CN1304957C - Method for improving compute system disc synchronous writing property based on movable memory - Google Patents

Method for improving compute system disc synchronous writing property based on movable memory Download PDF

Info

Publication number
CN1304957C
CN1304957C CNB2005100111255A CN200510011125A CN1304957C CN 1304957 C CN1304957 C CN 1304957C CN B2005100111255 A CNB2005100111255 A CN B2005100111255A CN 200510011125 A CN200510011125 A CN 200510011125A CN 1304957 C CN1304957 C CN 1304957C
Authority
CN
China
Prior art keywords
data
file
storage device
writing
disk
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.)
Expired - Fee Related
Application number
CNB2005100111255A
Other languages
Chinese (zh)
Other versions
CN1648878A (en
Inventor
张悠慧
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tsinghua University
Original Assignee
Tsinghua University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Tsinghua University filed Critical Tsinghua University
Priority to CNB2005100111255A priority Critical patent/CN1304957C/en
Publication of CN1648878A publication Critical patent/CN1648878A/en
Application granted granted Critical
Publication of CN1304957C publication Critical patent/CN1304957C/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The present invention relates to a method for raising the magnetic disk synchronous writing performance of a computer system based on mobile storage, which belongs to the technical field of the storage system structure of computers. More specifically, the present invention relates to a method for raising the magnetic disk synchronous writing performance of data processing efficiency by using a system call transparent packing mechanism to use common computer mobile storage equipment (such as U disks, mobile hard disk, etc.) as synchronous write buffers of the computer system, and the method is MSYNC for short. The present invention comprises a computer and mobile storage equipment, and the computer is connected with the mobile storage equipment through a high-speed high-bandwidth interface (USB interface or IEEE-1394 interface); the computer maps the mobile storage equipment into a local hard disk. The file writing operation of applications are packed by using a system call transparent packing technique so that synchronous magnetic disk is converted into high-speed synchronous writing to the mobile storage equipment. Thus, under the precondition that data reliability is not lowered, the magnetic disk synchronous writing performance of the computers is enhanced.

