CN102298506B - Storage system and method for implementing the same - Google Patents
Storage system and method for implementing the same Download PDFInfo
- Publication number
- CN102298506B CN102298506B CN201010213533.XA CN201010213533A CN102298506B CN 102298506 B CN102298506 B CN 102298506B CN 201010213533 A CN201010213533 A CN 201010213533A CN 102298506 B CN102298506 B CN 102298506B
- Authority
- CN
- China
- Prior art keywords
- solid
- state disk
- response
- document
- 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.)
- Active
Links
Abstract
The invention discloses a storage system and a method for implementing the same. The storage system comprises a disc array and a storage control device, wherein the disc array comprises at least one solid-state disc and at least one non-solid-state disc; the storage control device is configured to respond to entering into a clearing mode, scanning and moving data blocks in the at least one non-solid-state disc in the disc array to form more continuous idle blocks. The storage system provided by the invention has better reading and writing performance, higher data reliability and usability and lower cost.
Description
Technical field
The present invention relates to computer realm, be specifically related to a kind of storage system and for realizing the method for storage system.
Background technology
Storage system is one of most important parts in computer system.People wish that storage system has at a high speed (high IOPS and high throughput), high reliability and low cost always.
Most of existing servers or personal computer are still used hard disk (HDD) as main storage system.Although the storage system based on HDD has lower cost, the improvement of HDD access latency in recent years lags far behind the improvement of its capacity.The high stand-by period will greatly affect IOPS (input/output operations number per second) value.Especially for the dominant working load of random I/O, data base workload for example, HDD will become performance bottleneck.And for the dominant working load of sequential access, the access performance of HDD is better, for example, current single HDD can obtain the transfer rate up to the continuation of 160MB/s.
Compare with HDD, the solid-state disk (SSD) occurring in recent years has lower random read waiting time (5ms of the 0.03ms of SSD to HDD), the higher IOPS (35000 couples of HDD of SSD 200) that reads.Although originally the capacity of SSD is less, also caught up with gradually in recent years the capacity of HDD.For example, have at present some SSD as the 2TB of high-end product.The energy consumption of SSD is also lower than HDD.Yet SSD has following several shortcoming: the cost of the one, SDD is far above HDD (take current market price as example, the $ 0.6/GB of the $ 10/GB of SSD to HDD).In SSD, controller is extremely expensive.Each unit of the 2nd, SSD only has limited write cycle time, and this can affect its reliability.Although introduced several technology, overcome this problem, they can increase the cost of controller, and reduce IOPS and throughput.The readwrite performance of the 3rd, SSD is asymmetrical.The write latency ratio of SDD is read large 10 times of stand-by period.This is because first write operation must wipe whole (each piece is approximately 0.5-1MB), then legacy data in piece is write to this piece together with new data, and this operation is very slow.
How making storage system have higher readwrite performance, reliability and low cost, is a kind of important problem.Exist at present following several solution.
The one, by HDD, form RAID (RAID array), random read-write is distributed in some HDD, to improve readwrite performance and reliability.But the energy consumption cost of this solution is higher, racks of data centers space hold is larger, and improvement in performance is limited.
The 2nd, utilize lower the reading the stand-by period of SSD, use the part realizing in controller firmware to take that piece merges and abrasion equilibrium algorithm (wear-leveling algorithm) alleviates the negative consequence of writing at random simultaneously, this comprises: with large continuous segment, write data, at backstage assembling section, take piece and free block is carried out to refuse collection, and avoid constantly writing some focus pieces.Yet frequent backstage file system scans identification division to take quite a lot of bandwidth that piece and free block can consume SSD controller.And firmware is difficult to realize complicated garbage collection algorithm, and cost is very high.In addition, single SSD cannot provide high availability, and SSD array is too expensive.
The 3rd, the high-speed cache by SSD for HDD.This scheme does not have high reliability feature, and has the negative consequence of writing, and as limited in write cycle time and write latency is long etc.
The 4th, combination S SD and HDD, and carry out manual data subregion.For example, in GPFS (general parallel file system), by manual assignment by metadata store on SSD, and store data on HDD.This scheme is not only brought very large administrative burden, and cannot realize higher reading and writing data performance and the high availability of data.
Summary of the invention
For shortcoming of the prior art, technical scheme of the present invention has been proposed.
According to an aspect of the present invention, proposed a kind of storage system, having comprised: the disk array that comprises at least one solid-state disk and at least one non-solid-state disk; And memory control device, it is configured to: in response to entering cleaning pattern, scan and move the data block at least one non-solid-state disk in described disk array, form how continuous free block.
According to another aspect of the present invention, proposed a kind of for realizing the method for storage system, wherein, described storage system comprises disk array, this disk array comprises at least one solid-state disk and at least one non-solid-state disk, the method comprises: in response to entering cleaning pattern, scan and move the data block at least one non-solid-state disk in described disk array, form how continuous free block.
Advantage of the present invention comprises one or more in following several: storage system has not only obtained the good performance of reading by read data from solid-state disk, and by cleaning pattern, non-solid-state disk being cleared up to form more free block continuously, and there is good write performance; By data being write in solid-state disk and in non-solid-state disk, data reliability and availability have been improved; Driver by operating system is realized part and is taken the functions such as the merging of piece and abrasion equilibrium, writing speed and the serviceable life of having improved solid-state disk, and reduced cost.
Accompanying drawing explanation
In claims, set forth the creative feature that is considered to feature of the present invention.But, by with reference to accompanying drawing, read below to the detailed description of illustrative embodiment can understand better invention itself with and preferably use pattern, target, feature and advantage, in the accompanying drawings:
Fig. 1 shows storage system according to an embodiment of the invention;
Fig. 2 show according to an embodiment of the present for realizing the method for storage system; And
Fig. 3 show according to another kind of embodiment of the present invention for realizing the method for storage system.
Embodiment
Embodiments of the invention are described with reference to the accompanying drawings.In the following description, many details have been set forth to more fully understand the present invention.But, for those skilled in the art, being apparent that, realization of the present invention can not have some in these details.In addition, should be understood that, the present invention is not limited to introduced specific embodiment.On the contrary, can consider to implement the present invention by feature below and the combination in any of key element, and no matter whether they relate to different embodiment.Therefore, use for illustrative purposes only of aspect below, feature, embodiment and advantage and should not be counted as key element or the restriction of claims, unless clearly proposed in claim.
