JP2012247981A - Storage device and data transfer method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a storage device and a data transfer method that carry out data transfer in data duplication without lowering output performance of business while suppressing manufacturing costs.SOLUTION: The storage device includes a first storage part which stores data associated with business processing and a second storage part which stores duplicated data of the data. Further, the storage device includes a cache part having a first cache for duplication and a second cache for business processing; prediction means of predicting a transfer time of transfer data during duplication of the data; change means of dynamically changing the capacities of the first cache and second cache based upon the predicted transfer time predicted by the prediction means; and duplication means of duplicating the data using the first cache for duplication whose capacity is changed by the change means.

Description

  The present invention relates to a storage apparatus and a data transfer method for transferring data.

  In recent years, the storage capacity of disk devices has increased. As the capacity of the disk device increases, the amount of data handled by the storage device also increases. For example, the amount of replicated data when replicating data in the storage device is also increasing. Furthermore, as the amount of duplicate data increases, the time for transferring duplicate data also becomes longer.

  With such a long data replication time, a technique for reducing the time required for replication has been proposed. For example, in Patent Document 1 below, the updated data is arranged on the cache memory, and is used not only as a write-back cache to the disk device but also as an area for arranging snapshot difference data. A technique for speeding up data replication is disclosed.

  In general, a cache memory is mounted in the storage device in order to improve data input / output speed. Comparing the cache memory and the disk device, the cache memory is faster than the disk device in terms of data input / output, but the price per capacity is more expensive than the disk device.

JP 2010-122761 A

  As described above, the time required for data replication is prolonged because the amount of data handled by the storage device increases while the processing speed of the disk device is not sufficiently improved.

  Therefore, in order to solve such a problem, for example, it is conceivable to apply the technique described in Patent Document 1 to data transfer at the time of data replication. That is, by applying a technology for storing backup data on the cache memory, and further increasing the storage capacity of the cache memory, the data transfer at the time of replication is performed at a high speed, and the time required for the data transfer is shortened. It is possible.

  However, since the price per capacity of the cache memory is expensive, the more the cache memory is added, the higher the manufacturing cost of the storage device. Also, for example, when replicating data, it is necessary to store replication (complete replication) instead of differential snapshots, so it is necessary to install a large amount of cache memory, which significantly increases manufacturing costs. Become.

  That is, if the storage capacity of the cache memory is increased to increase the data transfer speed, the manufacturing cost will increase. If the storage cost is not increased without increasing the storage capacity of the cache memory, the data transfer will take a long time. . If the data transfer takes time, the load imposes the capacity of the cache memory that can be used for business, and the output performance related to the business processing of the storage apparatus decreases.

  The present invention has been made in view of the above circumstances, and its object is to provide a storage device capable of transferring data at the time of data duplication without reducing manufacturing costs and reducing the output performance of business. It is to provide a data transfer method.

  The present invention is a storage apparatus having a first storage unit that stores data relating to business processing and a second storage unit that stores data replication data, and includes a first cache for replication and a business processing A cache unit having a second cache; a predicting unit for predicting a transfer time of the data based on a data writing state at the time of data replication; and a first based on the predicted transfer time predicted by the predicting unit And changing means for dynamically changing the capacities of the cache and the second cache, and duplicating means for duplicating data using the first cache for duplication whose capacity is changed by the changing means. And

  Another aspect of the present invention is a data transfer method for a storage apparatus having a first storage unit that stores data related to business processing and a second storage unit that stores data replication data. The step of predicting the transfer time of the data based on the write status and the capacity of the first cache for replication and the second cache for business processing are dynamically changed based on the predicted transfer time predicted And a step of replicating data using the first cache for replication whose capacity has been changed.

  According to the present invention, it is possible to provide a storage apparatus and a data transfer method capable of performing data transfer at the time of data duplication without reducing manufacturing costs and reducing business output performance.

