JPH11219312A - Data cache method and data access method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To designate the precision of data and to efficiently access a large size data object. SOLUTION: A data access request containing a requested precision of data is issued from a client 1. A cache device 2 judges whether the designated data exists in a cache or not. When the designated data exists, compression attribute of the cached data is referred to. When it satisfies the precision designated in the data access request, the compressed data is returned. When it does to exist in the cache or precision is not satisfied, data received from a data supply source is compressed in a range satisfying the designated precision, it is registered in the cache and is returned to the client 1.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a data object caching method and a data access method in an information system.

[0002]

2. Description of the Related Art With the development of the information society and the advance of computer technology, not only text information but also various multimedia information such as voice, image, and moving image have been exchanged within a system. These pieces of information are different from conventional text information in that the data size per object is very large. Also, with regard to text information, a database has been constructed in which the entire text of a dissertation or magazine is treated as one object, and the amount of data handled by the entire information system has become extremely large. In addition, companies are rapidly introducing data warehouse systems that store and analyze huge amounts of raw data, and the tables held by database systems are increasing to several terabytes in large systems. ing.

In order to share such a vast amount of information on the Internet, a technology for efficiently accessing such information is essential. Worl especially on the Internet
In a system such as dWide Web (hereinafter, WWW), there is a problem that a load on a network increases with an increase in data amount, and a time required for data transfer increases.

[0004] As a technique for solving such a problem, there is a caching server. The caching server is a technology that has emerged with the development of the WWW system on the Internet, for example, "Michael Baentsc
h, et.al., “World Wide WebCaching: The Application
-Level View of the Internet ”, IEEE CommunicationsM
agazine, June 1997. "and its history. In this, a technique for maintaining cache coherency and a technique for load balancing of a caching server are described.

[0005] When a huge amount of data such as video information is stored in the caching server, there is a high possibility that other useful data is expelled. In order to avoid this, as a mechanism for controlling the size of data stored in the caching server, there is a method of compressing and caching data without caching data having a certain size or more as it is. This conventional method is described in "M. Oguchi and K. On
o, “A Proposal for a World-Wide Web Caching Proxy
Mechanism ”, In Proc. Of Third InternationalSymposi
um on Interworking (INTERWOKING'96), pp.531-540, O
ctober 1996. "

[0006] The compression described here is a method used when digital or analog data is made smaller than its original size (compression). Compression includes lossless compression (Lossless), which does not allow distortion to restore the original data when decompressing data, and lossy compression (Loss), which allows distortion where the amount of information may be less than the original data.
y), compression of character information, programs, etc., generally uses a compression method that does not tolerate distortion to reduce the data size, and allows compression of information to the extent that moving images and audio data cannot be recognized. Method is used.

There is a compression algorithm such as a run-length method and a Huffman method for the lossless compression which does not allow the distortion. In addition, JPEG used for image (still image / moving image) compression,
MPEG is a compression method that incorporates both lossless compression that does not allow distortion and irreversible compression that allows distortion. If a designation is made that allows distortion, a particularly high compression ratio can be obtained.

With irreversible compression, the original data cannot be reproduced, so that the data quality deteriorates. However, for multimedia data such as video and images, the user does not always request data of the same quality as the original data. . Therefore, in this paper, in response to an access request from a client, if the client hits a compressed cache, the client is asked if the quality based on this compression is acceptable, and if the client allows the quality, A protocol for transferring the compressed data is disclosed.

FIG. 8 is a diagram for explaining a protocol using a compression cache described in this paper.

FIG. 8A is a diagram showing the flow of processing when a client request hits a compression cache of a cache device (cache server) and the client allows deterioration of data quality due to the compression. First,
The client 1 requests data access to the cache device 2 (801). Next, the cache device:
The cache 81 of the original data and the cache 82 of the compressed data held in the apparatus are checked to see if there is any corresponding data. Here, there is no data in the cache of the original data (cache miss),
If there is data only in the compressed cache (cache hit), the cache device returns the accuracy of the data stored in the compressed cache to the client (802). The client determines whether or not the data accuracy notified from the cache device is sufficient (83), and if so, notifies the cache device of the determination (803). Upon receiving the acknowledgment from the client, the cache device retrieves the compressed data from the cache and transfers it (804).

FIG. 8B is a diagram showing the flow of processing when a client request hits the compression cache of the cache server, but deterioration of data quality due to the compression cannot be tolerated. First, the client 1 requests data access to the cache device 2 (80
5). Next, the cache device checks the original data cache (not shown) and the compressed data cache (not shown). If the cache hits only in the compression cache, the cache device responds to the client with the accuracy of the data (806). Then, the client determines the accuracy of the data quality (84), and returns a negative response indicating that the data quality is unacceptable to the cache device 2 (8).
07). Then, the cache device issues an access request to the original data of the data supply source 5 (80
8) The original data 85 is transferred to the client (809).

FIG. 8C shows that the request from the client is
FIG. 14 is a diagram showing a flow of processing when a mistake is also made in the compression cache of the cache server 2; First, the client 1 requests data access to the cache server 2 (810). Next, the cache server checks the cache of the original data and the cache of the compressed data held in the device. Here, if you miss both caches,
The cache server 2 issues an access request to the original data 85 stored in the data source 5 (8
11). The data source 5 transfers the original data 85 to the client 1 via the cache server (81
Along with 2), the cache server 2 registers the transferred original data 85 in the cache. If the size of the transferred original data is too large to be stored in the original cache, the data is compressed into data of a size determined according to the state of the system (86) and registered in the compressed cache. (87).