Referring now to Fig. 1, it shows the storage system according to one embodiment of the invention.As shown in the figure, this storage system comprises: disk array 110, and this disk array 110 comprises at least one solid-state disk 111 and at least one non-solid-state disk 112; And memory control device 120, this memory control device is configured to: in response to entering cleaning pattern, scan and move the data block at least one non-solid-state disk in described disk array, form more (comparing before more with scanning with movement) continuous free block.
Described non-solid-state disk 112 refers to for random read-write can cause tracking time delay and the storage medium of speed during the sequential write higher writing rate of solid-state disk (for example, close to).Typical example is hard disk (HDD), and it is by when disc rotation, and in road, (track) and sector (sector) upper moving head reads and writes data, and therefore, when random read-write, can cause tracking time delay; And when sequential write, can save tracking time delay.Other example is optical disc and the magneto-optic disks such as some erasable CD, DVD, and the speed of its sequential write is higher, and close to SSD.
Although only show a solid-state disk 111 and a non-solid-state disk 112 in figure, the present invention is not limited to this.The disk array 110 of storage system also can comprise a plurality of solid-state disks 111 and a non-solid-state disk 112 (can be described as N+1 pattern) according to an embodiment of the invention, or comprise a solid-state disk 111 and a plurality of non-solid-state disk 112 (can be described as 1+M pattern), or comprise a plurality of solid-state disks 111 and a plurality of non-solid-state disk 112 (can be described as N+M pattern).First the disk array 110 of storage system of take below comprises that a solid-state disk 111 and non-solid-state disk 112 illustrate embodiments of the invention as example.
In a preferred embodiment of the invention, described memory control device 120 is that device driver in operating system (OS) is realized.In other embodiments of the invention, described memory control device 120 also can be realized by other devices or functional module, for example, can be realized by the controller of disk array 110.Although in a preferred embodiment of the invention, the repertoire of described memory control device 120 is realized by software, and in other embodiments of the invention, all or part of function of described memory control device 120 also can be realized by firmware or hardware.
Memory control device 120, for receiving the storage operation request from upper layer application by file system, is processed storage operation request by disk array 110 is carried out to corresponding operating, and is returned to result to upper layer application.
According to an embodiment of the present, described memory control device 120 is also configured to: in response to entering reading mode, from described solid-state disk 111, read corresponding data; And in response to entering WriteMode, to described solid-state disk 111 and described non-solid-state disk 112, write corresponding data.When memory control device 120 receives and start to process the read request from upper layer application, enter reading mode.In reading mode, memory control device 120 will only read asked data from solid-state disk 111.Because the read access speed of solid-state disk 111 is higher, can meet quickly this read request like this.When memory control device 120 receives and process the write request from upper layer application, enter WriteMode.In WriteMode, the data that memory control device 120 is write request write respectively solid-state disk 111 and non-solid-state disk 112 both.Because memory control device 120 can write data the continuous free block that cleaning pattern produces on non-solid-state disk 112 and the free block producing on solid-state disk 111, so solid-state disk 111 and non-solid-state disk 112 can provide very high write performance.Owing to identical data being write to solid-state disk 111 and non-solid-state disk 112 these two positions, for example, when data that like this, can be in a dish are unavailable fault or the wearing and tearing of dish (, due to), data in another dish are provided, thereby have improved the reliabilty and availability of data.
According to an embodiment of the present, described memory control device 120 is also configured to: after described solid-state disk 111 reads corresponding data, verify the integrality of described corresponding data; In response to described, be verified, to upper layer application, return to described corresponding data; And do not pass through in response to described checking, from described non-solid-state disk 112, read corresponding data, this corresponding data is returned to described solid-state disk 111, and return to this corresponding data to upper layer application.
As known to persons skilled in the art, for example can be by using the check bit of storing together with data to carry out the integrality of verification msg.Reading corresponding data from described non-solid-state disk 112, also can first verify the integrality of these data, if the verification passes, these data are returned to solid-state disk 111, and return to this data to upper layer application; If checking is not passed through, return to mistake.In addition,, if checking is not passed through, also can further this event be charged to daily record, and can give notice, so that keeper adopts remedial measures.
In execution, data are returned to after the operation of solid-state disk 111, also can first judge whether data are successfully returned to solid-state disk 111.If data are successfully returned to solid-state disk 111, to upper layer application, return to this data; Otherwise, return to mistake.In addition, in data, fail return to solid-state disk 111 in the situation that, also can further this event be charged to daily record, and can send notice, so that keeper adopts remedial measures.
According to an embodiment of the present, to non-solid-state disk 112, write corresponding data and comprise: find continuous free block in non-solid-state disk 112, and write corresponding data to continuous free block in non-solid-state disk 112.As known to persons skilled in the art, when writing data in the non-solid-state disks 112 such as HDD, if continuous free block is write data wherein, can save seek time, thereby can improve writing speed, may reach and write to solid-state disk 111 speed that data are suitable, so just making storage system according to an embodiment of the invention not only there is the good performance of reading, and there is good write performance.
According to an embodiment of the present, if do not find enough continuous free blocks in non-solid-state disk 112, can write corresponding data to discontinuous free block in non-solid-state disk 112.Like this, write performance can decrease.Therefore, can be by corresponding event (for example, in non-solid-state disk 112, there is no enough continuous free blocks) charge to daily record or send notice, to adopted remedial measures by memory control device 120 or keeper, for example make memory control device 120 enter hereinafter described cleaning pattern, so that the data in non-solid-state disk 112 are cleared up to form how continuous free block, or shift the data in non-solid-state disk 112, or change non-solid-state disk 112 etc.
Can find in many ways continuous free block in non-solid-state disk 112.For example, can be safeguarded by memory control device 120 record of whole free blocks in non-solid-state disk 112.Whenever to non-solid-state disk 112 data writings or while deleting the data in non-solid-state disk 112, this record of corresponding renewal, makes this record record all the time the whole free blocks in non-solid-state disk 112; In addition, when as described below when non-solid-state disk 112 is cleared up in cleaning pattern, also this record of corresponding renewal.Like this, can record to find free block continuous in solid-state disk 111 by inquiring about this.Certainly, also can find by other means continuous free block in non-solid-state disk 112.For example, also can be by directly non-solid-state disk 112 being scanned to find wherein continuous free block.
As known to persons skilled in the art, when writing data in solid-state disk 111, can adopt abrasion equilibrium algorithm.This function normally realizes in the prior art in the controller firmware of solid-state disk 111.And in an embodiment of the present invention, this function is to realize, like this, just saved greatly the firmware cost of solid-state disk 111 in the driver of operating system.