FIG. 2 is a diagram for schematically explaining the configuration of the storage apparatus according to the first embodiment of the present invention. It is a figure for demonstrating an example of the write information which concerns on the said 1st Embodiment. It is a figure for demonstrating an example of the writing status information which concerns on the 1st Embodiment. It is a figure for demonstrating the relationship between the writing condition which concerns on the 1st Embodiment, and the estimated transfer time of data. It is a flowchart which shows the update process of the write status information which concerns on the 1st Embodiment. It is a figure for demonstrating an example of the write condition information which concerns on the 2nd Embodiment of this invention. It is a figure for demonstrating the relationship between the writing condition which concerns on the said 2nd Embodiment, and the estimated transfer time of data. It is a flowchart which shows the update process of the write status information which concerns on the said 2nd Embodiment. It is a figure for demonstrating schematically the structure of the storage apparatus which concerns on the 3rd Embodiment of this invention. It is a figure for demonstrating the prediction parameter which concerns on the said 3rd Embodiment. It is a figure for demonstrating the prediction parameter which concerns on the 4th Embodiment of this invention. It is a figure for demonstrating the prediction parameter which concerns on the 4th embodiment. It is a figure for demonstrating an example for determining the prediction formula of the data transfer time which concerns on the 4th Embodiment.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

(First embodiment)
FIG. 1 is a block diagram schematically showing the configuration of the storage apparatus 100. The storage apparatus 100 includes an input / output unit 11, a cache unit 12, a first storage unit 13, a second storage unit 14, a duplication unit 15, a write information acquisition unit 16, and a write status storage unit 17. , A replication performance predicting means 18 and a cache switching means 19 which is a changing means.

  The input / output unit 11 is an interface that performs input / output to the storage apparatus 100. This interface is, for example, a Fiber Channel HBA or an Ethernet (registered trademark) NIC, but is not limited thereto.

  The cache unit 12 caches input / output to / from the first storage unit 13 or the second storage unit 14. The cache unit 12 is, for example, a flash memory or a DRAM. These are merely examples, and instead of these, for example, a disk device that is faster than the disk device used in the first storage unit 13 or the second storage unit 14 may be used.

  The cache unit 12 includes an I / O cache 12a and a duplication cache 12b. The I / O cache 12 a is a business processing cache, and caches data for input / output of the first storage means 13. The duplication cache 12b is a data duplication cache, and caches differential data in order to duplicate (backup) data from the first storage means 13 to the second storage means 14. In the present embodiment, a case where differential data used for a snapshot is cached will be described. However, when replication is performed, duplicate data is cached.

  The first storage unit 13 stores the latest data (business process data) input from the input / output unit 11. An example of the first storage means 13 is a hard disk drive.

  The second storage unit 14 stores a backup of data at a certain point of time stored in the first storage unit 13, that is, differential data. An example of the second storage means 14 is a hard disk drive.

  The duplication unit 15 transfers the data difference in order to back up the data in the first storage unit 13 to the second storage unit 14. An example of the duplicating unit 15 is a memory that holds a table in which dirty blocks of the first storage unit 13 are recorded, a CPU or DMA controller that controls transfer, and a buffer RAM. These are merely examples and the present invention is not limited to these.

  The write information acquisition unit 16 receives write information as an input when a write request to the storage apparatus 100 from a host device (not shown) is generated, and receives data stored in the write status storage unit 17. Update. FIG. 2 is a diagram for explaining an example of the write information input to the write information acquisition means 16. As shown in the figure, the write information has a sector number T101 and a sector number T102. The sector number T101 indicates the number of the first sector of data write in the data write request to the storage apparatus 100, and the sector number T102 indicates the number of sectors of the data.

  The writing status storage unit 17 stores writing status information used for predicting the transfer time of difference data. This writing status information is updated by the writing information acquisition means 16. FIG. 3 is a diagram for explaining an example of the writing status information stored in the writing status storage unit 17. As shown in the figure, the write status information has a write count T201. The write count T201 is the number of data write requests to the storage apparatus 100. In other words, assuming that the area where data writing occurs is uniform, the amount of dirty data, that is, the copy size of the difference data is proportional to the number of write requests, so the transfer of the difference data from the write count T201. The amount can be predicted.

