CN103257935B - Of the cache management method and its application - Google Patents

Of the cache management method and its application Download PDF

Info

Publication number
CN103257935B
CN103257935B CN 201310138419 CN201310138419A CN103257935B CN 103257935 B CN103257935 B CN 103257935B CN 201310138419 CN201310138419 CN 201310138419 CN 201310138419 A CN201310138419 A CN 201310138419A CN 103257935 B CN103257935 B CN 103257935B
Authority
CN
Grant status
Grant
Patent type
Prior art keywords
cache
queue
data
cr
buffer
Prior art date
Application number
CN 201310138419
Other languages
Chinese (zh)
Other versions
CN103257935A (en )
Inventor
陈俭喜
刘景宁
冯丹
黄赛
王璞
Original Assignee
华中科技大学
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
Grant date

Links

Abstract

本发明公开了一种缓存管理方法,包括:若缓存中有空白缓存空间,则请求数据写入空白缓存空间;若缓存已被写满,同时请求数据块号未记录在缓存筛选队列中,则将请求数据块号写入缓存筛选队列的尾部;若缓存已被写满,同时请求数据块号已被记录在缓存筛选队列当中,将此数据块号从缓存筛选队列中删除,并将此请求数据写入缓存;其中,缓存筛选队列为LRU队列,用于记录最近被访问却未命中缓存的数据的磁盘块号。 The present invention discloses a buffer management method, comprising: if the blank space in the cache buffer, the buffer empty space data write request; if the buffer has been filled, while the requested data block number is not recorded in the cache screening queue, the request data block number of the write cache filter tail of the queue; if the buffer has been filled, while the request for the data block number has been recorded in the buffer queue screening among this data removed from the cache block number screening queue and request the data written to the cache; wherein the cache LRU filter queue is a queue for recording disk block number has not been recently accessed data cache hits. 本发明还公开了所述方法在读写请求执行控制中的应用。 The present invention also discloses a method of read and write requests the application control execution. 本发明不仅可以提高缓存命中率,还可以延长缓存设备的使用寿命,同时对不同特征的应用负载都可以在保证命中率的前提下尽可能减少缓存设备的写操作,无需手动进行参数调优。 The present invention not only can improve the cache hit rate, the cache may also extend the life of the device, while the application can be loaded with different characteristics while ensuring the hit rate of the cache apparatus to minimize a write operation, no manual tuning parameters.

Description

一种缓存管理方法及其应用 Of the cache management method and its application

技术领域 FIELD

[0001 ]本发明属于计算机数据存储领域,具体地,涉及一种缓存管理方法。 [0001] The present invention belongs to the field of computer data storage, in particular, to a cache management method.

背景技术 Background technique

[0002]缓存是指可以进行高速数据交换的存储器。 [0002] The cache memory means may be a high-speed data exchange. 在磁盘存储系统中,由于磁盘在读写数据时需要寻道和定位这些机械操作,因此响应速度远远慢于存储系统的处理速度。 In the disk storage system, because the disk needs to seek to locate these mechanical operations and read and write data, so that the response speed is much slower than the processing speed of the storage system. 磁盘存储系统中缓存的作用就是存储磁盘中被频繁访问的热点数据,当磁盘存储系统需要这些数据时,直接从缓存中读取,提供快速的数据响应,提高整个磁盘存储系统的性能。 Disk storage system is now cached data storage disk hot frequently accessed, when the data is needed disk storage system, read directly from the cache, provides fast response data, improve the performance of the disk storage system. 但是能够进行高速数据交换的缓存设备一般价格比较高昂,因此缓存容量有限。 However, a cache device capable of high speed data exchange is generally higher prices high, and therefore the cache capacity is limited. 缓存管理就是要在有限的缓存容量中存储频繁访问的数据,尽量提高磁盘存储系统的缓存命中率(所请求数据能够在缓存中读取到的比例),以提高整个磁盘存储系统的性能。 Cache management is to be stored in the limited capacity of the cache frequently accessed data, a cache hit ratio to maximize the disk storage system (ratio can be requested to read data in the cache) to improve the performance of the disk storage system.

[0003] 相比传统磁盘,固态硬盘(Solid State Drive)无需机械部件,没有耗时的寻道和定位操作,能够提供更高的随机访问性能,还具有低功耗、低噪音、抗震性好等优点;相比RAM(随机存储器),在相同成本预算下,固态硬盘的容量大大高于RAM,且固态硬盘具有断电非易失的特性,因此在很多存储产品中,固态硬盘被作为磁盘的缓存,以较低的成本换取存储系统性能的显著提升。 [0003] Compared to the traditional magnetic disk, SSD (Solid State Drive) without mechanical parts, no time-consuming and positioning a seek operation, to provide a higher random access performance, but also has low power consumption, low noise, good shock resistance etc; compared RAM (random access memory), in the same budget costs, much higher than the capacity of the SSD RAM, and has a characteristic solid state drives off non-volatile, so in many storage products, SSD is used as disk cache memory at a lower cost return system performance significantly.

[0004] 尽管如此,固态硬盘也有自己的缺陷,最主要的问题是擦除问题。 [0004] However, SSDs also has its own shortcomings, the main problem is to erase the problem. 固态硬盘是以闪存(Flash)为介质的存储设备,闪存(Flash)是以浮栅晶体管为单元构成的半导体存储器件,通过编程,每个单元可处于多种不同状态,从而表示一个或者多个比特的数据。 SSD is a flash memory (Flash) storage medium device, a flash memory (Flash) is a floating gate transistor is a semiconductor memory device units by programming, each cell may be in a variety of different states, thereby representing the one or more bits of data. 根据门电路结构的不同,闪存可分为NAND和NOR两类。 Depending on the configuration of the gate, NAND and NOR flash memory can be divided into two categories. 其中NAND闪存具有较高的存储密度,常用于大规模数据存储,如固态硬盘等。 Wherein the NAND flash memory having a higher storage density, commonly used in large-scale data storage, such as a solid-state disks. NAND闪存芯片通常被组织成多个块(block),一个块通常包括64〜256个页面(page)。 NAND flash memory chips are usually organized into a plurality of blocks (block), a block generally comprises 64~256 pages (page). 根据存储密度的不同,NAND闪存又分为SLC和MLC两类。 Depending on the storage density, NAND flash memory is divided into two categories SLC and MLC. SLC闪存中每个存储单元只保存I个比特的数据,而MLC闪存中每个单元可以保存多个比特,因此存储密度更高。 Each SLC flash memory cell only save the data bits I, while the MLC flash memory can store multiple bits per cell, and therefore a higher storage density. 很多大容量固态硬盘都采用MLC闪存作为介质。 Many large-capacity MLC flash SSDs are used as a medium. 闪存的物理特性决定了要更新其中存储的数据必须首先进行擦除操作,然后才能写新的数据。 Determine the physical characteristics of flash memory to update the data stored therein must first erase operation before writing new data. NAND闪存的擦除操作以块为单位进行,一次擦除将块中所有比特置为O。 NAND flash memory erase operation in units of blocks, set all bits erase block is O. 闪存的擦除次数是有限的,SLC闪存可以擦除约100,000次,而MLC缓存只能擦除大约10,000次甚至更少。 Number of erase flash memory is limited, SLC flash memory can be erased about 100,000 times, while MLC cache can only erase about 10,000 or even less. 因此,频繁的写操作会缩短使用寿命。 Therefore, frequent write operations will shorten the service life. 而在随机写密集的情况下,固态硬盘本身的日志和垃圾回收机制还将增加额外的写和擦除,进一步缩短使用寿命。 In the case of random writes dense, solid-state hard drive itself, logs and garbage collection will add an additional write and erase, to further shorten life.