As mentioned above, according to an embodiment of the present, in described cleaning pattern, the non-solid-state disk 112 of memory control device 120 scanning, and move data block wherein, to form how continuous free block.Like this, when writing new data, just more likely find continuous free block, and data are write to continuous free block, rather than data are write to discontinuous free block, thereby improve writing speed.After Mobile data piece, memory control device 120 can the above-mentioned whole free block records of corresponding renewal, to reflect in non-solid-state disk 112, through data block, move the new free block of formation.
According to an embodiment of the present, described memory control device 120 is also configured to: in described cleaning pattern, scan and merge the data block that in described disk array 110, the part in solid-state disk 111 takies, forming how complete free block.That is to say, by the data in the data block that a plurality of parts are taken, merge in one or more, make described one or more to become the data block taking completely, can save how complete free block (as known in the art, because solid-state disk is a kind of random access storage device, to the access time of data and data, the physical location on dish is irrelevant, so free block needs not to be continuous).Like this, when writing data in solid-state disk 111, just more likely directly in complete free block, write data.As known to persons skilled in the art, to complete free block in solid-state disk, write fashionablely, can save the work of reading this piece legacy data, erasable whole data block, thereby greatly improve the performance that solid-state disk writes.In current solid-state disk technology, the function of this assembling section free block realizes in controller firmware, and in one embodiment of this invention, this function is to realize in the driver of operating system, thereby has saved firmware cost.
Can come regulation memory control device 120 how to enter and exit cleaning pattern with multiple strategy.Each strategy has been stipulated the described particular event that enters cleaning pattern for triggering.Only, as example, the first strategy can be stipulated, when not receiving read request and write request in schedule time threshold value T, to enter cleaning pattern; When the ratio that continuous free block accounts for all free blocks in non-solid-state disk 112 surpasses predetermined ratio threshold value C, or when receiving read request or write request, can exit cleaning pattern.
The second strategy can be stipulated, when Disk bandwidth utilization rate is during lower than predetermined utilization rate threshold value BW, to enter cleaning pattern; And when Disk bandwidth utilization rate is during higher than predetermined utilization rate threshold value BW, or the ratio that continuous free block accounts for all free blocks in non-solid-state disk 112 is while surpassing predetermined ratio threshold value C, can exit cleaning pattern.
The third strategy can be stipulated in memory control device 120 when receiving length from the read request on upper strata and the request queue of write request and be less than predetermined length threshold value QL, to enter cleaning pattern; And when the length of request queue is during higher than predetermined threshold QL, or the ratio that continuous free block accounts for all free blocks in non-solid-state disk 112 is while surpassing predetermined ratio threshold value C, can exit cleaning pattern.
In addition, also can enter cleaning pattern by the manual order of keeper, or in non-solid-state disk 112, there is no the notice of enough continuous free blocks and enter cleaning pattern in response to receiving.Certainly, can in each strategy, further stipulate to only have and when non-solid-state disk 112, check and regulate the ratio that free block continuous in row 110 accounts for all free blocks and when (it can be less than or equal to above-mentioned predetermined ratio threshold value C), just enter cleaning pattern lower than a predetermined ratio threshold value D; And the ratio that continuous free block occupies institute's free block in non-solid-state disk 112 is while being equal to or greater than predetermined ratio threshold value D, does not enter and knows pattern.Or, memory control device 120 enters after the particular event of knowing pattern in the triggering that receives specified specific policy regulation, further judge that whether ratio that in non-solid-state disk 112, continuous free block accounts for all free blocks is lower than predetermined ratio threshold value D, and only have when this is judged as YES, just enter cleaning pattern, and when this is judged as NO, abandon entering cleaning pattern.Whether memory control device 120 can be by inquiring about ratio that the record of all free blocks in above-mentioned non-solid-state disk 112 judges that in non-solid-state disk 112, continuous free block accounts for all free blocks lower than predetermined ratio threshold value.In addition, can also set a priority scheduling strategy, regulation is less than the read request that produced by upper layer application and the priority Q of write request by the priority P of clearing up the I/O request of mode producing, and dispatched according to priority by memory control device 120, like this, just can guarantee that the I/O that cleaning pattern is produced asks can not disturb the operation of application.
Above strategy is only example, rather than limitation of the present invention.In some embodiments of the invention, can adopt other strategies to stipulate how to enter and exit cleaning pattern.
According to an embodiment of the present, this memory control device 120 can (for example file system or operating system) provide a policy definition interface to upper strata, keeper can specify for entering and exit the strategy of cleaning pattern by this interface, and sets the required parameters such as threshold value of this strategy.Can be for different upper layer application specific, specify different strategies, and set different parameters.
For example, access Web server frequently for one, its average read-write was spaced apart for 1 second, and CPU is 200MBPS to the bandwidth of disk, can specify above-mentioned the first strategy, and sets T=0.5 second.That is to say, if while there is no read and write access in 0.5 second, enter cleaning pattern.Or, can specify above-mentioned the second strategy, and set BW=50MBPS.That is to say, when Disk bandwidth is used lower than 50MBPS, enter cleaning pattern.If current bandwidth is used higher than 50MBPS, exit cleaning pattern.Or, can specify above-mentioned the third strategy, and set QL=100.That is to say, the queue length of only having the request being produced by upper layer application in driver lower than 100 time, just enters cleaning pattern.Once queue length surpasses 100, exits cleaning pattern.In addition, can also specify above-mentioned priority scheduling strategy, and the priority of setting the I/O request of cleaning mode producing is 5, the priority of the I/O request that upper layer application produces is 15.Like this, the scheduling feature in memory control device 120 just can guarantee that the request of clearing up mode producing can not disturb the operation of application.In addition, also can set above-mentioned for exiting the predetermined ratio threshold value C=25% of cleaning pattern.That is to say, the continuous space that can use when system surpass free space 25% time, just can exit cleaning pattern.By specifying and setting as mentioned above strategy, can optimize for the frequent access of Web server and for example, for the read-write of small documents (the various little picture on webpage).
Again for example, for the file server of Yi Ge enterprise, its average read-write is spaced apart 20 seconds, can specify above-mentioned the first strategy, and sets T=10 second.That is to say, if while there is no read and write access in 10 seconds, enter cleaning pattern.Or, can specify above-mentioned the second strategy 2, and set BW=20MBPS.That is to say to only have when Disk bandwidth is used lower than 20MBPS, enter cleaning pattern.If current utilized bandwidth, higher than 20MBPS, exits cleaning pattern.Or, can specify above-mentioned the third strategy, and set QL=200, that is to say, the queue length of only having the request being produced by upper layer application in driver lower than 200 time, just enters cleaning pattern, once queue length surpasses 200, exits cleaning pattern.In addition, can also set above-mentioned priority scheduling strategy, and appointment is 8 by the priority of clearing up the I/O request of mode producing, the priority of the I/O request that upper layer application produces is 15, like this, the scheduling feature in memory control device 120 just can guarantee that the I/O that clears up mode producing asks can not disturb the operation of application.In addition, can also set above-mentioned for exiting the predetermined ratio threshold value C=50% of cleaning pattern.That is to say, the continuous space that can use when system surpass free space 50% time, just can exit cleaning pattern; Lower than 50% time, will proceed disk scanning, calculating and data block and move, so that available space reaches the more than 50% of free space continuously.By specifying and setting strategy as mentioned above, can optimize for file server access frequency compared with low and each large numbers of read-writes (such as hundreds of million, or even the file of several GB sizes is uploaded).
