operating system calls FLUSH (flushing). During the cache operation, the secondary index table cache and the reference counting block cache will be flushed to the file.

For the original implementation, as long as there is an item update in the cache item, the file will be written in a whole block during the cache. Let's take a cluster size of 4M as an example, assuming that the user write model is random write 4K, when random write When a new cluster is allocated in a 4K area, these write operations will cause the flushing of 4M secondary index table cache entries and 4M reference counting table cache entries, which will additionally cause 8M metadata write amplification.

When adding a dirty bitmap to the cache item, assuming that each dirty bit represents a 4KB area, then this metadata write amplification can reduce 4KB+4KB=8KB, which greatly reduces the need to allocate new clusters for write operations Time to bring the write amplification of metadata.

Referring to FIG. 5, FIG. 5 is a schematic diagram of functional modules of an embodiment of the disk performance optimization apparatus of the present application. In this embodiment, the disk performance optimization device includes:

The obtaining module 10 is configured to obtain a cluster offset, and allocate a cache entry based on the cluster offset, where the cluster offset is the difference between the initial size of the cluster and the actual size of the cluster;

The setting module 20 is configured to set flushing parameters in the cache item. The flushing parameters include dirty bit granularity and dirty bitmap size, where the dirty bitmap size is equal to the actual size of the cluster and the dirty bitmap. The ratio of the granularity, the size of the dirty bitmap is used to indicate the total number of bits of the dirty bitmap storing dirty data, and the granularity of the dirty bit is used to indicate the amount of data represented by each bit;

The dirty bitmap allocating module 30 is configured to determine a target dirty bitmap according to the size of the dirty bitmap and a preset formula and initialize the target dirty bitmap;

The data flashing judgment module 40 is used to judge whether the disk has received a request for flashing data;

The dirty data identification module 50 is configured to, if the disk receives the request for flushing data, identify the corresponding dirty data through the target dirty bitmap, and flush the dirty data.

In this embodiment, the dirty data identification module 30 includes:

The reference count item determining unit is configured to determine whether any reference count item in the reference count block has been updated if the disk receives a request for flushing data, wherein the reference count block is used to record references in the disk A module for counting items, where the reference counting item is used to record a bitmap with data buffered;

The target dirty bitmap identification unit is configured to, if any reference count item in the reference count block is updated, identify the corresponding dirty data through the target dirty bitmap, and flush the dirty data.

In this embodiment, the disk performance optimization device further includes:

The historical secondary index table index judgment module is used to judge whether the primary index entry is indexed to the historical secondary index table, the primary index entry has a preset mapping relationship with the historical secondary index table, and the primary index Item is used to index the historical secondary index table;

The loading module is configured to load the historical secondary index table into the cache entry if the primary index entry is indexed into the historical secondary index table.

In this embodiment, the disk performance optimization device further includes:

The target secondary index table index module is used to index the target secondary index table through the primary index item;

The secondary index table allocating cluster module is used for allocating clusters for the target secondary index table.

Optionally, the disk performance optimization device further includes:

The cluster directly usable judgment module is used to judge whether the cluster is directly usable;

The calculation module is used to calculate the number of directly available clusters if the clusters are directly available.

In this embodiment, the cluster directly usable judging module includes:

A starting secondary index entry acquiring unit, configured to acquire a starting secondary index entry according to a target secondary index table, wherein the target secondary index table and the starting secondary index entry have a preset mapping relationship;

The highest bit determination unit of the cluster is configured to determine whether the highest bit of the cluster is a preset value according to the initial secondary index entry;

The cluster directly available determining unit is configured to determine that the cluster is directly available if the highest bit of the initial secondary index entry is a preset value.

In this embodiment, the calculation module includes the following units:

Whether the values are equal or not judging unit, used to calculate separately

and

The value of to determine the

The value of the

The number of directly available clusters obtains the unit, which is used if

And the value of

The present application also provides a computer-readable storage medium. The computer-readable storage medium may be non-volatile or volatile.

In this embodiment, a disk performance optimization program is stored on the computer-readable storage medium, and the disk performance optimization program is executed by a processor to implement the steps of the disk performance optimization method described in any of the above embodiments.