[0005]传统的缓存管理方法以提高缓存命中率为唯一目标。 [0005] The traditional cache management methods to improve the cache hit rate only goal. 基于时间局部性原理(最近访问过的数据块在不久的将来很可能再次被访问),这些方法通常会把每一个访问过的数据块都写入缓存,以备将来访问。 Based on the principle of temporal locality (recently accessed data blocks are likely to be accessed again in the near future), these methods will usually visited each block of data are written to the cache for future access. 这些方法并非针对固态硬盘设计的,它们并没有考虑到固态硬盘的写寿命问题。 These methods are not designed for SSDs, they do not take into account the SSD write life issues. 当固态硬盘用作缓存设备的时候,缓存替换会带来额外的写操作。 When SSDs used as cache devices, cache replacement will bring additional write operation. 特别是对于以读为主的应用负载,缓存设备的写操作主要来自于替换操作。 Particularly for read-mostly applied load, write cache device mainly from the replacement operation. 另一方面,实际应用负载中数据块的访问频率分布是不均衡的,频繁访问的数据块通常只占一小部分,其他大部分数据块的访问频率较低,将这些数据块缓存在固态硬盘上并不能显著提升性能,反而无谓的消耗的固态硬盘的写寿命。 On the other hand, practical application access frequency distribution of the data block in the load is not balanced, frequently accessed data blocks are usually only a small portion of most other low access frequency of data blocks, the data blocks in the cache SSDs It did not significantly improve the performance of solid state drives instead of wasteful consumption of writing life.

发明内容 SUMMARY

[0006]针对现有技术的以上缺陷或改进需求,本发明提供了一种缓存管理方法及其在读写请求执行中的应用,其目的在于通过对数据块的筛选,只缓存被频繁访问的热点数据块,减少了缓存写满后的替换操作,由此解决传统缓存管理方法快速损耗固态硬盘缓存使用寿命的问题,同时解决了传统缓存管理方法将部分非频繁访问数据块写入缓存后缓存命中率不够高的问题。 [0006] To solve the above drawbacks of the prior art or the need for improvement, the present invention provides a cache management method and its application to read and write requests execution, an object of frequently accessed data blocks by screening, only the cached hot data blocks, reducing the cache replace operation is full, thereby solving the problems of the conventional cache management method is rapid loss SSD cache lifetime, while addressing the traditional cache management method of the frequently accessed data portion of the non-block write cache buffer hit rate is not high enough problems.

[0007]按照本发明的一个方面,提供一种缓存控制方法,用于对未命中缓存的请求数据进行控制,以筛选写入缓存的请求数据,减少缓存替换操作,具体包括: [0007] According to an aspect of the invention, there is provided a cache control method, a request for data cache misses is controlled to filter the requested data written to the cache, reducing cache replace operation, specifically comprising:

[0008]若缓存未被写满,则将请求数据块号写入缓存的空白空间; [0008] If the buffer is not full, the blank-space data block number of the request is written to the cache;

[0009]若缓存已被写满,同时请求数据块号未记录在缓存筛选队列中,则不将请求数据块号写入缓存,而将请求数据块号写入缓存筛选队列的尾部; [0009] When the buffer has been filled, while the request for the data block number screening is not recorded in queue, the requested data is not cached in the write cache block number, the block number of the write request data cache filter tail of the queue;

[0010]若缓存已被写满,同时请求数据块号已被记录在所述缓存筛选队列当中,则将此数据块号从该缓存筛选队列中删除,并将此请求数据写入缓存; [0010] When the buffer has been filled, while the request for the data block number has been recorded among the screened queue, the queue of this data block number is deleted from the buffer in the screening buffer, and written to the cache data request;

[0011]其中,所述缓存筛选队列为LRU队列,用于记录最近被访问却未命中缓存的数据的磁盘块号。 [0011] wherein the cache LRU filter queue is a queue for recording disk block number has not been recently accessed data cache hits.

[0012]优选地,缓存筛选队列的容量Cr可根据读写请求进行自适应调整。 [0012] Preferably, the buffer queue capacity Cr screening may be adaptively adjusted according to read and write requests.

[0013]优选地,缓存筛选队列的容量Cr可动态调整为Cr-C/(C_Cr)或Cr+C/Cr,其中,C为缓存能够存储的数据块的最大数量。 [0013] Preferably, the buffer queue capacity Cr screening can be dynamically adjusted to the maximum number of Cr-C / (C_Cr) or Cr + C / Cr, where, C is capable of storing the cache data block.

[0014]优选地,在读写请求数据未命中缓存时,所述缓存筛选队列容量Cr自适应调整为Cr+C/Cr,在读写请求数据未命中缓存时,所述缓存筛选队列容量Cr自适应调整为Cr+C/Cr。 [0014] Preferably, when a data write request misses the cache, the cache filter adaptively adjusted to queue capacity Cr Cr + C / Cr, when a data write request misses the cache, the cache filter queue capacity Cr adaptive adjustment of Cr + C / Cr.