According to an embodiment of the present, described memory control device 120 is also configured to: in described cleaning pattern, judge whether to receive the particular event that cleaning pattern is exited in triggering, this particular event can be the particular event that cleaning pattern is exited in above described various tactful triggering of stipulating, the ratio that for example in non-solid-state disk, continuous free block accounts for all free blocks surpasses predetermined ratio threshold value C, etc.; And if described in be judged as NO, continue the cleaning operation to non-solid-state disk 112 and solid-state disk 111 as mentioned above; And if this is judged as YES, exit cleaning pattern.
More than described storage system according to an embodiment of the invention, wherein, memory control device 120 is write data to solid-state disk 111 and non-solid-state disk 112 in both in WriteMode; And memory control device 120 in reading mode only from solid-state disk 111 reading out datas.Yet this is not limitation of the present invention.In other embodiments of the invention, can not that the data of all write requests are write to solid-state disk 111 and non-solid-state disk 112 in both, but only the data with the write request that high availability requires are write to solid-state disk 111 and non-solid-state disk 112 in both, and the data without the write request that high availability requires are only write in non-solid-state disk 112; In addition, though can also after this will not have high availability to require the data that are accessed frequently to transfer to solid-state disk 111 from non-solid-state disk 112 in cleaning pattern.By this way, not only guarantee discriminatively high reliability and the high availability of the data of application-specific, and guaranteed discriminatively the data access speed of application-specific.
Therefore, according to another embodiment of the present invention, described memory control device 120 is configured to: in response to the read request that starts to process to a file, judge in solid-state disk 111, whether there is this document; In response to having this document in judgement solid-state disk 111, from solid-state disk 111, read this document; In response to not having this document in judgement solid-state disk 111, from non-solid-state disk 112, read this document; In response to starting, process the write request to a new file, judge whether this new file has high availability requirement; In response to this new file of judgement, there is high availability requirement, in solid-state disk 111 and non-solid-state disk 112, write new this document; And do not there is high availability requirement in response to this new file of judgement, only in non-solid-state disk 112, write this new file.Certainly, according to further embodiment of the present invention, described memory control device 120 also can be configured to: in response to judgement solid-state disk 111, have described file, and from solid-state disk 111, read after described file, can verify the integrality of described file data; In response to described, be verified, to upper layer application, return to described corresponding data; And do not pass through in response to described checking, from described non-solid-state disk 112, read described file, and return to this corresponding data to upper layer application
According to a further embodiment of the invention, described memory control device 120 is also configured to: in response to read a file from solid-state disk 111 or non-solid-state disk 112, upgrade the access frequency of this document; In cleaning pattern, in response to the access frequency that judges a file, higher than predetermined threshold, this document exists only in non-solid-state disk 112, and in solid-state disk 111, has enough free spaces, and described file is transferred in solid-state disk 111 from non-solid-state disk 112; And in cleaning pattern, in response to the access frequency of judgement one file, lower than predetermined threshold, and this document exists only in solid-state disk 111, and described file is transferred to non-solid-state disk 112 from solid-state disk 111.
Certainly, in these embodiment of the present invention, memory control device 120 is also configured to the operations of the data block for the non-solid-state disk of integral body and solid-state disk in the cleaning pattern as described above of carrying out.
The storage array of take above comprises that a solid-state disk 111 and non-solid-state disk 112 described storage system according to an embodiment of the invention as example, but this is not limitation of the present invention.
In other embodiment of the present invention, the storage array of storage system can comprise a plurality of solid-state disks 111 and a non-solid-state disk 112 (N+1 pattern), for example 4 solid-state disks 111 and 1 non-solid-state disk 112.This pattern is suitable for the less situation of solid-state disk 111 capacity.Because the cost of current solid-state disk 111 is still higher, capacity is less, and the cost of jumbo solid-state disk 111 is higher, for example, the cost of the solid-state disk 111 of 4 low capacities may be lower than the cost of 1 jumbo solid-state disk 111, and therefore this embodiment can reduce the cost of storage system.In such an embodiment, the memory control device 120 in storage system will be considered as a large solid-state disk 111 a plurality of solid-state disks 111, and carries out operations as described above.That is to say, in reading mode, memory control device 120 can read corresponding data by a solid-state disk from described a plurality of solid-state disks; In WriteMode, memory control device 120 can be write corresponding data in a solid-state disk in described a plurality of solid-state disks and described non-solid-state disk; And in cleaning pattern, memory control device 120 can be cleared up described a plurality of solid-state disks alternatively, the data block taking with assembling section.
In other embodiment of the present invention, the storage array of storage system can comprise a solid-state disk 111 and a plurality of non-solid-state disk 112 (1+M pattern), for example 1 solid-state disk 111 and 3 non-solid-state disks 112.This pattern is suitable for concurrent writing and counts and write the situation that bandwidth requirement is very high.When needs written document, can, according to the different reliability requirements of different files, the file with high reliability request be write at least two non-solid-state disks 112, and the file without high reliability request is only write to a non-solid-state disk 112.And in cleaning pattern, the frequency reading according to file, transfers to solid-state disk 111 by the file that reading frequency is higher from non-solid-state disk 112.Certainly, in cleaning pattern, also to scan and arrange to form how continuous free block to each non-solid-state disk 112, and alternatively, solid-state disk 111 be scanned to the data block taking with assembling section.