Description

Compute system disc synchronous writing property raising method based on mobile storage
Technical field
Compute system disc synchronous writing property raising method (being called for short MSYNC) based on mobile storage belongs to computer memory system structure field.
Background technology
The main performance bottleneck of computer nowadays system is the I/O link.This mainly be since the raising of I/O system performance well below the raising of CPU and internal memory performance.According to the Moore law, cpu performance and memory size just can be doubled in per 18 months.But the major equipment as external memory---hard disc, because the essential characteristic of mechanical motion causes the raising of performance very limited, annual about 7%.Can be drawn by the Amadl law like this, nowadays I/O occupies critical role in the entire system performance.In fact really the principal element that limits disk performance is that disk access postpones, (access time of hard disk is mainly by data locking can only to increase about 10% every year, tracking and rotational latency, reading and writing data are formed, and wherein tracking refers to magnetic head with rotation and moves to the relevant position of hard disk so that carry out data access.Because the two is mechanically actuated, its performance and data locking and accessing operation differ two more than the order of magnitude, and this has just hindered the raising of disk access performance.)。
From the software angle, can adopt multiple optimisation technique to reduce the influence of I/O visit, such as caching technology to system performance.By various buffer scheduling algorithms, caching technology can reduce the number of times to disk reading and writing operations---application program to the write operation of disk in fact being converted the write operation of buffer memory and is in time write disk with these data by operating system according to system running state.Disk write operation repeatedly can be simplified to single job like this, thereby reduce magnetic disc access times, improve overall performance.This write operation is commonly referred to as asynchronous write.
But it is (different with asynchronous write to carry out synchronous write, synchronous write is not used buffer memory, but just calculate complete operation after directly data being write the relevant position of hard disk, so performance is very low, normally one of percentage of asynchronous write or per mille) time these methods inoperative.In the universal document system, write operation is in the great majority, and is about 57%, and the synchronous operation ratio accounts for 50% to 75%, and the visit to metadata accounts for 67% to 78% in the write operation.The synchronous writing operation that can draw I/O from these data is the main cause of performance bottleneck.Why like this, mainly be because file system must be finished write operation to metadata and responsible consumer data by the method for synchronization for the integrality that guarantees data.For Database Systems, synchronous operation has critical influence for the performance of issued transaction in addition.
From hardware point of view, adopt non-volatile NVRAM to replace traditional internal memory can improve computer memory system under the prerequisite that does not reduce system reliability performance, but this mechanism need special hardware (comprising mainboard etc.).Therefore for the PC of the cheapness of widely using at present, this method cost is too high, is difficult to practicality.
Therefore crucial problem is how to utilize general and cheap hardware, improves the synchronous write performance of disk of computer system comparatively easily, should have higher portability (promptly realize principle do not rely on certain or certain type operating system) simultaneously.
At present, the use of movable storage device (as portable hard drive, flash disk) has more and more widely almost become the standard configuration of PC; Simultaneously increasing digital equipment (as mp3 player, digital camera etc.) uses the storer based on Flash Memory (flash memory) to come and the PC swap data.Flash Memory is a development in recent years internal memory rapidly, belong to non-voltile memory (Non-volatile promptly cut off the power supply data also can preserve), it has the advantages that electricity is wiped, advantage such as also have low-power consumption, density height, volume is little, reliability is high, can wipe, can rewrite (for details, see the appendix 2).Like this, the performance that writes the operation of data in the movable storage device synchronously can be far above to the writing of PC hard disk synchronously, and main cause is tracking and the rotational delay time that the former need not disk.Simultaneously, support the development (comprising usb protocol, IEEE-1394 etc.) of the technology of high bandwidth peripheral hardware to make that the data transmission between this type of mobile storage and main frame is more and more quicker, make that the former performance advantage is embodied.
Based on above-mentioned analysis, the utilization that the present invention proposes movable storage device at a high speed is as the reliable disk buffering of some I/O intensive applications.
Summary of the invention
The invention reside in provides a kind of with the high performance synchronous write method of mobile storage as the computer disk buffer memory, so that make full use of the performance (USB of present computer system high bandwidth peripheral hardware, interfaces such as IEEE-1394), to improve disk synchronous write speed, solve the problem that exists in the current computer storage organization level.MSYNC is an I/O intensive applications program---handles service as WebServer and db transaction, a high performance storage means is provided.