[0015] 优选地,所述缓存筛选队列的容量Cr在调整后若Cr小于0.1 XC,则令Cr=0.1 XC,若大于0.9 XC,则令Cr=0.9 XC,其中,C为缓存能够存储的数据块的最大数量。 [0015] Preferably, the screening capacity Cr buffer queue after the adjustment when the Cr is less than 0.1 XC, then let Cr = 0.1 XC, if it exceeds 0.9 XC, then let Cr = 0.9 XC, wherein, C is capable of storing the cache the maximum number of data blocks.

[0016]优选地,所述缓存筛选队列容量Cr的初始值优选设置为0.1 XC,其中C为缓存能够存储的数据块的最大数量。 [0016] Preferably, the screening buffer queue capacity Cr is preferably set to an initial value of 0.1 XC, wherein the maximum number of data blocks can be stored as a cache C.

[0017]按照本发明的另一方面,提供一种读写请求的执行控制方法,应用上述缓存管理方法进行对读写请求数据进行缓存控制,该方法具体包括: [0017] According to another aspect of the present invention, there is provided a control method of performing read and write requests, application of the above method of managing cache data in cache control read and write requests, the method comprises:

[0018] (I)建立LRU队列作为缓存筛选队列,其中该缓存筛选队列用于记录最近被访问却未命中缓存的数据的磁盘块号; [0018] (I) as established LRU queue buffer queue screening, wherein the screening queue buffer for recording disk block number has not been recently accessed data cache hits;

[0019] (2)根据读写请求对所述缓存筛选队列容量进行自适应调整; [0019] (2) The write request queue capacity of the cache filtering adaptively adjusted;

[0020] (3)根据读写请求选择写入缓存的数据,包括: [0020] (3) The selection data read and write requests write cache, comprising:

[0021 ]若是请求数据命中缓存,则不需要进行筛选;若请求数据既未命中缓存,则执行上述权利要求1-6中任一项所述的缓存管理方法,进行缓存控制 [0021] If the request data cache hit, there is no need to filter; if neither request data cache hit, the cache management method of claim any one of the preceding claims 1-6 is performed, cache control

[0022] (4)重定向读写请求并执行 [0022] (4) read and write requests and performs redirection

[0023] (4.1)对于未被写入缓存的请求数据,则直接将读写请求重定向到磁盘上对应的地址,执行读写操作; [0023] (4.1) for the requested data is not written to the cache, the request is redirected to read and write directly to the corresponding address on the disk, read and write operations;

[0024] (4.2)对于已经存在于缓存,或者筛选后写入缓存的请求数据,将缓存中对应的数据块移到缓存队列的尾部,然后将读写请求重定向到缓存中对应的地址,执行读写操作。 Data Request [0024] (4.2) with respect to the write cache already exists in the cache, or screening, the corresponding data block in the cache buffer moved to the tail of the queue, and then redirect the request to write the address corresponding to the cache, perform read and write operations.

[0025]优选地,所述LRU队列在内存中建立,作为缓存筛选队列并初始化其容量为Cr,Cr的初始值优选设置为0.1 XC,C为缓存能够存储的数据块的最大数量。 [0025] Preferably, the LRU queue is constructed in memory as a cache and initialize the queue capacity screening of Cr, the initial value is preferably set Cr is 0.1 XC, C the maximum number of data blocks can be stored as a cache. 缓存筛选队列是一个设置在内存中的LRU(最近最少使用)队列,它的容量为Cr,Cr的初始值优选设置为0.1 XC,也可以为其他值,C为缓存能够存储的数据块的最大数量,容量Cr会随应用负载的变化而自适应调整。 The maximum cache filter queue is a set LRU in memory (Least Recently Used) queue, which has a capacity of Cr, the initial value is preferably set Cr is 0.1 XC, may be other values, C is a cache capable of storing blocks of data number, capacity Cr will vary with application load changes adaptively adjusted.

[0026]优选地,所述缓存筛选队列容量的自适应调整方法如下: [0026] Preferably, the adaptive adjustment method of screening buffer queue capacity is as follows:

[0027] (2.1)判断读写请求数据是否命中缓存,若命中,转入步骤(2.2);否则,转入步骤(2.3); [0027] (2.1) determines whether the read and write request data cache hit, if hit, proceeds to the step (2.2); otherwise, go to step (2.3);

[0028] (2.2)自动将缓存筛选队列容量Cr的大小调整为Cr-C/(C-Cr);如果调整后的Cr小于0.1 XC,则另Cr=0.1 XC,其中C为缓存能够存储的数据块的最大数量; [0028] (2.2) automatically cache sizing screening queue capacity Cr is Cr-C / (C-Cr); if Cr adjusted less than 0.1 XC, the other Cr = 0.1 XC, where C is a cache capable of storing the the maximum number of data blocks;

[0029] (2.3)自动将缓存筛选队列容量Cr的大小调整为Cr+C/Cr;如果调整后的Cr大于0.9XC,则令Cr=0.9 XC,其中C为缓存能够存储的数据块的最大数量。 [0029] (2.3) Screening buffer queue capacity automatically resized to Cr Cr + C / Cr; maximum when the Cr is greater than the adjusted 0.9XC, then let Cr = 0.9 XC, wherein C is capable of storing cache blocks of data number.

[0030]优选地,将请求数据写入缓存时,如果缓存中没有空白块,则先清空缓存队列头部对应的数据块,将清空的缓存块分配给此请求数据,然后将此请求数据块移动到缓存队列的尾部。 [0030] Preferably, the request for data written to the cache, the cache if there is no empty block, the head of the queue to empty the cache corresponding data block, the cache block assigned to the empty data request, and this request data block moving to the end of the buffer queue.

[0031]优选地,所述缓存设备为闪存。 [0031] Preferably, the buffer is a flash memory device.

[0032]本发明中,缓存筛选队列用于记录最近被访问却未命中缓存的数据的磁盘块号,它并不记录实际的磁盘数据。 [0032] In the present invention, the buffer queue is screened for data recording disk block number has not been recently accessed cache hit, it does not record the actual disk data. 当缓存筛选队列记录的数据块号的数量大于它的容量Cr,那么缓存筛选队列根据LRU(最近最少使用)算法只保留最近被访问的Cr个数据块号。 When the number of data cache filter block number recorded in the queue is larger than its capacity Cr, then the cache filtering in accordance with the queue LRU (least recently used) algorithm only retains Cr block number most recently accessed. 缓存筛选队列的头部记录的是最近最少被访问的数据块号,而缓存筛选队列的尾部记录的是最近最常被访问的数据块号。 Screening recording head of the queue buffer is a data block number of the least recently accessed, and the cache filter queue tail block number is recorded most recently accessed. 因此每次向缓存筛选队列写入一个新的数据块号时,将新数据块号插入缓存筛选队列的尾部,然后检测缓存筛选队列是否超出了容量Cr,若超出则删除缓存筛选队列的头部数据块号。 Thus each time a new data is written to the cache block number screening queue, the new data block number inserting the tail of the queue buffer screening, detection and screening buffer queue exceeds the capacity of Cr, exceeding the head buffer if the queue is deleted screening data block number.