Therefore, according to these embodiment of the present invention, memory control device 120 is configured to: in response to starting, process the read request to a file, judge in described solid-state disk 111 whether have this document; In response to having this document in the described solid-state disk 111 of judgement, from described solid-state disk 111, read this document; In response to not having this document in the described solid-state disk 111 of judgement, from described a plurality of non-solid-state disks 112, read this document; In response to starting, process the write request to a new file, judge whether this new file has high availability requirement; In response to this new file of judgement, have high availability requirement, at least two dishes in a plurality of non-solid-state disks 112 are write this new file; And do not there is high availability requirement in response to this new file of judgement, only to a dish in described a plurality of non-solid-state disks 112, write this new file.According to these embodiment of the present invention, described memory control device 120 is also configured to: in response to read a file from solid-state disk 111 or a plurality of non-solid-state disk 112, upgrade the access frequency of this document; In cleaning pattern, in response to judgement one file access frequency higher than predetermined threshold, this document exists only in described a plurality of non-solid-state disk 112, and in solid-state disk 111, has enough free spaces, and described file is transferred in solid-state disk 111 from described a plurality of non-solid-state disks 112; And in cleaning pattern, in response to the access frequency of judgement one file, lower than predetermined threshold, and this document exists only in described solid-state disk 111, and described file is transferred to a plurality of non-solid-state disks 112 from solid-state disk 111.According to these embodiment of the present invention, described memory control device 120 is also configured to: in response to entering cleaning pattern, scan described a plurality of non-solid-state disk 112, and move data block wherein, to form how continuous free block.According to further embodiment of the present invention, described memory control device 120 is also configured to: in described cleaning pattern, scan and merge the data block that in described disk array 110, the part in solid-state disk 111 takies, forming how complete free block.
In some embodiment more of the present invention, the storage array of storage system can comprise a plurality of solid-state disks 111 and a plurality of non-solid-state disk 112 (N+M pattern), for example 2 solid-state disks 111 and 3 non-solid-state disks 112.In this pattern, can define the reliability requirement of different stage, and the higher file of the rank of reliability requirement is write in more dish.For example, for the storage system that comprises 2 solid-state disks 111 and 3 non-solid-state disks 112, can be defined as follows 5 grades of reliability requirements:
Rank 1 (default situation): new file is write to 1 non-solid-state disk 112, and reliability is minimum;
Rank 2: new file is write to 2 non-solid-state disks 112;
Rank 3: new file is write to 3 non-solid-state disks 112;
Rank 4: new file is write to 3 non-solid-state disks 112 and 1 solid-state disk 111;
Rank 5: new file is write to 3 non-solid-state disks 112 and 2 solid-state disks 111, and reliability is the highest.
When needs written document, can, according to the different reliability requirement ranks of different files, file be write in the non-solid-state disk 112 of varying number or solid-state disk 111.And in cleaning pattern, the frequency reading according to file, transfers to solid-state disk 111 by the file that reading frequency is higher from non-solid-state disk 112.Certainly, in cleaning pattern, also to scan and arrange to form how continuous free block to each non-solid-state disk 112, and alternatively, solid-state disk 111 be scanned to the data block taking with assembling section.
The rank of the reliability requirement shown in certainly, above is only exemplary.In an embodiment of the present invention, can adopt other reliability requirement rank.
Therefore, according to these embodiment of the present invention, memory control device 120 is configured to: in response to starting, process the read request to a file, judge in described a plurality of solid-state disk 111 whether have this document; In response to having this document in the described a plurality of solid-state disks 111 of judgement, from described a plurality of solid-state disks 111, read this document; In response to not having this document in the described a plurality of solid-state disks 111 of judgement, from described a plurality of non-solid-state disks 112, read this document; In response to starting, process the write request to a new file, judge the height of the availability requirement of this new file; According to the height of the availability requirement of this new file, at least one dish in described a plurality of solid-state disks 111 and a plurality of non-solid-state disk 112, write this new file.That is to say, for the higher new file of availability requirement, in dishes more in described a plurality of solid-state disks 111 and non-solid-state disk 112, write this new file.According to these embodiment of the present invention, described memory control device 120 is also configured to: in response to read a file from a plurality of solid-state disks 111 or a plurality of non-solid-state disk 112, upgrade the access frequency of this document; In cleaning pattern, in response to judgement one file access frequency higher than predetermined threshold, this document exists only in a plurality of non-solid-state disks 112, and in a plurality of solid-state disk 111, has enough free spaces, and described file is transferred in a plurality of solid-state disks 111 from a plurality of non-solid-state disks 112; And in cleaning pattern, in response to the access frequency of judgement one file, lower than predetermined threshold, and this document exists only in a plurality of solid-state disks 111, and described file is transferred to a plurality of non-solid-state disks 112 from a plurality of solid-state disks 111.According to these embodiment of the present invention, described memory control device 120 is also configured to: in response to entering cleaning pattern, scan described a plurality of non-solid-state disk 112, and move data block wherein, to form how continuous free block.According to further embodiment of the present invention, described memory control device 120 is also configured to: in described cleaning pattern, scan and merge the data block that in described disk array 110, the part in a plurality of solid-state disks 111 takies, forming how complete free block.
In other embodiment of the present invention, the storage array of storage system can comprise a plurality of solid-state disks 111 and a plurality of non-solid-state disk 112 of equal number, and each solid-state disk 111 forms a unit (can be described as N+N pattern) with each non-solid-state disk 112, for example 4 solid-state disks 111 and 4 non-solid-state disks 112 form 4 unit, and each unit comprises a solid-state disk 111 and a non-solid-state disk 112.In this pattern, memory control device 120 is carried out various operations as described above for each unit, and combines as required between unit, such as forming RAID0, RAID 1 or RAID 5 etc.
More than described the storage system according to various embodiments of the present invention, be to be noted that above description is only example, rather than limitation of the present invention.In other embodiments of the invention, that this storage system can have is not more, more not or different modules, and the memory control device of this storage system to can be configured to execution more, more not or different operations.
Describe according to an embodiment of the invention for realizing the method for storage system with reference to the accompanying drawings.Each step in the method is corresponding to the operations of above-described storage system according to an embodiment of the invention.For simplicity's sake, omit in the following description the part details repeating with above description, therefore, can obtain the more detailed understanding to the method with reference to above description.
According to embodiments of the invention, proposed a kind of for realizing the method for storage system, wherein said storage system comprises disk array, this disk array comprises at least one solid-state disk and at least one non-solid-state disk, the method comprises the following steps: in response to entering cleaning pattern, scan and move the data block at least one non-solid-state disk in described disk array, form how continuous free block.
Fig. 2 show according to one embodiment of the invention for realizing the specific implementation of the method for storage system.As shown in the figure, the method comprises the steps:
In step 201, carry out pattern arbitration, to determine, enter which in reading mode, WriteMode and cleaning pattern.