The core content of MSYNC technology is---utilize the high-speed mobile memory device (linking to each other) that is connected on the computing machine as some uses the disk buffering of (often needing to carry out more timing disk writes) on the main frame with interfaces such as USB, IEEE-1394, the feasible data that need write synchronously are stored on the mobile device with higher speed reliably, and on main frame, just carry out asynchronous write fast, under the prerequisite that does not reduce data reliability, improve the performance of these application with this.Concrete realization key is that the file that writes synchronously that the present invention has intercepted and captured operating system pellucidly calls, and it is directed on movable storage device at a high speed.
Its main innovative point is as follows:
1. utilize movable storage device as disk buffering, it links to each other (comprising usb interface, IEEE-1394 interface) with the high bandwidth Peripheral Interface with intercomputer.
2. adopt the transparent technique for packing of system call to make that the realization of this method is transparent fully for operating system and destination application, promptly application program need not change.
3. mobile device writes the Journaled management of data efficiently, makes can support large-scale IO intensive applications than the movable storage device of low capacity, comprises that large database uses.
System of the present invention is characterized in that:
It is the high bandwidth Peripheral Interface (as USB, IEEE-1394 etc.) by computing machine, utilizes movable storage device as the system of computer disk buffer memory with the raising data-handling efficiency, and it contains:
The computing machine that the high bandwidth Peripheral Interface is arranged: promptly desire improves the computing machine of disk synchronous write performance, operational objective application program.The high bandwidth Peripheral Interface comprises USB interface (1.1 and 2.0 standard), IEEE-1394 interface etc.
Movable storage device: as USB flash disk, portable hard drive etc., as the disk synchronous write buffer memory of computing machine.Current have many high-speed mobile memory devices, distinguishes from storage medium to comprise portable hard drive and multiple storer based on Flash Memory, and the latter comprises flash disk (flash disk), the CompactFlash card, the SmartMedia card, Memory Stick, MultiMedia Card card etc.Wherein USB flash disk has progressively replaced the standard configuration that floppy disk is called PC.
Both link to each other with the high bandwidth Peripheral Interface.At present, the memory device of supporting the high bandwidth Peripheral Interface comprises Windows series, Linux, MAC OS, Solaris etc. by the operating system support of most of main flows.Generally speaking, be connected to these equipment on the computing machine after, operating system utilize its driver with these equipment as independent partition holding---as Windows 2000 times, be used as a new disk; And under Linux, then can be used as one independently disk be mapped to a certain catalogue, then to its operation with as broad as long to the hard disk operation of this machine.Therefore, the installation of movable storage device is solved by operating system with using, this illustrates that also the MSYNC technology has bigger portability and ease for use.
The method of the invention is characterised in that:
It utilizes movable storage device as the computer disk buffer memory, and it is local disk with this device map that the computing machine of high bandwidth Peripheral Interface is arranged.(essence is exactly to insert code between system call that the user uses and the actual function that realizes by wrapping up for system call then, so that intercept and capture the information such as call parameters of user program), the associative disk that can directly intercept and capture application program like this writes and calls, then disk synchronous write wherein is converted to asynchronous write, and the data that need write synchronously write mobile storage with reliable fashion.Because the latter's writing speed is fast more a lot of than the former, so there is higher performance to improve leeway.
It contains successively and has the following steps:
The 1st step: computer system is mapped as a system disk to the outside movable storage device of the high bandwidth that is connected to oneself, with its buffer memory of writing as system;
The 2nd step: carry out initialization with movable storage device as the computer system of writing buffer memory to step 1 is resulting:
The parcel program that use is preset is wrapped up the file operation in the target program, promptly inserts code in the target program execute file, to rewrite the inlet of file operation in execute file that will wrap up;
The 3rd step: operational objective program;
The 4th step: when the computer system of having used MSYNC intercepts the file operation of destination application, carry out following steps:
The 4.1st step: for writing file operation? if then the insertion code in the above-mentioned steps 2 writes movable storage device with data with the form of asynchronous write; Simultaneously data are write host hard drive in the asynchronous write mode; Changeed for the 4th step then, continue.
The 4.2nd step: be that file data is brushed into disk operating (being synchronous writing operation)? if then the insertion code in the above-mentioned steps 2 brushes the data that write on the movable storage device into this movable storage device; Changeed for the 4th step then, continue; The data management policies of MSYNC usage log formula writes the data of movable storage device according to following management by methods:
1, after the file that step 4 intercepts destination application writes data, write data in the backup file default on the movable storage device, writing format each time is: size of data, the reference position of data in the computer disk file, data volume;
2, if after writing data, when the backup file size surpassed certain predetermined threshold value, then the insertion code in the above-mentioned steps 2 just forced the respective file on the host hard drive to carry out the synchronous operation of brushing, and empties this backup file after the success; The 4.3rd step: if the alternative document operation is then directly transferred to the file operation that is wrapped and handled; Changeed for the 4th step then, continue.
The Journaled management method that described movable storage device writes data is as follows:
1.MSYNC initialization: create one or more backup files on movable storage device, what each was used to store related application writes data synchronously.
2. after file that above-mentioned step 4 intercepts application program writes data, data are write backup file, writing format each time is: size of data, the reference position of data in the computer disk file, data volume.
3. if when writing digital backup part file size and surpassing certain predetermined threshold value, MSYNC just forces the respective file on the host hard drive to carry out synchronously brush to go into operation, empty this backup file after the success.
The main reason of so doing is: generally speaking, the capacity of movable storage device is much smaller than host disk, and the file size that the needs that application program is used write synchronously may be greater than the capacity of mobile storage, and therefore having designed a Journaled data management policies solves this contradiction.In case in fact this mechanism be the same with the disk buffering of operating system---data cached brush is gone into hard disk, and these buffer memorys can be released and reuse.
The present invention realizes that in the Windows system its disk synchronous write performance has improved 1.5-2.5 doubly than the write performance that does not use this invention.And, improved 1.5-1.9 doubly at the performance of its transaction operation (being that data are to write disk synchronously) for the MySQL database.
Description of drawings:
Fig. 1, the structural drawing of system of the present invention, i.e. MSYNC system model.
Fig. 2, the FB(flow block) of system of the present invention.
Fig. 3, Win32 execute file form.
Fig. 4, the execute file form behind the function parcel.
Embodiment:
The present invention is made up of computing machine, a movable storage device that has high bandwidth Peripheral Interface (comprising USB, IEEE-1394).Utilize movable storage device as the computer disk buffer memory, it is local disk with this device map that the computing machine of high bandwidth Peripheral Interface is arranged.By carrying out transparent parcel for system call, the associative disk that can directly intercept and capture application program like this writes and calls then, wherein disk synchronous write is converted to asynchronous write, and the data that need write synchronously writes mobile storage with reliable fashion.Because the latter's writing speed is fast more a lot of than the former, so there is higher performance to improve leeway.
MSYNC provides service with a kind of to the complete transparent form of user.As long as the application program that user's transparent technique for packing of system call (being a software program in form) of utilizing MSYNC to provide improves write performance to desire is carried out single treatment (see for details " the transparent technique for packing of system call " joint), and specifies movable storage device Map Disk or the catalogue on computers that will use just passable.So just can guarantee that application program of user does not add any change.
1. the installation of high-speed mobile memory device and use
The memory device of supporting high bandwidth interface comprises Windows series, Linux, MAC OS, Solaris etc. by the operating system support of most of main flows at present.Generally speaking, be connected to these equipment on the computing machine after, operating system utilize its driver with these equipment as independent partition holding---as Windows 2000 times, be used as a new disk; And under Linux, then can be used as one independently disk be mapped to a certain catalogue, then to its operation with as broad as long to the hard disk operation of this machine.Therefore, the installation of movable storage device is solved by operating system with using, this illustrates that also the MSYNC technology has bigger portability and ease for use.
2. the transparent technique for packing of system call
The essence of system call technique for packing is exactly to insert code between system call that the user uses and the actual function that realizes, so that intercept and capture the information such as call parameters of user program.
Generally speaking, this technology is to call by name or realize in the storehouse that rewrites the function place by function; But MSYNC has adopted transparent code insertion technology, and dual mode is arranged:
The one, in the execute file of application program, insert code, realize by rewriteeing the system call inlet hereof that will wrap up.In case application program uses this to call like this, will be introduced into the code that MSYNC inserts, undertaken calling real system function again after the suitable processing or directly returning by the latter.
The 2nd, application program is loaded into carries out the function parcel behind the internal memory again, directly in internal memory, rewrite the image of related function, can not change the size of execute file like this, and effect is the same.