Through the description of the above implementation manners, those skilled in the art can clearly understand that the above-mentioned embodiment method can be implemented by means of software plus the necessary general hardware platform, of course, it can also be implemented by hardware, but in many cases the former is better.的实施方式。 Based on this understanding, the technical solution of this application essentially or the part that contributes to the existing technology can be embodied in the form of a software product, and the computer software product is stored in a storage medium (such as ROM/RAM), including Several instructions are used to make a terminal (which can be a mobile phone, a computer, a server or a network device, etc.) execute the methods described in the various embodiments of the present application.

The embodiments of the application are described above with reference to the accompanying drawings, but the application is not limited to the above-mentioned specific embodiments. The above-mentioned specific embodiments are only illustrative and not restrictive. Those of ordinary skill in the art are Under the enlightenment of this application, without departing from the purpose of this application and the scope of protection of the claims, many forms can be made, any equivalent structure or equivalent process transformation made by using the content of the description and drawings of this application, or It is directly or indirectly used in other related technical fields, and these all fall within the protection of this application.

Claims

A method for optimizing disk performance, wherein the method for optimizing disk performance includes:

Acquiring a cluster offset, and allocating a cache entry based on the cluster offset, where the cluster offset is the difference between the initial size of the cluster and the actual size of the cluster;

Set flushing parameters in the cache entry, the flushing parameters include dirty bit granularity and dirty bitmap size, where the dirty bitmap size is equal to the ratio of the actual size of the cluster to the dirty bit granularity, the The dirty bitmap size is used to indicate the total number of bits of the dirty bitmap storing dirty data, and the dirty bit granularity is used to indicate the amount of data represented by each bit;

Determine a target dirty bitmap according to the size of the dirty bitmap and a preset formula, and initialize the target dirty bitmap;

Determine whether the disk has received a request for flashing data;

If the disk receives the request for flushing data, the corresponding dirty data is identified through the target dirty bitmap, and the dirty data is flushed.
The method for optimizing the performance of a disk according to claim 1, wherein if the disk receives the request for flushing data, the target dirty bitmap is used to identify the corresponding dirty data, and the dirty Data flashing, including:

If the disk receives a request for flushing data, it is determined whether any reference count item in the reference count block has been updated, wherein the reference count block is a module for recording reference count items in the disk, and the reference count The count item is used to record the bitmap with data in the buffer;

If any reference count item in the reference count block is updated, the corresponding dirty data is identified through the target dirty bitmap, and the dirty data is flushed.
5. The method for optimizing disk performance according to claim 1, wherein after said obtaining the cluster offset and allocating a cache entry based on the cluster offset, a flushing parameter is set in the cache entry, and the flushing parameter is Before including the dirty bit granularity and the dirty bit bitmap size, it also includes:

Judging whether the primary index entry is indexed to the historical secondary index table, the primary index entry has a preset mapping relationship with the historical secondary index table, and the primary index entry is used to index the historical secondary index table;

If the primary index entry is indexed into the historical secondary index table, the historical secondary index table is loaded into the cache entry.
The method for optimizing disk performance according to claim 3, wherein, in the obtaining of the cluster offset, a cache entry is allocated based on the cluster offset, and the cluster offset is the difference between the initial size of the cluster and the actual size of the cluster. Before the difference, it also includes:

Index the target secondary index table through the primary index item;

Allocate clusters for the target secondary index table.
The method for optimizing disk performance according to claim 4, wherein after the clusters are allocated to the target secondary index table, after the cluster offset is obtained, a cache entry is allocated based on the cluster offset, and the cluster Before the offset is the difference between the initial size of the cluster and the actual size of the cluster, it also includes:

Determine whether the cluster is directly available;

If the clusters are directly available, calculate the number of directly available clusters.
7. The method for optimizing disk performance according to claim 5, wherein said determining whether the cluster is directly available comprises:

Obtaining a starting secondary index entry according to the target secondary index table, where the target secondary index table and the starting secondary index entry have a preset mapping relationship;

Judging whether the highest bit of the cluster is a preset value according to the initial secondary index item;

If the highest bit of the initial secondary index entry is a preset value, it is determined that the cluster is directly available.
The method for optimizing disk performance according to claim 5, wherein, if the clusters are directly available, calculating the number of directly available clusters includes:

Calculate separately
and
The value of to determine the
The value of the
Whether the values of are equal, where P is the cluster offset, L is the preset length of the cluster, B is the actual size of the cluster, and N is the number of secondary index entries in the cluster;

like
And the value of
If the value of is not equal, the number of secondary index entries in the cluster is subtracted by 1 to obtain the number of directly available clusters.
A disk performance optimization device, wherein the disk performance optimization device includes the following modules:

An obtaining module, configured to obtain a cluster offset, and allocate a cache entry based on the cluster offset, where the cluster offset is the difference between the initial size of the cluster and the actual size of the cluster;

The setting module is used to set flashing parameters in the cache item, the flashing parameters including dirty bit granularity and dirty bitmap size, wherein the dirty bitmap size is equal to the actual size of the cluster and the dirty bit granularity The dirty bitmap size is used to indicate the total number of bits of the dirty bitmap storing dirty data, and the dirty bit granularity is used to indicate the amount of data represented by each bit;

A dirty bitmap allocation module, configured to determine a target dirty bitmap according to the size of the dirty bitmap and a preset formula, and initialize the target dirty bitmap;

The flashing data judgment module is used to judge whether the disk has received a request for flashing data;

The dirty data identification module is configured to, if the disk receives the request for flushing data, identify the corresponding dirty data through the target dirty bitmap, and flush the dirty data.
A disk performance optimization device, wherein the disk performance optimization device includes a memory, a processor, and a disk performance optimization program that is stored on the memory and can run on the processor, and the disk performance optimization program is The steps of the method for optimizing disk performance as described below are implemented when the processor is executed:

Acquiring a cluster offset, and allocating a cache entry based on the cluster offset, where the cluster offset is the difference between the initial size of the cluster and the actual size of the cluster;

Set flushing parameters in the cache entry, the flushing parameters include dirty bit granularity and dirty bitmap size, where the dirty bitmap size is equal to the ratio of the actual size of the cluster to the dirty bit granularity, the The dirty bitmap size is used to indicate the total number of bits of the dirty bitmap storing dirty data, and the dirty bit granularity is used to indicate the amount of data represented by each bit;

Determine a target dirty bitmap according to the size of the dirty bitmap and a preset formula, and initialize the target dirty bitmap;

Determine whether the disk has received a request for flashing data;

If the disk receives the request for flushing data, the corresponding dirty data is identified through the target dirty bitmap, and the dirty data is flushed.
The disk performance optimization device according to claim 9, wherein the disk performance optimization program is executed by the processor, and if the disk receives the request for flushing data, it passes the target dirty bit When the graph identifies the corresponding dirty data and performs the steps of flashing the dirty data, the following steps are also performed:

If the disk receives a request for flushing data, it is determined whether any reference count item in the reference count block has been updated, where the reference count block is a module used to record the reference count item in the disk, and the reference count The count item is used to record the bitmap with data in the buffer;

If any reference count item in the reference count block is updated, the corresponding dirty data is identified through the target dirty bitmap, and the dirty data is flushed.
9. The disk performance optimization device according to claim 9, wherein the disk performance optimization program is executed by the processor to obtain the cluster offset and allocate a cache item based on the cluster offset. Before setting the flashing parameters, the flashing parameters include the dirty bit granularity and the dirty bitmap size, the following steps are also performed:

Judging whether the primary index entry is indexed to the historical secondary index table, the primary index entry has a preset mapping relationship with the historical secondary index table, and the primary index entry is used to index the historical secondary index table;

If the primary index entry is indexed into the historical secondary index table, the historical secondary index table is loaded into the cache entry.
The disk performance optimization device according to claim 11, wherein the disk performance optimization program is executed by the processor, the acquiring cluster offset, and the cache entry is allocated based on the cluster offset, and the cluster offset is Before the step for the difference between the initial size of the cluster and the actual size of the cluster, the following steps are also performed:

Index the target secondary index table through the primary index item;

Allocate clusters for the target secondary index table.
The disk performance optimization device according to claim 12, wherein the disk performance optimization program is executed by the processor to obtain the cluster offset, and allocate cache entries based on the cluster offset, the cluster offset being a cluster Before the step of the difference between the initial size of the cluster and the actual size of the cluster, the following steps are also performed:

Determine whether the cluster is directly available;

If the clusters are directly available, calculate the number of directly available clusters.
11. The disk performance optimization device of claim 13, wherein when the disk performance optimization program is executed by the processor in the step of determining whether the cluster is directly usable, the following steps are further performed:

Obtaining a starting secondary index entry according to the target secondary index table, where the target secondary index table and the starting secondary index entry have a preset mapping relationship;

Judging whether the highest bit of the cluster is a preset value according to the initial secondary index item;

If the highest bit of the initial secondary index entry is a preset value, it is determined that the cluster is directly available.
The disk performance optimization device according to claim 13, wherein when the disk performance optimization program is executed by the processor in the step of calculating the number of directly available clusters, if the clusters are directly available, the following steps are further executed:

Calculate separately
and
The value of to determine the
The value of the
Whether the values of are equal, where P is the cluster offset, L is the preset length of the cluster, B is the actual size of the cluster, and N is the number of secondary index entries in the cluster;

like
And the value of
If the value of is not equal, the number of secondary index entries in the cluster is subtracted by 1 to obtain the number of directly available clusters.
A computer-readable storage medium, wherein a disk performance optimization program is stored on the computer-readable storage medium, and when the disk performance optimization program is executed by a processor, the steps of the disk performance optimization method described below are implemented:

Acquiring a cluster offset, and allocating a cache entry based on the cluster offset, where the cluster offset is the difference between the initial size of the cluster and the actual size of the cluster;

Set flushing parameters in the cache entry, the flushing parameters include dirty bit granularity and dirty bitmap size, where the dirty bitmap size is equal to the ratio of the actual size of the cluster to the dirty bit granularity, the The dirty bitmap size is used to indicate the total number of bits of the dirty bitmap storing dirty data, and the dirty bit granularity is used to indicate the amount of data represented by each bit;

Determine a target dirty bitmap according to the size of the dirty bitmap and a preset formula, and initialize the target dirty bitmap;

Determine whether the disk has received a request for flashing data;

If the disk receives the request for flushing data, the corresponding dirty data is identified through the target dirty bitmap, and the dirty data is flushed.
The computer-readable storage medium according to claim 16, wherein the disk performance optimization program is executed by the processor, and if the disk receives the request for flushing data, it passes the target dirty bitmap When identifying the corresponding dirty data and flashing the dirty data, the following steps are also performed:

If the disk receives a request for flushing data, it is determined whether any reference count item in the reference count block has been updated, wherein the reference count block is a module for recording reference count items in the disk, and the reference count The count item is used to record the bitmap with data in the buffer;

If any reference count item in the reference count block is updated, the corresponding dirty data is identified through the target dirty bitmap, and the dirty data is flushed.
16. The computer-readable storage medium according to claim 16, wherein the disk performance optimization program is executed by the processor to obtain the cluster offset, and allocate a cache entry based on the cluster offset, and then store it in the cache entry Before setting the flashing parameters, the flashing parameters include the dirty bit granularity and the dirty bitmap size, the following steps are also performed:

Judging whether the primary index entry is indexed to the historical secondary index table, the primary index entry has a preset mapping relationship with the historical secondary index table, and the primary index entry is used to index the historical secondary index table;

If the primary index entry is indexed into the historical secondary index table, the historical secondary index table is loaded into the cache entry.
The computer-readable storage medium of claim 18, wherein the disk performance optimization program is executed by the processor, the acquiring cluster offset, and the cache entry is allocated based on the cluster offset, and the cluster offset is Before the step of shifting to the difference between the initial size of the cluster and the actual size of the cluster, the following steps are also performed:

Index the target secondary index table through the primary index item;

Allocate clusters for the target secondary index table.
19. The computer-readable storage medium of claim 19, wherein the disk performance optimization program is executed by the processor to obtain the cluster offset, and allocate cache entries based on the cluster offset, the cluster offset being Before the step of the difference between the initial size of the cluster and the actual size of the cluster, the following steps are also performed:

Determine whether the cluster is directly available;

If the clusters are directly available, calculate the number of directly available clusters.