Can use several different methods to carry out pattern arbitration.For example, when receiving from the read request of upper layer application and starting to process this request, determine and enter reading mode; When receiving from the write request of upper layer application and starting to process request, determine and enter WriteMode; And in the time receiving for example, triggering by specific policy (, above a kind of in described various strategies) regulation and enter the particular event of cleaning pattern, enter cleaning pattern.For example, can, in response to one of following event, enter described cleaning pattern: in schedule time threshold value, do not receive read request and write request; Judgement Disk bandwidth takies lower than bandwidth and takies threshold value; Judgement is for receiving from the read request of upper layer application and the queue length of write request lower than predetermined queue length threshold; And receive the order that enters cleaning pattern.
In step 202, in response to entering cleaning pattern, scan and move the data block at least one non-solid-state disk in described disk array, form how continuous free block.
In step 203, upgrade the record of whole free blocks in described at least one non-solid-state disk.
In step 204, scan and merge the data block that in described disk array, the part at least one solid-state disk takies, form how complete free block.
In step 205, judge whether to receive the particular event that cleaning pattern is exited in triggering, this particular event can be the particular event that cleaning pattern is exited in above described various tactful triggering of stipulating, the ratio that for example in non-solid-state disk, continuous free block accounts for all free blocks surpasses predetermined ratio threshold value C, etc.
If this is judged as NO, return to step 202.If this is judged as YES, exit cleaning pattern.
According to an embodiment of the present, the method also comprises the steps:
In step 206, in response to entering reading mode, open the read channel of described at least one solid-state disk.
In step 207, from described at least one solid-state disk, read corresponding data.
In optional step 208, verify the integrality of described corresponding data.
In optional step 209, judge whether described checking is passed through.
If this is judged as YES, forwards step 210 to, otherwise forward step 211 to.
In step 210, to upper layer application, return to read corresponding data.
In step 211, open the read channel of described at least one non-solid-state disk.
In step 212, from described at least one non-solid-state disk, read corresponding data.
In step 213, read corresponding data is returned to described at least one solid-state disk.
In step 214, whether judgement recovers successful.
If this is judged as YES, forward step 210 to, to upper layer application, return to read corresponding data.Otherwise forward step 215 to.
In step 215, return to mistake, and alternatively by this error logging daily record or transmission error notification.
According to an embodiment of the present, the method also comprises the steps:
In response to entering WriteMode, in step 216, open the write access of described at least one solid-state disk, and the write access of opening described at least one non-solid-state disk in step 217.
In step 218, find continuous free block in described at least one non-solid-state disk.
In step 219, in described at least one the non-solid-state disk of judgement, whether there is enough continuous free blocks.
If this is judged as YES, forward step 220 to; If this is judged as NO, forward step 221 to.
In step 220, the continuous free block in described at least one non-solid-state disk is write corresponding data.
In step 221, in discrete free block of described at least one non-solid-state disk, write corresponding data.
In optional step 222, corresponding event to be charged to daily record and sent notice, described corresponding event for example for cannot find abundant continuous free block in non-solid-state disk.
In step 223, upgrade the record of whole free blocks in described at least one non-solid-state disk.
In step 224, to described at least one solid-state disk, write corresponding data.
Certainly, the foregoing is only according to an embodiment of the invention for realizing a kind of specific implementation of the method for storage system, rather than limitation of the present invention.Should can there is other specific implementation for realizing the method for storage system.
Referring now to Fig. 3 describe according to another kind of embodiment of the present invention for realizing the method for storage system, wherein, described storage system comprises disk array, this disk array comprises at least one solid-state disk and at least one non-solid-state disk.In below to the description of this embodiment, omitted and step identical in the embodiment shown in Fig. 2, for example carry out the step of pattern arbitration, in cleaning pattern and the identical step that non-solid-state disk and solid-state disk are arranged in embodiment in Fig. 2, and open step of passage etc. in reading mode and WriteMode.
As shown in the figure, the method comprises the steps:
In step 301, in reading mode, in response to starting to process the read request to a file, in described at least one solid-state disk of judgement, whether there is this document.
If this step is yes, forward step 302 to; If this is judged as NO, forward step 304 to.
In step 302, from described at least one solid-state disk, read this document.
In step 303, upgrade the access frequency of this document.
In step 304, to upper layer application, return to read file data.
In step 305, from described at least one non-solid-state disk, read this document, then forward step 303 to.
In step 306, in WriteMode, in response to starting, process the write request to a new file, judge the height of the availability requirement of this new file.
In step 307, according to the height of the availability requirement of this new file, at least one dish in described at least one solid-state disk and described at least one non-solid-state disk, write this new file.
In step 308, in described cleaning pattern, in response to judgement one file access frequency higher than predetermined threshold, this document exists only in described at least one non-solid-state disk, and in described at least one solid-state disk, there is enough free spaces, described file is transferred to from described at least one non-solid-state disk in described at least one solid-state disk.
In step 309, in response to the access frequency that judges a file, lower than predetermined threshold, this document exists only in described at least one solid-state disk, and described file is transferred to described at least one non-solid-state disk from described at least one solid-state disk.
The priority of the I/O request producing in described cleaning pattern according to an embodiment of the present, is lower than the priority of the read request from upper layer application and write request.
According to an embodiment of the present, described at least one non-solid-state disk is hard disk or erasable optical disk.
According to an embodiment of the present, described method is to realize by the device driver in operating system.
More than described method according to an embodiment of the invention, be to be noted that above description is only example, rather than limitation of the present invention.In other embodiments of the invention, that the method can have is more, still less or different steps, and the relation between each step can be from described different.Although for the purpose of narrating and illustrate conveniently, each step in the above each embodiment has specific order, but between some step, can there is different order or can executed in parallel, for example, step 204 also can be carried out or executed in parallel with it before step 203, and step 309 also can be carried out or executed in parallel with it before step 308.Some steps in the above each step are optional step, or can merge with other steps.For example, step 206,211,216,217 is optional step, or can merge into a step with its subsequent step separately; Step 208,209,213,214,215,222,204 is also optional step.
Following table shows according to the readwrite performance of storage system of the readwrite performance of the storage system of one embodiment of the invention and cost and several prior aries and the comparison of cost:
| Read handling capacity | Read the stand-by period | Write handling capacity | Write latency | Cost ($/GB) | |
| SSD RAID1 | 250MB/s | 0.03ms | 170MB/s | 0.3ms | 20 |
| SATA RAID1 | Be less than 40MB/s | 2.5ms | Be less than 20MB/s | 8ms | 0.12 |
| SATA RAID 5 (for example, 3+1) | Be less than 60MB/s | 1.7ms | 5MB/s | 30ms | 0.08 |
| SSD | High | Low | Low | Long | High |
| RAID5 | |||||
| Storage system of the present invention | 150MB/s | 0.03ms | 170MB/s | Be less than 1ms | 6-7 |
As seen from the above table, compare with the existing array being comprised of SATA, storage system has much better performance and the write performance read according to an embodiment of the invention, and cost only slightly strengthens.Compare with the existing array being comprised of solid-state disk, storage system has quite and balanced performance and the write performance read substantially according to an embodiment of the invention, and cost significantly reduces.Certainly, except above-mentioned advantage, storage system also has advantages of high data reliability and availability according to an embodiment of the invention.