This technology can need not to revise source code the user, need not to recompilate to connect, and reaches and realizes required function under the prerequisite of not replacing any dynamic link libraries.What use in our reference realizes at present is first kind of mode.
3.Windows the reference under the system realizes
MSYNC has finished a realization in Windows 2000/XP system, its concrete system is composed as follows:
Windows?2000?Professional(sp4);
USB 2.0 portable hard drives;
The MSYNC program of version of window is used for the transparent parcel of system call;
The destination application that is used for performance test comprises MySQL database application and the synchronous write-in program of the hard disk of writing specially.
1) the transparent parcel of system call
Fig. 3 has shown the composition of execute file under the Windows, comprise file header, the PE head, text segment comprises program code, data segment comprises the input chained list of initialization data, used dynamic link libraries and system call, output (Exported) function table and debug symbol.
For revising the Windows run time version, MSYNC has created a new section (detours section) between debug symbol and output function table---and contain the code that the copy of original PE head, new function input table and MSYNC insert, revised original PE head simultaneously and make it to point to new input table.As Fig. 4.
Insert new input table two purposes are arranged: at first it has kept original input table so that the execute file that will revise is changeed back when needed.Secondly, new input table can comprise new dynamic link libraries, in order to finish required additional functionality.When destination application called the file write operation, in fact inquiry was the new function input table that MSYNC inserts, thereby has called the insertion code of MSYNC like this, and the latter comprises original call function.
2) code description of Cha Ruing
MSYNC has wrapped up four systems and has called under Windows---CreateFile, and CloseHandle, WriteFile, FlushFileBuffer has realized relevant insertion code simultaneously.The overall operation flow process is as follows:
The Step1:MSYNC initialization.Comprise the initialization backup listed files, and obtain the title (this needs artificial the appointment) of the fixed disk file that will back up.
Step2: intercept and capture related system and call, make the following judgment---
If the CreateFile operation is then called original createFile earlier and is created file.After enter the insertion code, judge whether file for backing up, if then on movable storage device, create backup file, and this information inserted backup file tabulation.Return.
If the CloseHandle operation judges then whether the file of closing backs up? if then this close file content is brushed into this machine hard disk, close the backup file on the mobile storage after the success, and delete it.Call original CloseHandle and return.
If WriteFile then at first calls original WritFile.Judge whether it is to write the file that has backed up then? if then the data that write are write backup file with asynchronous system (WriteFIle).Return.
If FlushFileBuffer (the synchronous brush under the Windows is gone into function) then at first judges whether it is that brush is gone into backup file? if then call original FlushFileBuffer, directly return at the backup file on the mobile storage.
Function behind the parcel is as follows:
New_CreateFile (LPCSTR a0, DWORD a1, DWORD a2, LPSECURITY_ATTRIBUTES a3, DWORD a4, DWORD a5, HANDLE a6) { HANDLE rv=0; _ try{ rv=Real_CreateFile (a0, a1, a2, a3, a4, a5, a6); // call real CreateFile if (rv!=INVALID_HANDLE_VALUE) // if operate successfully DuplicateFile (rv, (LPCSTR) szName, 0,0,0,0,0,0); // call the operation that MSYNC inserts } _ finally{ }; Return rv; New_CloseHandle (HANDLE a0) { BOOL rv=0; _ try{ CloseDuplicateFile (a0); Rv=Real_CloseHandle (a0);<!--SIPO<DP n=" 7 "〉--〉<dp n=" d7 "/_ finally{ }; Return rv; New_FlushFileBuffers (HANDLE a0) { BOOL rv=1; _ try{ if (! FlushDuplicateFileBuffers (a0)) rv=Real_FlushFileBuffers (a0); } _ finally{ }; Return rv; New_WriteFile (HANDLE a0, LPCVOID a1, DWORD a2, LPDWORD a3, LPOVERLAPPED a4) { BOOL rv=0; _ try{ rv=Real_WriteFile (a0, a1, a2, a3, a4); If (rv) WriteDuplicateFile (a0, a1, a2, a3, a4, a4!=NULL, rv); } _ finally{ }; Return rv;
3) performance evaluating
Realize at the reference under the Windows 2000, carried out the performance test of MSYNC.Concrete computer configuration is as follows:
Windows?2000Professional(sp4);
CPU:PIII?800Mhz
Internal memory: 128 internal memories
USB drives: the USB Mass Storage Device of Microsoft, 5.0.2195.1
Host hard drive: IBM-DJSA-220
USB 2.0 portable hard drives;
Concrete tested object application program is divided into two classes: disk synchronous write file routine of writing specially and MySQL database.For every kind of test, compared its write direct this machine hard disk and performance difference that uses after MSYNC technology.
Test 1
The disk synchronous write program of oneself writing has been adopted in this test, writes different big or small data synchronously 1024 times on hard disk at every turn, just enables MSYNC and whether carries out the comparison of deadline, and is as follows:
4KB*1024 2KB*1024 1KB*1024 512B*1024
Not enabled MSYNC (unit second) 39 37 37 40
Enable MSYNC (unit second) 25 18 16 16
The performance ratio 1.56 2.05 2.31 2.5
Test 2
The MySQL database under the Windwos has been adopted in this test, inserts 1024 different big or small data (adopt transaction mode, promptly data are to write disk synchronously) continuously in a table.Just enable MSYNC and whether carry out the comparison of deadline, as follows:
1KB*1024 512B*1024 64B*1024
Not enabled MSYNC (unit second) 155 143 110
Enable MSYNC (unit second) 83 75 72
The performance ratio 1.87 1.91 1.53