[0033]本发明中,缓存筛选队列的容量直接影响它筛选数据块的效果。 [0033] In the present invention, the capacity of the buffer queue screening filter data block which directly affects the results. 若是缓存筛选队列容量太小,则会使很多被频繁访问的数据块没有被筛选到;若是缓存筛选队列容量太大,则会使很多并非频繁访问的数据块被筛选出来。 If screening buffer queue capacity is too small, then make a lot of data blocks are frequently accessed is not to be screened; if screening buffer queue capacity is too large, many will not frequently accessed data blocks are filtered out. 本发明的缓存筛选队列能够根据应用负载对容量调整为最优值,从而控制筛选的尺度,在保证命中率的前提下,尽可能减少不必要的替换操作。 Buffer queues of the present invention can be screened in accordance with the application of the load capacity is adjusted to the optimum value, thereby controlling scale screening, to ensure the premise of the hit rate, to minimize unnecessary replacement operation.

[0034]实际应用负载中,频繁访问的热点数据块只占一少部分,而其他大多数的数据块则访问次数较少,甚至只访问一次,将这些不常用的数据块写入缓存并不能提高缓存命中率,反而浪费了有限的缓存空间,当缓存设备为固态硬盘时,还缩短了其使用寿命。 [0034] The practical application of the load, the hotspot frequently accessed data blocks only a small part, most other data block is accessed less frequently, or even only once accessed, the write cache is not used these data blocks can not be improve the cache hit rate, but a waste of limited buffer space when the cache device is a solid-state hard drive, but also shorten its service life. 因此先筛选出这些被频繁访问的数据块,再写入容量有限的缓存中能够提高缓存的命中率,同时减少缓存数据的替换操作,延长固态硬盘缓存的使用寿命。 Thus to filter out these frequently accessed data blocks, then the write cache limited capacity cache hit rate can be improved, while reducing the cache data replacement operation, extend the service life of solid state disk cache.

[0035]本发明提出的缓存管理方法及其应用,利用缓存筛选队列,筛选出访问频率更高的数据块保存在缓存中,提高了缓存命中率;同时避免不必要的缓存替换操作,从而减少用作缓存的固态硬盘的写负载,延长固态硬盘的使用寿命。 [0035] The cache management method proposed by the present invention and its application, the use of screening buffer queue, the higher the access frequency selected data blocks stored in the cache, the cache hit rate increase; while avoiding unnecessary cache replacement operations, thereby reducing SSDs used as cache write load and prolong the life of the SSD.

附图说明 BRIEF DESCRIPTION

[0036]图1为本发明实施例的方法应用的系统的示意图; Application of the method a schematic diagram of a system [0036] Figure 1 embodiment of the present invention;

[0037]图2为本发明实施例的方法的实施流程图; Flowchart of a method embodiment [0037] FIG. 2 embodiment of the present invention;

[0038]图3为本发明实施例的缓存数据块筛选流程图。 [0038] Example 3 Screening of cache data blocks flowchart embodiment of the present invention.

具体实施方式 detailed description

[0039]为了使本发明的目的、技术方案及优点更加清楚明白,以下结合附图及实施例,对本发明进行进一步详细说明。 [0039] To make the objectives, technical solutions and advantages of the present invention will become more apparent hereinafter in conjunction with the accompanying drawings and embodiments of the present invention will be further described in detail. 应当理解,此处所描述的具体实施例仅仅用以解释本发明,并不用于限定本发明。 It should be understood that the specific embodiments described herein are only intended to illustrate the present invention and are not intended to limit the present invention.

[0040]本发明的缓冲管理方法可应用在如图1所示的以固态硬盘作为缓存设备存储系统上。 [0040] The buffer management method of the present invention may be applied as shown in the SSD as the storage system 1 caching apparatus. 本发明提出的缓存管理方法优选是针对以闪存(如固态硬盘)为缓存介质的磁盘存储系统而设计,但其应用不限于此类磁盘存储系统。 The proposed method of the present invention is preferably a buffer management system for a magnetic disk storage flash memory (e.g., SSD) is designed as a cache medium, but its use is not limited to such a disk storage system.

[0041]如图2所示,本实施例的缓存管理方法,用于对未命中缓存的请求数据进行控制,以筛选写入缓存的请求数据,减少缓存替换操作,该方法具体包括: [0041] 2, the cache management method of the present embodiment, a request for data cache misses is controlled to filter requests a data write buffer, decrease cache replacement operation, which method comprises:

[0042]若缓存未被写满,则将请求数据块号写入缓存的空白空间; [0042] If the buffer is not full, the blank-space data block number of the request is written to the cache;

[0043]若缓存已被写满,同时请求数据块号未记录在缓存筛选队列中,则不将请求数据块号写入缓存,而将请求数据块号写入缓存筛选队列的尾部;其中,所述缓存筛选队列为LRU队列,用于记录最近被访问却未命中缓存的数据的磁盘块号。 [0043] When the buffer has been filled, while the request for the data block number screening is not recorded in queue, the requested data is not cached in the write cache block number, the block number of the write request data cache filter end of the queue; wherein, the cache LRU filter queue is a queue for recording disk block number has not been recently accessed data cache hits.