Based on the write status information stored in the write status storage unit 17, the replication performance prediction unit 18 predicts and outputs a replication time for copying the data difference, that is, a transfer time of the differential data. With reference to FIG. 4, an example of the prediction of the data transfer time during replication by the replication performance prediction means 18 will be described in detail. FIG. 4 is a diagram for explaining the relationship between the number of writes n stored as the write status information and the estimated transfer time T of the difference data. The replication performance predicting means 18 sets the difference data transfer time T to the number N of times of writing stored as the write status information.
T = T (N) = KN (1)
Predict by (K: positive constant). Assuming that the data transfer time is proportional to the transfer amount, the transfer time at the time of data duplication can be predicted by the above equation (1).

The cache switching unit 19 dynamically changes the capacity of the I / O cache 12a and the duplication cache 12b according to the duplication time (predicted transfer time) of the difference data predicted by the duplication performance prediction unit 18. In other words, the cache switching unit 19 determines a storage capacity to be used as the replication cache 12b in the cache unit 12. For example, if the predicted transfer time T of the difference exceeds a certain threshold value T0, the cache switching unit 19 sets the capacity C of the replication cache 12b to C = C0 (T−T0) (2)
(C0: positive constant) This determination method is an example, and the present invention is not limited to this.

  In the present embodiment, when the amount of difference data is sufficiently small, the duplication cache 12b is omitted to eliminate the influence on the business, the amount of difference data is large, and the time required for duplication is likely to exceed T0. Allocates the storage capacity of the cache unit 12 so as not to exceed the threshold T0 as much as possible. Thereby, the time required for data replication can be suppressed.

  Next, the operation of the storage apparatus 100 will be described.

  When a write request is issued from the input / output unit 11 to the storage device 100, the storage device 100 writes data to the I / O cache 12a and the first storage unit 13, and at the same time, the write information acquisition unit 16 The writing status information stored in the recording status storage means 17 is updated.

  With reference to FIG. 5, an example of the update process of the write status information by the write information acquisition unit 16 will be described in detail. The write information acquisition unit 16 acquires write information (write generation request) for the storage apparatus 100 (step S101). Next, the writing information acquisition unit 16 reads the writing count stored as the writing status from the writing status storage unit 17 and increments it (step S102).

  Next, at the timing of transferring differential data for data replication, the replication performance prediction means 18 reads the write status information from the write status storage means 17 and calculates the data difference. More specifically, the replication performance prediction unit 18 first reads the write status information from the write status storage unit 17. As described above, the write count is stored as the write status information, and the replication performance predicting means 18 sets the value of the read write count to n0 and substitutes this n0 into the above equation (1). For example, the point on the graph corresponding to the substituted result corresponds to P301 (see FIG. 4), and Kn0 is obtained as the predicted transfer time corresponding to P301. Thus, after obtaining the predicted transfer time Kn0, the replication performance predicting means 18 resets the stored contents of the write status storing means 17.

  On the other hand, the cache switching unit 19 determines the storage capacity to be used as the duplication cache 12b according to the predicted transfer time obtained from the duplication performance prediction unit 18.

  Next, the duplication unit 15 transfers the difference data from the first storage unit 13 to the second storage unit 14 using the duplication cache 12b.

  According to the storage apparatus 100 configured as described above, it is necessary to transfer the data at the time of duplication by predicting the transfer time of the differential data using the write status information stored in the write status storage unit 17. Time can be predicted. The storage apparatus 100 can appropriately allocate the storage capacity of the cache unit 12 as the duplication cache 12b by predicting the required time.

  Therefore, the storage apparatus 100 can appropriately allocate the storage capacity of the cache unit 12 according to the predicted transfer time without increasing the storage capacity of the cache unit 12 by adding a plurality of cache memories. The data transfer at the time of data duplication can be performed appropriately without reducing the manufacturing cost of 100 and reducing the input / output performance of the business.

(Second Embodiment)
Next, a second embodiment will be described. The second embodiment is different from the first embodiment described above in the writing status information and the process of predicting the transfer time using the writing status information. Therefore, these will be described in detail below. The same components as those of the first embodiment are denoted by the same reference numerals, and detailed description thereof is omitted.

