CN105808152B - A method of optimization RAID global metadata readwrite performance - Google Patents
A method of optimization RAID global metadata readwrite performance Download PDFInfo
- Publication number
- CN105808152B CN105808152B CN201410838457.XA CN201410838457A CN105808152B CN 105808152 B CN105808152 B CN 105808152B CN 201410838457 A CN201410838457 A CN 201410838457A CN 105808152 B CN105808152 B CN 105808152B
- Authority
- CN
- China
- Prior art keywords
- metadata
- memory
- global
- copy
- raid
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 31
- 238000005457 optimization Methods 0.000 title claims abstract description 17
- 230000006870 function Effects 0.000 claims description 18
- 230000004048 modification Effects 0.000 claims description 4
- 238000012986 modification Methods 0.000 claims description 4
- 230000008901 benefit Effects 0.000 abstract description 3
- 230000015556 catabolic process Effects 0.000 abstract description 3
- 230000009885 systemic effect Effects 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000009471 action Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000000151 deposition Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000004321 preservation Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Stored Programmes (AREA)
Abstract
The present invention relates to a kind of methods for optimizing RAID global metadata readwrite performance to be added memory metadata copy, be realized the optimization of global metadata readwrite performance based on the mode of in SBB system preceding four crucial disk management global metadatas；Specially：Initialize memory metadata copy；Global metadata is directly read from memory, optimizes reading performance；Global metadata is updated, new metadata information is written in key plate, and back up the metadata of the memory metadata copy.SBB system caused by this method avoid being damaged as key plate can not read and write metadata and systemic breakdown；It has given full play to from memory read data more than the advantage fast from disk reading speed, while the readwrite performance for improving global metadata, has greatly reduced burden when system operation.
Description
Technical field
The present invention is a kind of to be related to the optimization method of computer technology, and in particular to a kind of optimization RAID global metadata read-write
The method of performance.
Background technique
As high performance storage system, RAID of today (Redundant Array of Independent Disk,
Raid-array) it has been obtained and is more and more widely used.Wherein, for the overall situation of the software PLC of linux system
Metadata management is divided into two kinds, one is the super 1.x version metadata management of default, another kind be using ddf mode into
Row management.For the metadata management of super 1.x version, the layout of metadata is by the way of as shown in figure 3, each RAID
Member disk store respective metadata information in the position that it starts.
When all member disks of the RAID are all in off-line state, system can not perceive the presence of the RAID.To solve
The problem, the prior art is in the RAID module of SBB disk array, using the overall situation of the metadata based on super 1.x version
Way to manage.
As shown in figure 4, using the disk of first four slot position as the key plate of system, dedicated for depositing in SBB system
The metadata information of RAID in place system, even if the member disk for being RAID in system is all offline, system can also be perceived at the RAID
In off-line state.
However, the medium for using key plate to store as global metadata, there is also defects, are mainly manifested in following
Aspect：
1) in disc array system, due to the influence of the factors such as resonance, the damage of disk is not can avoid；When first four slot
The key plate of position is when all damaging, and SBB system will be unable to read-write metadata, the system of in turn resulting in can not normal use, to user with
Come unnecessary trouble and loss.
2) writing mode for using global metadata, must be sequentially written in four key plates, seriously affect metadata
Readwrite performance；
3) there are multiple RAID in systems, keep the operation of more new metadata more frequent, exacerbate when system is run
Burden.
Summary of the invention
In order to overcome the drawbacks of the prior art, the present invention proposes a kind of side for optimizing RAID global metadata readwrite performance
Method, this method avoid being damaged as key plate caused by SBB system can not read and write metadata and systemic breakdown；It gives full play to
From memory read data more than the advantage fast from disk reading speed, and delay more new metadata is to preceding four key plate, without
It is sequentially written in；While the readwrite performance for improving global metadata, burden when system operation is greatly reduced.
The purpose of the present invention is adopt the following technical solutions realization：
A method of optimization RAID global metadata readwrite performance, the method includes closing based in SBB system preceding four
The mode of Keyboard management global metadata adds memory metadata copy, realizes the optimization of global metadata readwrite performance；
Preferably, the method for the optimization RAID global metadata readwrite performance includes the following steps：
(1) memory metadata copy is initialized；
Wherein, the detailed process for initializing the memory metadata copy includes：Step 1-1, using reserved memory
Mode is allocated the memory metadata copy；The global metadata is read the memory metadata by step 1-2 in advance
In copy；
(2) global metadata directly is read from memory, optimizes reading performance；
(3) global metadata is updated, and new metadata information delay is updated into preceding four key plate, optimizes the property write
Energy；
(4) metadata of the memory metadata copy is backed up.