[0044]若缓存已被写满,同时请求数据块号已被记录在所述缓存筛选队列当中,则将此数据块号从该缓存筛选队列中删除,并将此请求数据写入缓存; [0044] When the buffer has been filled, while the request for the data block number has been recorded among the screened queue, the queue of this data block number is deleted from the buffer in the screening buffer, and written to the cache data request;

[0045]缓存筛选队列的容量Cr可根据读写请求进行自适应调整,即调整为Cr-C/(C_Cr)或Cr+C/Cr。 [0045] Screening buffer queue capacity Cr can be adaptively adjusted according to read and write requests, i.e., adjusted to a Cr-C / (C_Cr) or Cr + C / Cr. 即在读写请求数据未命中缓存时,缓存筛选队列容量Cr自适应调整为Cr+C/Cr,在读写请求数据未命中缓存时,所述缓存筛选队列容量Cr自适应调整为Cr+C/Cr,缓存筛选队列的容量Cr在调整后若Cr小于0.1 XC,则令Cr=0.1 XC,若大于0.9 XC,则令Cr=0.9 XC,其中,C为缓存能够存储的数据块的最大数量。 I.e., when a data write request misses the cache, the cache queue capacity Cr adaptive filter is adjusted to Cr + C / Cr, when a data write request misses the cache, the cache filter adaptively adjusted to queue capacity Cr Cr + C / Cr, cache filter capacity Cr queue after the adjustment when the Cr is less than 0.1 XC, then let Cr = 0.1 XC, if it exceeds 0.9 XC, then let Cr = 0.9 XC, wherein the maximum number of data blocks C is a cache capable of storing the . 缓存筛选队列容量Cr的初始值本实施例中优选设置为0.1 XC,其中C为缓存能够存储的数据块的最大数量,也可以设置为其他值。 Screening buffer queue capacity value Cr initial preferred embodiment according to the present embodiment is set to 0.1 XC, wherein the maximum number of data blocks that can be stored as a cache C, may be set to other values.

[0046]本实施例中应用上述缓存管理方法进行读写请求控制执行的方法,其通过对读写请求数据进行缓存控制,筛选出访问频率更高的数据块保存在缓存中,实现缓存命中率的提高,同时避免不必要的缓存替换操作。 [0046] The above-described embodiment the cache management method of the present embodiment application execution method for controlling read and write requests, the read and write through cache control request data, filtering out the higher the access frequency of data blocks stored in the cache, the cache hit rate achieved increase, while avoiding unnecessary cache replacement operation. 该方法具体包括:1.建立LRU队列作为缓存筛选队列并初始化;2.缓存筛选队列容量的自适应调整;3.筛选频繁访问的数据块写入缓存;4.缓存管理及读写请求重定向。 The method includes: establishing a queue as a cache LRU filter queue and initialize; 2 adaptively filtering buffer queue capacity; 3 Screening frequently accessed data block write cache; 4 redirect cache management and read and write requests.... . 步骤具体如下: Steps are as follows:

[0047] (I)建立LRU队列作为缓存筛选队列并初始化 [0047] (I) as established LRU queue buffer queue and initialize screening

[0048]在内存中建立缓存筛选队列,缓存筛选队列是一个设置在内存中的LRU(最近最少使用)队列,它的容量为Cr,Cr的初始值本实施例中优选设置为0.1 XC,C为缓存能够存储的数据块的最大数量,容量Cr会随应用负载的变化而自适应调整。 [0048] Create cache filter queue in memory, cache filter queue is provided in memory LRU (Least Recently Used) queue, which has a capacity of Cr, the initial value Cr according to the present embodiment is preferably set to 0.1 XC, C the maximum number of data blocks that can be stored as a cache capacity Cr will change with the load application adaptively adjusted.

[0049]缓存筛选队列用于记录最近被访问却未命中缓存的数据的磁盘块号,它并不记录实际的磁盘数据。 [0049] The queue buffer used for screening recording disk block number has not been recently accessed data cache hits, it does not record the actual disk data. 当缓存筛选队列记录的数据块号的数量大于它的容量Cr,那么缓存筛选队列根据LRU(最近最少使用)算法只保留最近被访问的Cr个数据块号。 When the number of data cache filter block number recorded in the queue is larger than its capacity Cr, then the cache filtering in accordance with the queue LRU (least recently used) algorithm only retains Cr block number most recently accessed.

[0050]缓存筛选队列的头部记录的是最近最少被访问的数据块号,而缓存筛选队列的尾部记录的是最近被访问的数据块号。 [0050] The cache filter head of the queue is a data recording block number of the least recently accessed, and the cache filter queue tail block number is recorded most recently accessed. 因此每次向缓存筛选队列写入一个新的数据块号时,将新数据块号插入缓存筛选队列的尾部,然后检测缓存筛选队列是否超出了容量Cr,若超出则删除缓存筛选队列的头部数据块号。 Thus each time a new data is written to the cache block number screening queue, the new data block number inserting the tail of the queue buffer screening, detection and screening buffer queue exceeds the capacity of Cr, exceeding the head buffer if the queue is deleted screening data block number.

[0051 ] (2)缓存筛选队列容量的自适应调整 [0051] (2) adaptively filtering buffer queue capacity

[0052]缓存筛选队列的容量直接影响它筛选数据块的效果。 [0052] The queue buffer directly affect the screening effect of the screening data block it. 若是缓存筛选队列容量太小,则会使很多被频繁访问的数据块没有被筛选到;若是缓存筛选队列容量太大,则会使很多并非频繁访问的数据块被筛选出来。 If screening buffer queue capacity is too small, then make a lot of data blocks are frequently accessed is not to be screened; if screening buffer queue capacity is too large, many will not frequently accessed data blocks are filtered out. 本发明的实施例中采用一种负载自适应方式,能够根据应用负载的特征自动将缓存筛选队列的容量调整为最优值,从而控制筛选的尺度,在保证命中率的前提下,尽可能减少不必要的替换操作。 Embodiments of the present invention uses a load adaptive mode, the cache capacity can be automatically adjusted according to the filter characteristic of the queue is the optimum value of the applied load, thereby controlling scale screening, to ensure the hit rate of the premise, to minimize unnecessary replacement operation.

[0053]缓存筛选队列容量的自适应调整方法如下: [0053] Adaptive filter adjustment method buffer queue capacity is as follows:

[0054] (2.1)每次接收到读写请求,判断请求数据是否命中缓存,若命中,转入步骤(2.2);否则,转入步骤(2.3); [0054] (2.1) each time it receives read and write requests, determining whether the requested data cache hit, if hit, proceeds to the step (2.2); otherwise, go to step (2.3);

[0055] (2.2)自动将缓存筛选队列容量Cr的大小调整为Cr-C/(C_Cr);如果调整后的Cr小于0.1 XC,则另Cr=0.1 XC,其中C为缓存能够存储的数据块的最大数量; [0055] (2.2) automatically cache filter queue capacity Cr resized to Cr-C / (C_Cr); if Cr adjusted less than 0.1 XC, the other Cr = 0.1 XC, where C is a cache capable of storing blocks of data the maximum number;

[0056] (2.3)自动将缓存筛选队列容量Cr的大小调整为Cr+C/Cr;如果调整后的Cr大于0.9XC,则令Cr=0.9 XC,其中C为缓存能够存储的数据块的最大数量。 [0056] (2.3) Screening buffer queue capacity automatically resized to Cr Cr + C / Cr; maximum when the Cr is greater than the adjusted 0.9XC, then let Cr = 0.9 XC, wherein C is capable of storing cache blocks of data number.