  First, an example of the write status information stored in the write status storage means 17 will be described in detail. The write status information includes a head sector T211 and a tail sector T212 shown in FIG. The head sector T211 is the minimum sector number of a sector where a write request has occurred (that is, dirty). The last sector T212 is the maximum sector number of a sector where a write request has occurred. Assuming that data writing occurs in one continuous area, the amount of dirty data (write width) is the difference between the tail sector T212 and the head sector T211.

Next, with reference to FIG. 7, an example of prediction of data transfer time during replication by the replication performance prediction unit 18 will be described in detail. FIG. 7 is a diagram for explaining the relationship between the writing status and the predicted transfer time. As shown in FIG. 7, when the last sector T212 is stored as “Stail” and the first sector T211 is stored as “Shead” as the writing status, the duplication performance predicting means 18 uses S = Stail−Shead as the difference data transfer time T. T = T (S) (3)
To predict.
Assuming that a cache of size Sthreshold (threshold) can be used and the transfer time of both the cache and the disk device is proportional to the transfer amount, the data transfer time at the time of duplication is represented by T = T (S) as shown in FIG. Predictable.

  Next, the operation of the storage apparatus 100 will be described.

  Each time a write request is issued to the storage apparatus 100, the write information acquisition unit 16 updates the write status stored in the write status storage unit 17. With reference to FIG. 8, an example of the update process of the write status information by the write information acquisition unit 16 will be described in detail.

  The write information acquisition unit 16 acquires write information (generation of write request) for the storage apparatus 100 (step S201). Next, the writing information acquisition unit 16 reads the head sector T211 stored in the writing status storage unit 17, and compares the head sector T211 with the sector number T101 of the writing information (step S202).

  When the head sector T211 is less than the sector number T101 (step S201: YES), the write information acquisition unit 16 updates the head sector T211 to the sector number T101 (step S203).

  When the leading sector T211 is updated to the sector number T101 as described above (S203), or when the leading sector T211 is equal to or greater than the sector number T101 (step S201: NO), the writing information acquisition unit 16 determines whether the writing status acquisition unit 16 The end sector T212 stored in the storage means 17 is read, and the sum of the sector number T101 and the sector number T102 of the write information is compared with the end sector T212 (step S204).

  When it is determined that the tail sector T212 is small (step S204: YES), the writing information acquisition unit 16 updates the tail sector T212 to the sum of the sector number T101 and the sector number T102 (S205).

  When updated to the sum of the sector number T101 and the sector number T102 (S205), or when it is determined that the tail sector T212 is small (S204: NO), the write information acquisition unit 16 ends this process.

  Next, the prediction of the transfer time of the difference data by the replication performance prediction unit 18 will be described. In the second embodiment, the graph of FIG. 7 is used instead of the graph of FIG. 4 described above.

  At the timing when the difference data is transferred for data replication, the replication performance prediction unit 18 reads the write status information from the write status storage unit 17 and calculates the difference of the data.

  First, the replication performance prediction unit 18 reads the write status information from the write status storage unit 17. A write width S is obtained based on the end sector T212 and the start sector T211 included in the write status information.

  As shown in FIG. 7 described above, if S = s1, the predicted transfer time T = K1s1 is obtained corresponding to the point P311 on the graph. If S = s2, the predicted transfer time T = K2s2 + C2 is obtained corresponding to the point P312 on the graph.

  Even if the storage apparatus 100 is configured as in the second embodiment, the same effects as in the first embodiment can be obtained.

(Third embodiment)
Next, a third embodiment will be described. The third embodiment is different from the first embodiment described above in that the feedback unit 20 and the prediction parameter storage unit 21 are provided. Therefore, these will be described in detail below. The same components as those of the first embodiment are denoted by the same reference numerals, and detailed description thereof is omitted.

  As illustrated in FIG. 9, the storage apparatus 100 includes a feedback unit 20 and a prediction parameter storage unit 21 in addition to the configuration described in the first embodiment. In FIG. 9, illustration of a configuration not directly related to the feedback unit 20 and the prediction parameter storage unit 21 is omitted.

  The prediction parameter storage unit 21 stores a prediction parameter used for prediction of data transfer time during replication by the replication performance prediction unit 18. FIG. 10 is a diagram for explaining an example of a prediction parameter stored in the prediction parameter storage unit 21. The prediction parameter shown in FIG. 10 is the predicted transfer rate T301. This predicted transfer rate T301 is v / N when the number of sectors v transferred per unit time and the average number of sectors N per write request are v / N. Is required.