Although specifically illustrated and illustrated the present invention with reference to preferred embodiment, those technician in this area should be understood that and can to it, carry out various changes and can not deviate from the spirit and scope of the present invention in form and details.More than describing is only the exemplary illustration to embodiments of the invention, is intended to make those skilled in the art to understand better the present invention, rather than limitation of the present invention.Scope of the present invention is only defined by the following claims.
Claims (22)
1. a storage system, comprising:
The disk array that comprises at least one solid-state disk and at least one non-solid-state disk; And
Memory control device, it is configured to: in response to entering cleaning pattern, scan and move the data block at least one non-solid-state disk in described disk array, form how continuous free block, described memory control device is further configured to data to write the continuous free block that cleaning pattern forms on non-solid-state disk
Wherein, described memory control device is also configured to:
In response to entering reading mode, from described at least one solid-state disk, read corresponding data; And
In response to entering WriteMode, to described at least one solid-state disk and described at least one non-solid-state disk, write corresponding data.
2. according to the storage system of claim 1, wherein, described memory control device is also configured to:
After described at least one solid-state disk reads corresponding data, verify the integrality of described corresponding data;
In response to described, be verified, to upper layer application, return to described corresponding data; And
In response to described checking, do not pass through, from described at least one non-solid-state disk, read corresponding data, this corresponding data is returned to described at least one solid-state disk, and return to this corresponding data to upper layer application.
3. according to the storage system of claim 1, wherein, to described at least one non-solid-state disk, write corresponding data and comprise:
Find continuous free block in described at least one non-solid-state disk; And
To continuous free block in described at least one non-solid-state disk, write described corresponding data.
4. according to the storage system of claim 1, wherein, described memory control device is also configured to:
In response to starting to process the read request to a file, in described at least one solid-state disk of judgement, whether there is this document;
In response to having this document in described at least one solid-state disk of judgement, from described at least one solid-state disk, read this document; And
In response to not having this document in described at least one solid-state disk of judgement, from described at least one non-solid-state disk, read this document.
5. according to the storage system of claim 4, wherein, described memory control device is also configured to:
In response to starting, process the write request to a new file, judge the height of the availability requirement of this new file; And
According to the height of the availability requirement of this new file, in at least one dish in described at least one solid-state disk and described at least one non-solid-state disk, write this new file, comprise for the higher new file of availability requirement, in more solid-state disks and non-solid-state disk, write this new file.
6. according to the storage system of claim 4, wherein, described memory control device is also configured to:
In response to read a file from described at least one solid-state disk or described at least one non-solid-state disk, upgrade the access frequency of this document;
In described cleaning pattern, in response to judgement this document access frequency higher than predetermined threshold, this document exists only in described at least one non-solid-state disk, and in described at least one solid-state disk, there is enough free spaces, this document is transferred to from described at least one non-solid-state disk in described at least one solid-state disk; And
In described cleaning pattern, in response to the access frequency of judgement this document, lower than predetermined threshold, and this document exists only in described at least one solid-state disk, and this document is transferred to described at least one non-solid-state disk from described at least one solid-state disk.
7. according to the storage system of claim 1, wherein, described memory control device is also configured to: in described cleaning pattern, scan and merge the data block that in described disk array, the part at least one solid-state disk takies, forming how complete free block.
8. according to the storage system of claim 1, wherein, described memory control device is also configured to:
In response to one of following event, enter described cleaning pattern:
In schedule time threshold value, do not receive read request and write request;
Judgement Disk bandwidth takies lower than bandwidth and takies threshold value;
Judgement is for receiving from the read request of upper layer application and the queue length of write request lower than predetermined queue length threshold; And
Receive the order that enters cleaning pattern.
9. storage system according to Claim 8, wherein, described memory control device is also configured to:
The ratio that accounts for all free blocks in response to continuous free block in the non-solid-state disk of judgement surpasses predetermined ratio threshold value, exits described cleaning pattern.
10. storage system according to Claim 8, wherein, the processing priority of the I/O request that described cleaning pattern produces is lower than the processing priority of the read request from upper layer application and write request.
11. according to the storage system of claim 1, and wherein, the function of described memory control device is to realize by the device driver in operating system.
12. 1 kinds for realizing the method for storage system, and wherein, described storage system comprises disk array, and this disk array comprises at least one solid-state disk and at least one non-solid-state disk, and the method comprises:
In response to entering cleaning pattern, scan and move the data block at least one non-solid-state disk in described disk array, form how continuous free block, so that data are written into the continuous free block that cleaning pattern forms on non-solid-state disk,
Described method also comprises:
In response to entering reading mode, from described at least one solid-state disk, read corresponding data; And
In response to entering WriteMode, to described at least one solid-state disk and described at least one non-solid-state disk, write corresponding data.
13. according to the method for claim 12, also comprises:
After described at least one solid-state disk reads corresponding data, verify the integrality of described corresponding data;
In response to described, be verified, to upper layer application, return to described corresponding data; And
In response to described checking, do not pass through, from described at least one non-solid-state disk, read corresponding data, this corresponding data is returned to described at least one solid-state disk, and return to this corresponding data to upper layer application.
14. according to the method for claim 12, wherein, writes corresponding data comprise to described at least one non-solid-state disk:
Find continuous free block in described at least one non-solid-state disk; And
To continuous free block in described at least one non-solid-state disk, write described corresponding data.
15. according to the method for claim 12, also comprises:
In response to starting to process the read request to a file, in described at least one solid-state disk of judgement, whether there is this document;
In response to having this document in described at least one solid-state disk of judgement, from described at least one solid-state disk, read this document; And
In response to not having this document in described at least one solid-state disk of judgement, from described at least one non-solid-state disk, read this document.
16. according to the method for claim 15, also comprises:
In response to starting, process the write request to a new file, judge the height of the availability requirement of this new file; And
According to the height of the availability requirement of this new file, in at least one dish in described at least one solid-state disk and described at least one non-solid-state disk, write this new file, comprise for the higher new file of availability requirement, in more solid-state disks and non-solid-state disk, write this new file.