[0057] (3)筛选出被频繁访问的数据块写入缓存 [0057] (3) filter out data written in the cache frequently accessed blocks

[0058]实际应用负载中,频繁访问的热点数据块只占一少部分,而其他大多数的数据块则访问次数较少,甚至只访问一次,将这些不常用的数据块写入缓存并不能提高缓存命中率,反而浪费了有限的缓存空间,当缓存设备为固态硬盘时,还缩短了其使用寿命。 [0058] The practical application of the load, the hotspot frequently accessed data blocks only a small part, most other data block is accessed less frequently, or even only once accessed, the write cache is not used these data blocks can not be improve the cache hit rate, but a waste of limited buffer space when the cache device is a solid-state hard drive, but also shorten its service life. 因此先筛选出这些被频繁访问的数据块,再写入容量有限的缓存中能够提高缓存的命中率,同时减少缓存数据的替换操作,延长固态硬盘缓存的使用寿命。 Thus to filter out these frequently accessed data blocks, then the write cache limited capacity cache hit rate can be improved, while reducing the cache data replacement operation, extend the service life of solid state disk cache.

[0059] —般来说,在一段较短的时间内被访问过两次或以上的数据块未来很可能被频繁访问,这些数据块就是潜在的热点数据块。 [0059] - In general, in a short period of time has been visited twice or more data blocks of the future is likely to be frequently accessed, the data block is a potential hot data blocks. 当缓存空间填满之后,只有发现了新的热点数据块,才会进行缓存替换操作,将其写入缓存。 When the buffer space fills, only to find new hot data blocks, only cache replace operation, the write buffer.

[0060]如图3所示,具体筛选方法如下: [0060] As shown in FIG. 3, the specific screening methods are as follows:

[0061 ] (3.1)若是请求数据命中缓存,则不需要进行筛选; [0061] (3.1) If the requested data cache hit, there is no need to filter;

[0062] (3.2)若请求数据既未命中缓存,但是缓存仍有空白缓存空间,则分配一块空白缓存空间存储请求数据,并将请求数据块移动到缓存队列的尾部(缓存队列一般也是LRU队列,缓存队列的尾部表示最近被访问的数据块)。 [0062] (3.2) If the request is neither a data cache hit, but the cache buffer space is still empty, an empty buffer space allocated data storage request, and the requested block of data is moved to the tail of the queue buffer (LRU queue buffer queue is typically , data indicating the tail of the queue buffer block most recently accessed).

[0063]缓存队列记录的是存储于缓存中的数据块的块号,它的尾部数据块号表示对应的数据块最近最常被访问;当需要用新的数据替换缓存中原有的数据时,就是根据缓存队列来判断应该删除哪个数据块,腾出空间给新的数据块。 [0063] buffer queue record is stored in the cache block number of the data block, its tail block number indicates that the corresponding data block is accessed most recently; when the cache needs to be replaced any original data with the new data, It is based buffer queue to determine which data blocks should be deleted to make room for new data block.

[0064] (3.3)若请求数据既未命中缓存,且缓存已被写满,同时请求数据块号未记录在缓存筛选队列中,则将请求数据块号写入缓存筛选队列的尾部。 [0064] (3.3) if the request data is neither a cache hit and the cache has been filled, while the requested data block number is not recorded in the cache screening queue, the data written to the cache block number in the request is screened tail of the queue.

[0065]若是此时缓存筛选队列的长度超过了它的自适应调整容量Cr,则删除缓存筛选队列的头部数据块号。 [0065] At this time, if the queue length exceeds the cache filter adaptive to adjust its capacity Cr, then delete the queue header data cache filter block number.

[0066] (3.4)若请求数据未命中缓存,且缓存已被写满,同时请求数据块号已被记录在缓存筛选队列当中,那么说明此数据块在最近一段时间已经是第2次被访问,视此数据块为频繁访问数据块,将此数据块号从缓存筛选队列中删除,并将此请求数据写入缓存。 [0066] (3.4) If the requested data cache misses, and the cache has been filled, while the request for the data block number has been recorded in the buffer queue screening them, then that data block is already in the most recent is the second to be accessed , depending on the data block of the data blocks frequently accessed, this data is deleted from the cache filter block number queue, and the request data is written to the cache.

[0067]将请求数据写入缓存前,先清空缓存队列头部对应的数据块,将清空的缓存块分配给此请求数据,然后将此请求数据块移动到缓存队列的尾部。 [0067] the requested data before the write cache to empty the cache block of data corresponding to the head of the queue will empty the cache block assigned to a data request, and this request to the cache block of data is moved to the tail of the queue.

[0068] (4)重定向读写请求并执行: [0068] (4) read and write requests and performs redirection:

[0069] (4.1)对于未被写入缓存的请求数据,则直接将读写请求重定向到磁盘上对应的地址,执行; [0069] (4.1) for the requested data is not written to the cache, the request is redirected to read and write directly to the corresponding address on the disk is executed;

[0070] (4.2)对于已经存在于缓存,或者筛选后写入缓存的请求数据,将缓存中对应的数据块移到缓存队列的尾部,然后将读写请求重定向到缓存中对应的地址,执行。 Data Request [0070] (4.2) with respect to the write cache already exists in the cache, or screening, the corresponding data block in the cache buffer moved to the tail of the queue, and then redirect the request to write the address corresponding to the cache, carried out.

[0071]本领域的技术人员容易理解,以上所述仅为本发明的较佳实施例而已,并不用以限制本发明,凡在本发明的精神和原则之内所作的任何修改、等同替换和改进等,均应包含在本发明的保护范围之内。 [0071] Those skilled in the art will readily appreciate, the above-described preferred embodiment of the present invention only but are not intended to limit the present invention, any modifications within the spirit and principle of the present invention, equivalent substitutions, and improvements, etc., should be included within the scope of the present invention.