  The replication performance predicting means 18 predicts the transfer time for transferring the data difference based on the parameter stored in the predicted parameter storage means 21 and the write status information stored in the write status storage means 17. Output.

  Next, prediction of data transfer time during replication by the replication performance prediction unit 18 will be described in detail. When the predicted transfer rate T301 is V = 1000 and the write count T201 is N = 100, the replication performance predicting means 18 calculates the predicted transfer time T as T = N / V = 100/1000 = 0.100.

  The feedback unit 20 compares the predicted transfer time T predicted by the replication performance prediction unit 18 with the actual transfer time measured in the actual data transfer, and uses the comparison result as a prediction parameter stored in the prediction parameter storage unit 21. provide feedback. For example, the feedback means 20 uses the value V of the predicted transfer rate T301 reflecting the feedback as V = N by using the value N of the number of times of writing T201 and the actual transfer time Treal actually applied (hereinafter referred to as Tr). The value stored in the prediction parameter storage unit 21 is updated using the calculated value V. Further, for example, when the predicted transfer time T is longer than the actual transfer time Tr, the feedback unit 20 increases the predicted transfer rate T301 stored in the prediction parameter storage unit 21 by a certain amount, so that the predicted transfer time T is greater than the actual transfer time Tt. If it is shorter, the predicted transfer rate T301 may be decreased by a certain amount.

  Even if the storage apparatus 100 is configured as in the third embodiment, the same effects as in the first embodiment can be obtained.

  Furthermore, the storage apparatus 100 includes the feedback unit 20, so that even if the tendency of the write request is not known in advance, the trend of the write request is updated by updating the predicted transfer rate T 301 that is a prediction parameter based on the actual transfer time Tr. It is possible to predict the transfer time according to.

  Further, the feedback of the feedback means 20 can be performed by other methods. For example, the variable t (0 <t <1) is stored in the prediction parameter storage unit 21, and the replication performance prediction unit 18 uses T = tT1 (N) for a plurality of prediction formulas T1 (N) and T2 (N). The replication time is predicted by + (1-t) T2 (N). The feedback means 20 compares | X−T1 (N) | and | X−T2 (N) | with respect to the actual result X, and if the result X is close to T1 (N), the variable t is increased. , The variable t may be reduced if it is close to T2 (N).

(Fourth embodiment)
Next, a fourth embodiment will be described. The fourth embodiment is different from the third embodiment described above in the feedback method of the feedback means and the prediction parameters stored in the prediction parameter storage means. Therefore, these will be described in detail below. The same components as those in the third embodiment are denoted by the same reference numerals, and detailed description thereof is omitted.

  The prediction parameter storage unit 21 stores a prediction parameter used for prediction of data transfer time during replication by the replication performance prediction unit 18. 11 and 12 are diagrams for explaining an example of a prediction parameter stored in the prediction parameter storage unit 21. FIG. As shown in FIG. 11, the prediction parameter has a time T401, a write count T402, a transfer time T403, and a primary coefficient T411 and a constant term T412 as shown in FIG.

  Time T401 is the time when the snapshot was created. The write count T402 is the number of write requests to the storage apparatus 100 issued between the snapshot creation shown at time T401 and the snapshot creation immediately before. In the example shown in FIG. 11, a case is shown in which there are 100 write requests before the snapshot creation at time 120 and 153 write requests between the snapshot creations at time 120 and 240. The transfer time T403 is a transfer time of difference data for creating a snapshot. In the example shown in FIG. 11, it is shown that 11.0 time was required for transferring the snapshot difference data at time 120, and 12.2 time was required for the transfer of snapshot difference data at time 240. It shows what was needed. The primary coefficient T411 is a primary coefficient a1 when the transfer time T is approximated by a linear expression T = T (N) = a1N + a0 of the number of times of writing N. The constant term T412 is a constant term a0 of an approximate expression for transfer time. In the example shown in FIGS. 11 and 12, the transfer time is approximated as T (N) = 0.05N + 5.00 with respect to the write count N.