Preferably, the step 1-1 includes that after system crash, the kernel started using the scene crash finds institute
State the position of memory metadata copy and preservation.
Further, the step 1-1 further includes that the kernel executes setup_arch function, and calls the execution
E820_remove_range function in journey reserves the memory of global metadata total size.
Preferably, the step 1-2 includes during initializing memory metadata copy, passing through execution User space journey
Sequence mdadm (the User space program of management software PLC), and by the initial of the memory metadata copy by the way of netlink
Change information and is transmitted to the kernel；
After the completion of initialization, the metadata of preceding four key plate is identical, and the kernel is read in any key plate
Metadata is filled to the memory metadata copy.
Preferably, in the step (3), it is described update global metadata be will be described in the new metadata information write-in
In memory metadata copy, and the new metadata information is backuped to before the SBB system in four key plates；
When the events value in global metadata updates, then newest memory metadata is updated to the memory member number
According in the key plate of copy and the SBB system；Otherwise, only newest memory metadata is updated to memory metadata copy.
Further, a thread is created in the kernel, for periodically updating the overall situation of the memory copy
Metadata is into preceding four key plate.
Preferably, the step (4) includes under system reboot or power down condition, first passing through and calling register_
Reboot_notifier interface registers a notification interface, subsequently into kernel_restart_prepare function, calls
Blocking_notifier_call_chain function executes the call back function of registration；The call back function is used for the memory member
The metadata of data copy backups in SSD.
Preferably, under power blackout situation, the metadata of the memory metadata copy is backuped into SSD with the power supply of battery
In.
Preferably, in the case where system crash, start the memory scene of new kernel capture at that time, modification/etc/
Kdump_pre in kdump.conf configuration file executes customized script；The customized script is described for triggering
The memory metadata copy of backup.
Further, by mount order carry root file system, proc interface is set and triggers backup operation, makes to back up
Memory metadata copy be saved in the designated position in SDD.
Compared with the latest prior art, excellent effect of the invention is：
By archive memory metadata copy, first number can not be read and write by avoiding the SBB system as caused by key plate damage
Accordingly and systemic breakdown；It makes full use of from memory read data more than the advantage fast from disk reading speed, and need not successively incite somebody to action
Preceding four key plate is written in metadata, so that the readwrite performance of global metadata is improved, bearing when greatly reducing system operation
Load；Effective management to metadata in existing RAID array is realized, the user experience is improved to a certain extent.
Detailed description of the invention
The global metadata for addition memory metadata copy provided by the invention manages schematic diagram as shown in Figure 1.
The global metadata provided by the invention that pre-reads is illustrated in figure 2 to memory metadata copy flow chart.
The super1.x metadata management schematic diagram of background technique offer is provided.
The global metadata management schematic diagram of background technique offer is provided.
Specific embodiment
The present invention is described in further detail below in conjunction with the accompanying drawings.
The present invention provides a kind of method for optimizing RAID global metadata readwrite performance, and the method includes such as Fig. 1 institute
Show, based on the mode of in SBB system preceding four crucial disk management global metadatas, adds memory metadata copy；
Sysuuid is for indicating whether the disk belongs to the SBB system；
Summary is for the version information of identification metadata and the metadata information of the global HotSpare disk of storage；
Specific is used to store the super 1.x metadata information of RAID member disk.
As shown in Fig. 2, the method includes the following steps：
(1) initialization memory metadata copy includes the following steps：
Step 1-1 distributes the memory metadata copy with reserved memory mode；
After system crash, the kernel started using the scene crash finds the position of the memory metadata copy
And it saves.
The kernel executes setup_arch function, and calls the e820_remove_range function in the implementation procedure,
The memory of reserved global metadata total size.
Step 1-2 reads the global metadata in the memory metadata copy in advance；In initialization memory metadata
During copy, by executing User space program mdadm, and by the memory metadata copy by the way of netlink
Initialization information is transmitted to the kernel；
After the completion of initialization, the metadata of preceding four key plate is identical, and the kernel is read in any key plate
Metadata is filled to the memory metadata copy.
(2) global metadata directly is read from memory, optimizes reading performance；
(3) global metadata is updated, and new metadata information delay is updated into preceding four key plate, optimization is write
Performance；
In the step (3), the global metadata that updates is that the new metadata information is written to the memory member
In data copy, and the new metadata information is backuped to before the SBB system in four key plates；
When the events value in global metadata updates, then newest memory metadata is updated to the memory member number
According in the key plate of copy and the SBB system；Otherwise, only newest memory metadata is updated to memory metadata copy；
A thread is created in the kernel, for periodically updating the global metadata of the memory copy to institute
It states in preceding four key plate.
(4) metadata of the memory metadata copy is backed up.