Claims (9)

  1. 1.一种缓存管理方法,用于对未命中缓存的请求数据进行控制,以筛选写入缓存的请求数据,减少缓存替换操作,其特征在于,该方法具体包括: 若缓存未被写满,则将请求数据块写入缓存的空白空间,并将请求数据块号移动到缓存筛选队列的尾部; 若缓存已被写满,同时请求数据块号未记录在缓存筛选队列中,则不将请求数据块写入缓存,而将请求数据块号写入缓存筛选队列的尾部; 若缓存已被写满,同时请求数据块号已被记录在所述缓存筛选队列当中,则将此数据块号从该缓存筛选队列中删除,并将此请求数据块写入缓存,其中在将请求数据写入缓存前,先清空缓筛选存队列头部对应的数据块,将清空的缓存块分配给此请求数据; 其中,所述缓存筛选队列为LRU队列,用于记录最近被访问却未命中缓存的数据的磁盘块号,该缓存筛选队列的头部 A cache management method, for a data request misses the cache control request to filter the data written to the cache, reducing cache replacement operation, characterized in that the method comprises: if the buffer is not completely filled, blank-space data block is then written to the cache request, and the request data block number to the mobile screening tail of the queue buffer; if the buffer has been filled, while the requested data block number is not recorded in the cache screening queue, the request is not blocks the write cache block number of the write request data cache filter end of the queue; if the buffer has been filled, while the request for the data block number has been recorded among the screened queue, then this data block number in the cache from remove the filter buffer queue, and the cache writes the data block request, wherein the request before the data written to the cache, the first head of the queue empty buffer memory filter corresponding data block, the cache block assigned to this clear request data ; wherein the cache LRU filter queue is a queue for data recording disk block number has not been recently accessed cache hit, the head of the queue buffer screening 记录的是最近最少被访问的数据块号,而缓存筛选队列的尾部记录的是最近最常被访问的数据块号。 Recording block number is the least recently accessed, and the cache filter queue tail block number is recorded most recently accessed.
  2. 2.根据权利要求1所述的一种缓存管理方法,其特征在于,所述缓存筛选队列的容量Cr可根据读写请求进行动态调整,即调整为Cr-C/(C-Cr)或Cr+C/Cr,其中,C为缓存能够存储的数据块的最大数量。 A cache management method according to claim 1, wherein said buffer queue capacity Cr screening can be dynamically adjusted based on read and write requests, i.e., adjusted to a Cr-C / (C-Cr) or Cr + C / Cr, wherein the maximum number of data blocks can be stored as a cache C.
  3. 3.根据权利要求2所述的一种缓存管理方法,其特征在于,在读写请求数据未命中缓存时,所述缓存筛选队列容量Cr调整为Cr+C/Cr,其中,C为缓存能够存储的数据块的最大数量。 A cache management method according to claim 2, wherein, when a data write request misses the cache, the cache filter adjusted to queue capacity Cr Cr + C / Cr, where, C is a cache capable of the maximum number of data blocks stored therein.
  4. 4.根据权利要求2所述的一种缓存管理方法,其特征在于,所述缓存筛选队列的容量Cr在调整后若Cr小于0.1 XC,则将其容量重新设置为Cr = 0.1 XC,若大于0.9 XC,则将其容量设置为Cr = 0.9 XC。 A cache management method according to claim 2, wherein said buffer queue capacity Cr screened after the adjustment when the Cr is less than 0.1 XC, its capacity will be reset to Cr = 0.1 XC, if greater than 0.9 XC, then the capacity is set to Cr = 0.9 XC.
  5. 5.根据权利要求2-4中任一项所述的一种缓存管理方法,其特征在于,所述缓存筛选队列容量Cr的初始值优选设置为0.1XCo A cache management method according to any one of the 2-4 claims, wherein said initial buffer queue capacity value Cr is preferably provided for screening 0.1XCo
  6. 6.根据权利要求1-4中任一项所述的一种缓存管理方法,其特征在于,在将请求数据块号记录到所述缓存筛选队列后,若缓存筛选队列的实际长度超过其容量,则删除该缓存筛选队列的头部数据块号。 A cache management method as claimed in any one of the preceding claims, characterized in that the data block number in the request record to the queue buffer after screening, if screening buffer queue length exceeds its actual capacity , deleting the header data cache filter queue block number.
  7. 7.—种读写请求的执行控制方法,其应用上述权利要求1-6中任一项所述的缓存管理方法对读写请求数据进行缓存控制,其特征在于,该方法具体包括: (1)建立L RU队列作为缓存筛选队列,该缓存筛选队列用于记录最近被访问却未命中缓存的数据的磁盘块号; (2)根据读写请求对所述缓存筛选队列容量进行自适应调整; (3)根据读写请求选择写入缓存的数据,包括: 若请求数据命中缓存,则不进行筛选,转到步骤(4);若请求数据未命中缓存,则执行上述权利要求1-6中任一项所述的缓存管理方法进行缓存控制; (4)重定向读写请求并执行(4.1)对于未被写入缓存的请求数据,则直接将读写请求重定向到磁盘上对应的地址,执行读写操作; (4.2)对于已经存在于缓存,或者筛选后写入缓存的请求数据,将缓存中对应的数据块号移到缓存筛选队列的尾部,然后将读写请求 7.- kinds of read and write requests execution control method, its application, preceding the cache management method as claimed in any one of the requirements for read and write request data cache control, characterized in that the method comprises: (1 ) to establish a queue L RU screening buffer queue, the queue buffer screening for recording disk block number has not been recently accessed data cache hits; (2) according to the read or write request queue buffer capacity adaptive filter adjustment; (3) the selection data read and write requests to the write cache, comprising: a cache hit if the request data, the screening is not performed, go to step (4); if the request for data cache miss, performing the above claims 1-6 cache management method according to any one of the cache control is performed; (4) read and write requests and performs redirection (4.1) for the requested data is not written to the cache, the request is redirected to read and write directly to the corresponding address on the disk performing read and write operations; (4.2) for the requested data already exists in the cache, the write cache or after screening, the data block number in the cache buffer corresponding to the filter to move the tail of the queue, then the read and write requests 重定向到缓存中对应的地址,执行读写操作。 Redirected to the address corresponding to the cache, read and write operations.
  8. 8.根据权利要求7所述的执行控制方法,其特征在于,所述缓存筛选队列容量的自适应调整方法如下: (2.1)判断读写请求数据是否命中缓存,若命中,转入步骤(2.2);否则,转入步骤(2.3); (2.2)自动将缓存筛选队列容量Cr的大小调整为Cr-C/(C-Cr);如果调整后的Cr小于0.1XC,则将缓存筛选队列容量设置为Cr=0.1 XC,其中C为缓存能够存储的数据块的最大数量; (2.3 )自动将缓存筛选队列容量Cr的大小调整为Cr+C/Cr,如果调整后的Cr大于0.9 XC,则将缓存筛选队列容量设置为Cr = 0.9 Xe。 8. The execution control method according to claim 7, wherein said adaptive adjustment method of screening buffer queue capacity as follows: (2.1) determines whether the read and write request data cache hit, if hit, proceeds to step (2.2 ); otherwise, go to step (2.3); (2.2) screening buffer queue capacity automatically resized to Cr Cr-C / (C-Cr); If Cr is less than the adjusted 0.1XC, then the buffer queue capacity screening to Cr = 0.1 XC, the maximum number of data blocks where C is a cache capable of storing; (2.3) automatically cache filter queue capacity Cr resized to Cr + C / Cr, when the Cr adjusted greater than 0.9 XC, the the screening buffer queue capacity to Cr = 0.9 Xe.
  9. 9.根据权利要求7或8所述的执行控制方法,其特征在于,将请求数据写入缓存时,如果缓存中没有空白块,则先清空缓存筛选队列头部对应的数据块,将清空的缓存块分配给此请求数据,然后将此请求数据块号移动到缓存筛选队列的尾部。 Execution control method according to claim 7 or claim 8, wherein the request for data written to the cache, the cache if there is no empty block, the head of the queue to empty the cache filter corresponding data block, the empty cache blocks allocated to the request data, then this data block number to request the mobile screening tail of the queue buffer.
CN 201310138419 2013-04-19 2013-04-19 Of the cache management method and its application CN103257935B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201310138419 CN103257935B (en) 2013-04-19 2013-04-19 Of the cache management method and its application