  In addition, the prediction parameter memorize | stored in the above-mentioned prediction parameter memory | storage means 21 is an example, and is not restricted to this. For example, a past writing status information history may be used as a prediction parameter, or a statistic representing a past writing status information history may be used.

  The replication performance predicting means 18 predicts the time for transferring the data difference based on the prediction parameter stored in the prediction parameter storage means 21 and the write status information stored in the write status storage means 17, and outputs it. To do.

  Next, an example of data transfer time prediction by the replication performance prediction unit 18 will be described in detail. As shown in FIG. 12, the primary coefficient T411 is 0.05 and the constant term T412 is 5.00. For example, as shown in FIG. When the write count N = 100 is stored, the replication performance predicting means 18 calculates the predicted transfer time T as T = 0.05 × 100 + 5.00 = 10.0.

  This method for calculating the prediction of the data transfer time is an example, and is not limited to this. For example, it may be approximated not by a linear expression of past writing status information as described above but by a polynomial or other function.

  The feedback means 20 feeds back the actual transfer time measured in the actual data transfer to the transfer time prediction formula. An example of feedback by the feedback unit 20 will be described in detail with reference to FIGS. 11 to 13. FIG. 13 is a diagram for explaining an example in which the transfer time is approximated by a linear expression of the number of times of writing and the prediction expression is determined by the least square method.

  In the example illustrated in FIG. 13, as illustrated in FIGS. 11 and 12, the predicted transfer time is calculated by using the least square method based on the write status and the transfer time history stored as the prediction parameters. Is approximated as a linear expression.

  In FIG. 13, P501, P502, and P503 indicate points where a set of past write counts and transfer time histories stored in the prediction parameter storage unit 21 illustrated in FIG. 11 is plotted. P504 is a point in which a set (128, 11.2) of the actual write count and transfer time is plotted. In the example illustrated in FIG. 13, by applying the least square method to P501, P502, P503, and P504, the transfer time prediction formula T (N) is calculated, and T (N) = 0.041N + 6.37 is obtained. .

  The feedback unit 20 outputs the actual backup creation time, the number of writes, the transfer time, the primary coefficient 0.041 of the prediction formula, and the constant term 6.37, and the time T401 and the number of writes T402 of the prediction parameter storage unit 21. The transfer time T403, the primary coefficient T411, and the constant term T412 are updated. Note that the entire history of the writing status information may be saved, or only a part of the history may be saved.

  Note that the feedback method of the feedback unit 20 is not limited to the estimation of the prediction formula by the least square method described above, and other regression analysis methods may be used. Further, the feedback means 20 may instruct the prediction to be performed using a prescribed prediction formula when the history of the writing status information is not sufficient.

  In addition, the storage apparatus 100 described in each of the above embodiments can be used, for example, for scheduling (deciding a time zone for performing backup) when data is backed up from the first storage unit to the second storage unit. It is. More specifically, based on the write status information stored in the write status storage means 17, for example, the data write status of a certain day (or a predetermined period) is verified, and the time when the data is least written A band may be selected and backup may be performed during this time period. By adjusting the time zone for transferring the data at the time of backup in this way, the storage apparatus 100 can transfer the data at the time of backup during a time zone in which the I / O cache is rarely used in business. For this reason, the total used capacity of the cache memory in a predetermined time unit can be reduced, the capacity of the cache unit 12 can be reduced, and the manufacturing cost can be further reduced.

  In addition, this invention is not limited to the above-mentioned embodiment, In the implementation, various deformation | transformation are possible.

  A part or all of the above embodiment can be described as in the following supplementary notes, but is not limited thereto.

(Appendix 1)
A storage device having a first storage unit that stores data related to business processing and a second storage unit that stores duplicate data of the data,
A cache unit having a first cache for duplication and a second cache for business processing;
A predicting means for predicting a transfer time of the data based on a writing situation of the data when replicating the data;
Changing means for dynamically changing the capacities of the first cache and the second cache based on the predicted transfer time predicted by the prediction means;
Replication means for replicating the data using a first cache for replication whose capacity has been changed by the changing means;
A storage apparatus comprising:

(Appendix 2)
Write number acquisition means for acquiring the number of data write requests from the host device;
Write number storage means for storing the acquired number of writes,
The storage device according to appendix 1, wherein the write status is a write count stored in the count storage means.