17. according to the method for claim 15, also comprises:
In response to read a file from described at least one solid-state disk or described at least one non-solid-state disk, upgrade the access frequency of this document;
In described cleaning pattern, in response to judgement this document access frequency higher than predetermined threshold, this document exists only in described at least one non-solid-state disk, and in described at least one solid-state disk, there is enough free spaces, this document is transferred to from described at least one non-solid-state disk in described at least one solid-state disk; And
In described cleaning pattern, in response to the access frequency of judgement this document, lower than predetermined threshold, and this document exists only in described at least one solid-state disk, and this document is transferred to described at least one non-solid-state disk from described at least one solid-state disk.
18. according to the method for claim 12, also comprises:
In described cleaning pattern, scan and merge the data block that in described disk array, the part at least one solid-state disk takies, form how complete free block.
19. according to the method for claim 12, also comprises:
In response to one of following event, enter described cleaning pattern:
In schedule time threshold value, do not receive read request and write request;
Judgement Disk bandwidth takies lower than bandwidth and takies threshold value;
Judgement is for receiving from the read request of upper layer application and the queue length of write request lower than predetermined queue length threshold;
Receive the order that enters cleaning pattern.
20. according to the method for claim 19, also comprises:
The ratio that accounts for all free blocks in response to continuous free block in the non-solid-state disk of judgement surpasses predetermined ratio threshold value, exits described cleaning pattern.
21. according to the method for claim 19, and wherein, the processing priority of the I/O request producing in described cleaning pattern is lower than the processing priority of the read request from upper layer application and write request.
22. according to the method for claim 12, and wherein, described method is to realize by the device driver in operating system.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010213533.XA CN102298506B (en) | 2010-06-24 | 2010-06-24 | Storage system and method for implementing the same |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010213533.XA CN102298506B (en) | 2010-06-24 | 2010-06-24 | Storage system and method for implementing the same |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102298506A CN102298506A (en) | 2011-12-28 |
| CN102298506B true CN102298506B (en) | 2014-12-03 |
Family
ID=45358953
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201010213533.XA Active CN102298506B (en) | 2010-06-24 | 2010-06-24 | Storage system and method for implementing the same |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102298506B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102799395B (en) * | 2012-06-29 | 2015-07-22 | 华为技术有限公司 | Dynamic balancing processing method and device for disk spaces and disk system |
| CN102831088A (en) * | 2012-07-27 | 2012-12-19 | 国家超级计算深圳中心(深圳云计算中心) | Data migration method and device based on mixing memory |
| CN103970684A (en) * | 2013-02-04 | 2014-08-06 | 联想(北京)有限公司 | Data storage method and electronic device |
| CN107239225B (en) * | 2016-03-29 | 2020-05-26 | 群联电子股份有限公司 | Memory management method, memory storage device and memory control circuit unit |
| US10037149B2 (en) * | 2016-06-17 | 2018-07-31 | Seagate Technology Llc | Read cache management |
| WO2018103010A1 (en) * | 2016-12-07 | 2018-06-14 | 华为技术有限公司 | Memory management method and user terminal |
| CN107632780A (en) * | 2017-08-28 | 2018-01-26 | 深圳市云舒网络技术有限公司 | A kind of roll of strip implementation method and its storage architecture based on distributed memory system |
| CN108984130A (en) * | 2018-07-25 | 2018-12-11 | 广东浪潮大数据研究有限公司 | A kind of the caching read method and its device of distributed storage |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6467022B1 (en) * | 1998-04-16 | 2002-10-15 | International Business Machines Corporation | Extending adapter memory with solid state disks in JBOD and RAID environments |
| CN101238431A (en) * | 2005-08-03 | 2008-08-06 | 桑迪士克股份有限公司 | Mass data storage system |
| CN101673188A (en) * | 2008-09-09 | 2010-03-17 | 上海华虹Nec电子有限公司 | Data access method for solid state disk |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7921142B2 (en) * | 2007-01-08 | 2011-04-05 | International Business Machines Corporation | Method and apparatus for storing and accessing data records on solid state disks |
-
2010
- 2010-06-24 CN CN201010213533.XA patent/CN102298506B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6467022B1 (en) * | 1998-04-16 | 2002-10-15 | International Business Machines Corporation | Extending adapter memory with solid state disks in JBOD and RAID environments |
| CN101238431A (en) * | 2005-08-03 | 2008-08-06 | 桑迪士克股份有限公司 | Mass data storage system |
| CN101673188A (en) * | 2008-09-09 | 2010-03-17 | 上海华虹Nec电子有限公司 | Data access method for solid state disk |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102298506A (en) | 2011-12-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102298506B (en) | Storage system and method for implementing the same | |
| US9471240B2 (en) | Performing read and write operations with respect to at least one solid state disk and at least one non-solid state disk | |
| CN101038532B (en) | Data storage device and method thereof | |
| US8171207B2 (en) | Adaptive hybrid density memory storage device and control method thereof | |
| US8214586B2 (en) | Apparatus and method for mirroring data between nonvolatile memory and a hard disk drive | |
| US8775731B2 (en) | Write spike performance enhancement in hybrid storage systems | |
| CN100570546C (en) | Use the disk of first and second memory storages to quicken | |
| CN102576293B (en) | Data management in solid storage device and Bedding storage system | |
| CN101118477A (en) | Process for enhancing magnetic disc data accessing efficiency | |
| US7539816B2 (en) | Disk control device, disk control method | |
| Lin et al. | Hot random off-loading: A hybrid storage system with dynamic data migration | |
| JP6111575B2 (en) | Storage device, internal processing control method, and internal processing control program | |
| WO1997026600A1 (en) | Disk caching disk | |
| WO2011153478A2 (en) | Cache management and acceleration of storage media | |
| CN101727293A (en) | Method, device and system for setting solid state disk (SSD) memory | |
| CN101467122A (en) | Data progression disk locality optimization system and method | |
| CN103530237A (en) | Solid-state disc array garbage collecting method | |
| CN102637147A (en) | Storage system using solid state disk as computer write cache and corresponding management scheduling method | |
| CN100428193C (en) | Data preacquring method for use in data storage system | |
| Wang et al. | {BCW}:{Buffer-Controlled} Writes to {HDDs} for {SSD-HDD} Hybrid Storage Server | |
| US20080168226A1 (en) | Correction method for reading data of disk array system | |
| US10152242B1 (en) | Host based hints | |
| US11016889B1 (en) | Storage device with enhanced time to ready performance | |
| CN109375868B (en) | Data storage method, scheduling device, system, equipment and storage medium | |
| Oe et al. | On-the-fly automated storage tiering with caching and both proactive and observational migration |
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 |