The step (4) includes under system reboot or power down condition, first calling register_reboot_notifier
Interface registers a notification interface, subsequently into kernel_restart_prepare function, calls blocking_
Notifier_call_chain function executes the call back function of registration；The call back function is used for the memory metadata copy
Metadata backup in SSD.
Under power blackout situation, is powered with battery and backup to the metadata of the memory metadata copy in SSD.
In the case where system crash, start the memory scene of new kernel capture at that time, modification/etc/kdump.conf
Kdump_pre in configuration file executes customized script；The customized script is used to trigger the memory of the backup
Metadata copy.
By mount order carry root file system, proc interface is set and triggers backup operation, makes the memory member number of backup
The designated position in SDD is saved according to copy.
Finally it should be noted that：The above embodiments are merely illustrative of the technical scheme of the present invention and are not intended to be limiting thereof, to the greatest extent
Invention is explained in detail referring to above-described embodiment for pipe, and those of ordinary skill in the art still can be to this hair
Bright specific embodiment is modified or replaced equivalently, and these without departing from spirit and scope of the invention any modification or
Person's equivalent replacement, within the scope of the claims of the invention pending application.
Claims (9)
1. a kind of method for optimizing RAID global metadata readwrite performance, which is characterized in that the method includes being based on SBB system
The mode of preceding four crucial disk management global metadatas, adds memory metadata copy in system；
The method includes the following steps：
(1) initialization memory metadata copy includes the following steps：
Step 1-1 distributes the memory metadata copy with reserved memory mode；
Step 1-2 reads the global metadata in the memory metadata copy in advance；
(2) global metadata directly is read from memory, optimizes reading performance；
(3) global metadata is updated, and new metadata information delay is updated into preceding four key plate, optimizes the property write
Energy；
(4) metadata of the memory metadata copy is backed up.
2. the method for optimization RAID global metadata readwrite performance as described in claim 1, which is characterized in that the step 1-
1 includes, and after system crash, the kernel started using the scene crash finds the position of the memory metadata copy simultaneously
It saves.
3. the method for optimization RAID global metadata readwrite performance as claimed in claim 2, which is characterized in that the step 1-
1 includes, and the kernel executes setup_arch function, and calls the e820_remove_range function in the implementation procedure, in advance
Stay the memory of global metadata total size.
4. the method for optimization RAID global metadata readwrite performance as described in claim 1, which is characterized in that the step 1-
2 include, during initializing memory metadata copy, by executing User space program mdadm, and using the side of netlink
The initialization information of the memory metadata copy is transmitted to kernel by formula；
After the completion of initialization, the metadata of preceding four key plate is identical, and the kernel reads first number in any key plate
According to filling to the memory metadata copy.
5. the method for optimization RAID global metadata readwrite performance as described in claim 1, which is characterized in that the update is global
Metadata is one thread of creation in kernel, the global metadata of memory metadata copy is periodically updated, by new member
Data information is written in the memory metadata copy, and before the new metadata information is backuped to the SBB system
In four key plates；
When the event object in global metadata updates, then newest memory metadata is updated to the memory metadata pair
In the key plate of this and the SBB system；Otherwise, only newest memory metadata is updated to memory metadata copy.
6. the method for optimization RAID global metadata readwrite performance as described in claim 1, which is characterized in that described in the backup
The metadata of memory metadata copy includes under system reboot or power down condition, first calling register_reboot_
Notifier interface registers a notification interface, subsequently into kernel_restart_prepare function, calls blocking_
Notifier_call_chain function executes the call back function of registration；The call back function is used for the memory metadata copy
Metadata backup in SSD.
7. the method for optimization RAID global metadata readwrite performance as described in claim 1, which is characterized in that in power blackout situation
Under, it is powered with battery and backups to the metadata of the memory metadata copy in SSD.
8. the method for optimization RAID global metadata readwrite performance as described in claim 1, which is characterized in that as system crash
In the case where, start the memory scene of new kernel capture at that time, the kdump_ in modification/etc/kdump.conf configuration file
Pre executes customized script；The customized script is used to trigger the memory metadata copy of the backup.