(Appendix 3)
When there is a data write request from the host device, sector information acquisition means for acquiring sector information including the head sector and the tail sector storing the data,
Sector information storage means for storing the sector information,
The storage apparatus according to appendix 1, wherein the write status is a write width of a head sector and a tail sector included in sector information stored in the sector information storage means.

(Appendix 4)
Prediction parameter storage means for storing a prediction parameter for predicting the transfer time;
A feedback means for comparing the predicted transfer time predicted using the prediction parameter storage means and the actual transfer time, and for feeding back the result of the comparison to a prediction parameter stored in the prediction parameter storage means,
The storage apparatus according to claim 1, wherein the predictive transfer unit predicts a transfer time using a prediction parameter stored in the prediction parameter storage unit.

(Appendix 5)
The storage apparatus according to any one of appendices 1 to 4, wherein the data replication is a snapshot replication, and the transfer data is differential data.

(Appendix 6)
The storage apparatus according to any one of appendices 1 to 4, wherein the data replication is replication by replication, and the transfer data is replication data.

(Appendix 7)
A data transfer method for a storage apparatus having a first storage unit for storing data relating to business processing and a second storage unit for storing duplicate data of the data,
Predicting the transfer time of the data based on the data writing status when replicating the data;
Dynamically changing the capacity of the first cache for replication and the second cache for business processing based on the predicted transfer time predicted;
Replicating the data using a first cache for replication with the capacity changed;
A data transfer method characterized by comprising:

  The present invention can be widely applied to a storage apparatus capable of performing data replication and a data transfer method at the time of data replication.

DESCRIPTION OF SYMBOLS 11 ... Input / output means 12 ... Cache part 12a ... I / O cache 12b ... Duplicating cache 13 ... First storage means 14 ... Second storage means 15 ... Duplicating means 16... Write information obtaining means 17... Writing status storage means 18... Replication performance prediction means 19... Cache switching means 20. ... Storage devices

Claims (7)

A storage device having a first storage unit that stores data related to business processing and a second storage unit that stores duplicate data of the data,
A cache unit having a first cache for duplication and a second cache for business processing;
A predicting means for predicting the transfer time of the data based on the data writing status at the time of data replication;
Changing means for dynamically changing the capacities of the first cache and the second cache based on the predicted transfer time predicted by the prediction means;
Replication means for replicating the data using a first cache for replication whose capacity has been changed by the changing means;
A storage apparatus comprising: Write number acquisition means for acquiring the number of data write requests from the host device;
Write number storage means for storing the acquired number of writes,
The storage apparatus according to claim 1, wherein the write status is a write count stored in the count storage unit. When there is a data write request from the host device, sector information acquisition means for acquiring sector information including the head sector and the tail sector storing the data,
Sector information storage means for storing the sector information,
2. The storage apparatus according to claim 1, wherein the write status is a write width of a head sector and a tail sector included in sector information stored in the sector information storage unit. Prediction parameter storage means for storing a prediction parameter for predicting the transfer time;
A feedback means for comparing the predicted transfer time predicted using the prediction parameter storage means and the actual transfer time, and for feeding back the result of the comparison to a prediction parameter stored in the prediction parameter storage means,
The storage apparatus according to claim 1, wherein the prediction transfer unit predicts a transfer time using a prediction parameter stored in the prediction parameter storage unit.   The storage apparatus according to claim 1, wherein the data replication is a snapshot replication, and the transfer data is differential data.   The storage apparatus according to claim 1, wherein the data replication is replication by replication, and the transfer data is replication data. A data transfer method for a storage apparatus having a first storage unit for storing data relating to business processing and a second storage unit for storing duplicate data of the data,
Predicting the transfer time of the data based on the data writing status when replicating the data;
Dynamically changing the capacity of the first cache for replication and the second cache for business processing based on the predicted transfer time predicted;
Replicating the data using a first cache for replication with the capacity changed;
A data transfer method characterized by comprising:

Images