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201310138419 CN103257935B (en) 2013-04-19 2013-04-19 Of the cache management method and its application

Publications (2)

Publication Number Publication Date
CN103257935A true CN103257935A (en) 2013-08-21
CN103257935B true CN103257935B (en) 2016-07-13

Family

ID=48961865

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201310138419 CN103257935B (en) 2013-04-19 2013-04-19 Of the cache management method and its application

Country Status (1)

Country Link
CN (1) CN103257935B (en)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103577348A (en) * 2013-10-09 2014-02-12 广东欧珀移动通信有限公司 Method and mobile device for automatically counting application cache size and reminding user
CN103744624B (en) * 2014-01-10 2017-09-22 浪潮电子信息产业股份有限公司 Ssd one implementation, system cache data storage system architecture selective upgrading
CN104123238A (en) * 2014-06-30 2014-10-29 海视云(北京)科技有限公司 Data storage method and device
CN104360966B (en) * 2014-11-21 2017-12-12 浪潮(北京)电子信息产业有限公司 Method and apparatus for block data input and output operations
CN105095495A (en) * 2015-08-21 2015-11-25 浪潮(北京)电子信息产业有限公司 Distributed file system cache management method and system
CN107463509A (en) * 2016-06-05 2017-12-12 华为技术有限公司 Cache management method, cache controller and computer system

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102637147A (en) * 2011-11-14 2012-08-15 天津神舟通用数据技术有限公司 Storage system using solid state disk as computer write cache and corresponding management scheduling method
CN102945207A (en) * 2012-10-26 2013-02-27 浪潮(北京)电子信息产业有限公司 Cache management method and system for block-level data
CN103049394A (en) * 2012-11-30 2013-04-17 记忆科技(深圳)有限公司 Method and system for data caching of solid state disk

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1505506A1 (en) * 2003-08-05 2005-02-09 SAP Aktiengesellschaft A method of data caching

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102637147A (en) * 2011-11-14 2012-08-15 天津神舟通用数据技术有限公司 Storage system using solid state disk as computer write cache and corresponding management scheduling method
CN102945207A (en) * 2012-10-26 2013-02-27 浪潮(北京)电子信息产业有限公司 Cache management method and system for block-level data
CN103049394A (en) * 2012-11-30 2013-04-17 记忆科技(深圳)有限公司 Method and system for data caching of solid state disk

Also Published As

Publication number Publication date Type
CN103257935A (en) 2013-08-21 application

Similar Documents

Publication Publication Date Title
US7500063B2 (en) Method and apparatus for managing a cache memory in a mass-storage system
US7173863B2 (en) Flash controller cache architecture
US20100088459A1 (en) Improved Hybrid Drive
US20100011154A1 (en) Data accessing method for flash memory and storage system and controller using the same
US7472222B2 (en) HDD having both DRAM and flash memory
US20120166749A1 (en) Data management in solid-state storage devices and tiered storage systems
US20140095775A1 (en) Systems and methods for cache endurance
US20110145479A1 (en) Efficient use of hybrid media in cache architectures
US20090265506A1 (en) Storage device
US20070094445A1 (en) Method to enable fast disk caching and efficient operations on solid state disks
US20030088739A1 (en) Adaptive data insertion for caching
US7039765B1 (en) Techniques for cache memory management using read and write operations
US20140181432A1 (en) Priority-based garbage collection for data storage systems
US20110320687A1 (en) Reducing write amplification in a cache with flash memory used as a write cache
US20110055458A1 (en) Page based management of flash storage
US20140229654A1 (en) Garbage Collection with Demotion of Valid Data to a Lower Memory Tier
US20080114930A1 (en) Disk drive with cache having volatile and nonvolatile memory
US20120239851A1 (en) Prioritized erasure of data blocks in a flash storage device
US20120054421A1 (en) Information device equipped with cache memories, apparatus and program using the same device
US20140229656A1 (en) Multi-Tiered Memory with Different Metadata Levels
US20140095827A1 (en) Memory storage device, and a related zone-based block management and mapping method
US20140379968A1 (en) Memory system having a plurality of writing mode
CN101673188A (en) Data access method for solid state disk
US20160188223A1 (en) Promoting consistent response times in a data storage system having multiple data retrieval mechanisms
Huang et al. Improving flash-based disk cache with lazy adaptive replacement

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination
C14 Grant of patent or utility model