9. the method for optimization RAID global metadata readwrite performance as claimed in claim 8, which is characterized in that ordered by mount
Carry root file system is enabled, setting proc interface triggers backup operation, is saved in the memory metadata copy of backup in SSD
Designated position.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410838457.XA CN105808152B (en) | 2014-12-29 | 2014-12-29 | A method of optimization RAID global metadata readwrite performance |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410838457.XA CN105808152B (en) | 2014-12-29 | 2014-12-29 | A method of optimization RAID global metadata readwrite performance |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105808152A CN105808152A (en) | 2016-07-27 |
| CN105808152B true CN105808152B (en) | 2018-11-27 |
Family
ID=56980859
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410838457.XA Active CN105808152B (en) | 2014-12-29 | 2014-12-29 | A method of optimization RAID global metadata readwrite performance |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105808152B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101196832A (en) * | 2007-12-29 | 2008-06-11 | 北京航空航天大学 | A Resource Broker for Lighting Heterogeneous Devices |
| CN102368268A (en) * | 2011-10-25 | 2012-03-07 | 无锡城市云计算中心有限公司 | Method for achieving consistency of pieces of metadata |
| CN102567445A (en) * | 2011-10-25 | 2012-07-11 | 无锡城市云计算中心有限公司 | Method for guaranteeing consistency of metadata in distributed file system |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8150801B2 (en) * | 2008-08-20 | 2012-04-03 | Microsoft Corporation | Recovery of a computer that includes virtual disks |
-
2014
- 2014-12-29 CN CN201410838457.XA patent/CN105808152B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101196832A (en) * | 2007-12-29 | 2008-06-11 | 北京航空航天大学 | A Resource Broker for Lighting Heterogeneous Devices |
| CN102368268A (en) * | 2011-10-25 | 2012-03-07 | 无锡城市云计算中心有限公司 | Method for achieving consistency of pieces of metadata |
| CN102567445A (en) * | 2011-10-25 | 2012-07-11 | 无锡城市云计算中心有限公司 | Method for guaranteeing consistency of metadata in distributed file system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105808152A (en) | 2016-07-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8838918B2 (en) | Information processing apparatus and data backup method | |
| US9921758B2 (en) | Avoiding long access latencies in redundant storage systems | |
| TWI556104B (en) | Techniques to perform power fail-safe caching without atomic metadata | |
| US8275930B2 (en) | Storage system having volatile memory and non-volatile memory | |
| CN106354582B (en) | A kind of continuous data protection method | |
| US20120311261A1 (en) | Storage system and storage control method | |
| CN103970481A (en) | Method and device for reconstructing memory array | |
| CN112632643B (en) | Method for preventing flash memory data from being lost, solid state drive controller, solid state drive | |
| US9696922B2 (en) | Storage system | |
| CN104050056A (en) | File system backup of multi-storage-medium device | |
| CN102999564B (en) | Write the method for data, device and equipment | |
| CN103150128A (en) | Implementation method of solid state drive (SSD) and disk-based reliable mixed storage system | |
| US20140351523A1 (en) | System and Method of Rebuilding READ Cache for a Rebooted Node of a Multiple-Node Storage Cluster | |
| JP7318367B2 (en) | Storage control device and storage control program | |
| US9772782B2 (en) | Non-volatile complement data cache | |
| WO2021088367A1 (en) | Data recovery method and related device | |
| WO2014075586A1 (en) | Method and device for automatically recovering storage of jbod array | |
| CN103019623A (en) | Memory disc processing method and device | |
| US20060112221A1 (en) | Method and Related Apparatus for Data Migration Utilizing Disk Arrays | |
| CN108874312B (en) | Data storage method and storage device | |
| US20110119509A1 (en) | Storage system having power saving function | |
| CN102609224B (en) | Redundant array of independent disk system and initializing method thereof | |
| US7702864B2 (en) | Apparatus, system, and method for writing stripes in parallel to unique persistent storage devices | |
| CN110531925B (en) | Data reading method and device, electronic equipment and machine-readable storage medium | |
| CN105808152B (en) | A method of optimization RAID global metadata readwrite performance |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: 100193 Beijing, Haidian District, northeast Wang West Road, building 8, building 36, floor 5 Applicant after: Shuguang Cloud Computing Group Co.,Ltd. Address before: 100193 Beijing, Haidian District, northeast Wang West Road, building 8, building 36, floor 5 Applicant before: DAWNING CLOUD COMPUTING TECHNOLOGY Co.,Ltd. |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20210521 Address after: 100089 building 36, courtyard 8, Dongbeiwang West Road, Haidian District, Beijing Patentee after: Dawning Information Industry (Beijing) Co.,Ltd. Address before: 100193 5 floor, 36 building, No. 8 Northeast Road, Haidian District, Beijing. Patentee before: Shuguang Cloud Computing Group Co.,Ltd. |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20220728 Address after: 100089 building 36, courtyard 8, Dongbeiwang West Road, Haidian District, Beijing Patentee after: Dawning Information Industry (Beijing) Co.,Ltd. Patentee after: DAWNING INFORMATION INDUSTRY Co.,Ltd. Address before: 100089 building 36, courtyard 8, Dongbeiwang West Road, Haidian District, Beijing Patentee before: Dawning Information Industry (Beijing) Co.,Ltd. |