Claims (2)

1. based on the compute system disc synchronous writing property raising method of mobile storage, be called for short MSYNC, it is characterized in that it contains following steps successively:
The 1st step: computer system is mapped as a system disk to the outside movable storage device of the high bandwidth that is connected to oneself, with its buffer memory of writing as system;
The 2nd step: carry out initialization with movable storage device as the computer system of writing buffer memory to step 1 is resulting:
The parcel program that use is preset is wrapped up the file operation in the destination application, promptly inserts code in the destination application execute file, to rewrite the inlet of file operation in execute file that will wrap up;
The 3rd step: operational objective application program;
The 4th step: when the computer system of having used MSYNC intercepts the file operation of destination application, carry out following steps:
The 4.1st step: for writing file operation? if then the insertion code in the above-mentioned steps 2 writes movable storage device with data with the form of asynchronous write; Simultaneously data are write host hard drive in the asynchronous write mode; Changeed for the 4th step then, continue;
The 4.2nd step: whether be that file data is brushed into disk operating, i.e. synchronous writing operation? if then the insertion code in the above-mentioned steps 2 brushes the data that write on the movable storage device into this movable storage device, changeed for the 4th step then, continue; MSYNC usage log formula data management policies, management writes the data of movable storage device in such a way:
1, after the file that step 4 intercepts destination application writes data, write data in the backup file default on the movable storage device, writing format each time is: size of data, data reference position, the data volume in the computer disk file;
2, if after writing data, when the backup file size surpassed certain predetermined threshold value, then the insertion code in the above-mentioned steps 2 just forced the respective file on the host hard drive to carry out the synchronous operation of brushing, and empties this backup file after the success;
The 4.3rd step: if the alternative document operation is then directly transferred to the file operation that is wrapped and handled; Changeed for the 4th step then, continue.
CNB2005100111255A 2005-01-07 2005-01-07 Method for improving compute system disc synchronous writing property based on movable memory Expired - Fee Related CN1304957C (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CNB2005100111255A CN1304957C (en) 2005-01-07 2005-01-07 Method for improving compute system disc synchronous writing property based on movable memory

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CNB2005100111255A CN1304957C (en) 2005-01-07 2005-01-07 Method for improving compute system disc synchronous writing property based on movable memory

Publications (2)

Publication Number Publication Date
CN1648878A CN1648878A (en) 2005-08-03
CN1304957C true CN1304957C (en) 2007-03-14

Family

ID=34875467

Family Applications (1)

Application Number Title Priority Date Filing Date
CNB2005100111255A Expired - Fee Related CN1304957C (en) 2005-01-07 2005-01-07 Method for improving compute system disc synchronous writing property based on movable memory

Country Status (1)

Country Link
CN (1) CN1304957C (en)

Families Citing this family (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101751307A (en) * 2008-12-12 2010-06-23 比兹肯解决方法有限公司 Data storage system, method as well as data storage and backup system
CN102193842A (en) * 2010-03-15 2011-09-21 成都市华为赛门铁克科技有限公司 Data backup method and device
CN102203753B (en) * 2011-05-23 2014-04-16 华为技术有限公司 Data writing processing method, apparatus and terminal equipment
CN102541757B (en) * 2011-11-30 2014-11-05 华为技术有限公司 Write cache method, cache synchronization method and device
CN103500075A (en) * 2013-10-11 2014-01-08 张维加 Externally-connected computer accelerating device based on new materials
CN106469119B (en) * 2015-08-10 2020-07-07 北京忆恒创源科技有限公司 Data writing caching method and device based on NVDIMM
CN105511806B (en) 2015-11-30 2018-09-07 华为技术有限公司 The method and mobile terminal of processing write requests
CN108200217A (en) * 2018-03-05 2018-06-22 吉林化工学院 The synchronous method and system of a kind of data
CN108829835A (en) * 2018-06-19 2018-11-16 深圳市瑞驰信息技术有限公司 A kind of novel high speed cache file system architecture

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1476562A (en) * 2000-09-26 2004-02-18 ض� Posted write-through cache for flash memory
CN2603974Y (en) * 2002-12-27 2004-02-18 孙斌 Magnetic disc cache apparatus
US20040093454A1 (en) * 2002-11-07 2004-05-13 Teng Peter Chu Tin USB endpoint controller flexible memory management

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1476562A (en) * 2000-09-26 2004-02-18 ض� Posted write-through cache for flash memory
US20040093454A1 (en) * 2002-11-07 2004-05-13 Teng Peter Chu Tin USB endpoint controller flexible memory management
CN2603974Y (en) * 2002-12-27 2004-02-18 孙斌 Magnetic disc cache apparatus

Also Published As

Publication number Publication date
CN1648878A (en) 2005-08-03

Similar Documents

Publication Publication Date Title
CN1304957C (en) Method for improving compute system disc synchronous writing property based on movable memory
US20240303222A1 (en) Heuristic interface for enabling a computer device to utilize data property-based data placement inside a nonvolatile memory device
JP6910131B2 (en) Storage device including multiple partitions for multiple mode operation, and its operation method
KR101969883B1 (en) Data storage device and operating method thereof
US7433994B2 (en) On-device data compression to increase speed and capacity of flash memory-based mass storage devices
KR102585883B1 (en) Operating method of memory system and memory system
US20090164700A1 (en) efficient memory hierarchy in solid state drive design
CN1945537A (en) Method for realizing high speed solid storage device based on storage region network
Eshghi et al. Ssd architecture and pci express interface
JP5068754B2 (en) Improved host interface
CN1955939A (en) Backup and recovery method based on virtual flash disk
Choudhuri et al. Performance improvement of block based NAND flash translation layer
CN111796759A (en) Computer readable storage medium and method for fragment data reading on multiple planes
Park et al. New techniques for real-time FAT file system in mobile multimedia devices
CN116578234B (en) Flash memory access system and method
CN116185892B (en) Flash memory controller and method for accessing flash memory particles
TWI414940B (en) Block management and data writing method, flash memory storage system and controller
Lo et al. Swap-before-hibernate: a time efficient method to suspend an OS to a flash drive
Wu et al. CAGC: A content-aware garbage collection scheme for ultra-low latency flash-based SSDs
CN110096452B (en) Nonvolatile random access memory and method for providing the same
Ou et al. Gemini: a novel hardware and software implementation of high-performance PCIe SSD
Micheloni et al. Solid state drives (ssds)
Choudhuri et al. Real-time access guarantees for nand flash using partial block cleaning
Son et al. Optimizing garbage collection overhead of host-level flash translation layer for journaling filesystems
Micheloni et al. SSD architecture and PCI express interface

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20070314

Termination date: 20200107

CF01 Termination of patent right due to non-payment of annual fee