[0013]

In the above-described prior art for realizing a compression cache, in order to maintain consistency of an access interface between a client and a cache server, when an access request hits a compressed file, Only the protocol that asks the client whether the data accuracy is sufficient was adopted. However, such a protocol requires two round trips between the client and the cache server when hitting a compressed file. Therefore, there is a problem that a load on a network connecting the client and the server increases.

Further, in order to implement a protocol for confirming the accuracy of data, it is necessary to change the communication protocol on both the client side and the server side, and further, the client side has a judgment mechanism for approving the accuracy. Had to be equipped. Since the number of clients is huge compared to the number of cache servers,
Making such changes has been very difficult.

In FIG. 8, for convenience, the client 1 and the data supply source 5 are illustrated as being connected via one cache device 2, but in reality, FIG.
As shown in (1), a plurality of cache devices (201 to 203) are interposed between the client 1 and the data supply source 5. Therefore, similarly to the case where the accuracy of the compression cache is rejected by the client (901), when the cache device makes a data access request to the next cache device, the rejection of the data accuracy is repeatedly performed by the same protocol. (902-904), there is a problem that the latency of data access is remarkably reduced.

Further, in the conventional data access method,
When accessing data from a data source, the original data is always transferred, so that a large amount of data flows over the network (905 to 908) even if the user does not require such high quality data. There is a point.

An object of the present invention is to provide a method of caching large-scale data, compressing and storing the cached data according to a user's request or the state of a cache device, and returning the cached data to the client in response to the request. Aim. It is another object of the present invention to provide a method for compressing and transferring data at a data source in order to reduce network load. Furthermore, the present invention provides a mechanism for adding the accuracy of the data as header information when a data access request is made from a client, so that the cache server or the information supply source can efficiently provide the data requested by the user. The purpose is to provide an access method.

[0018]

In order to achieve the above object, according to the present invention, when a client issues a data access request, not only the data ID to be accessed but also the required accuracy of the data is added at the same time. If you want to obtain the original data, do not specify the required accuracy of the data. The server that has received the access request with the required precision searches the compressed cache area in addition to the normal cache area, and when the server hits the compressed cache area, compares the required precision with the compression precision of the stored data. If it is, return the compressed data. If the compression cache misses or the accuracy is not satisfied, the client issues a data access request with accuracy sent from the client to the data source or a cache server closer to the data source. To obtain data.

The data source is conventionally designed to always return the original data ignoring the required accuracy of the data. However, the server of the data source also has means for internally storing the compressed data and means for interpreting the access request with accuracy and determining whether or not the accuracy is satisfied, thereby providing the data source. Can respond to the compressed data access request with the compressed data. As a result, the amount of data to be transmitted can be greatly reduced, so that the data transmission cost of the data supply source and the network load can be significantly reduced.

In the data access according to the present invention, the data intended by the client can be obtained only by designating the required accuracy of the data at the time of data access. Therefore, client-side modifications are kept to a minimum. Further, when the data access accuracy is not specified, it is not necessary to modify the conventional client application which always assumes that accurate data is obtained by interpreting that the original data is requested.

Further, the required accuracy of data is embedded as header information at the time of data access, and the header information uses an optional area, so that a conventional cache server which does not correspond to a data access request with accuracy can be obtained. Can easily ignore this information, thus ensuring that new cache servers implementing the features of the present invention and traditional cache servers will work well even if they are mixed on the network. You.

For the above reasons, the caching method according to the present invention and the client application premised on using it can be easily added to the conventional system.

[0023]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a diagram showing an outline of a cache device embodying the present invention. The data access request with accuracy 101 issued from the client 1 is sent to the cache device 2. In the cache device 2, the data identifier array 21 corresponding to the data identifier of the access request is used.
To check if the data is in the cache. The cache device 2 has a data identifier array 21
In addition, there are a cache area 23 for storing data and a compression attribute array 22 for storing compression attributes. However, when the compression attribute is attached to the data itself and stored together in the cache area, the compression attribute array 22 is not required.

In the cache device 2, when the data requested to be accessed exists in the cache (hit), the compression attribute of the data is compared with the required accuracy of the data described in the access request, and the compression attribute is determined. If the precision of the based data satisfies the required precision of the data, the compressed data and the compression attribute are returned to the client 1. The client 1 expands the received compressed data according to the compression attribute, if necessary, and displays the data.

Here, as means used for compressing data, any moving image compression technology represented by MPEG or the like is used for video data, and JPE is used for image data.
Any still image compression technology represented by G or fractal compression can be applied, and various compression methods can be applied to audio. In addition, since it is permissible to reduce the accuracy, for example, for moving images, frames per second are thinned out, or only the representative image of each scene is compressed, and thus slightly different from normal compression. Data compression is also applicable. Also, regarding compression of text data,
A technique of extracting and caching only keywords from the contents of the entire text, or automatically summarizing and caching sentences is applicable. Further, various data compression (or data amount reduction) techniques such as caching of sample data can be applied to a relational data table or the like.

When the access request from the client 1 does not exist in the cache held by the cache device 2 (a cache miss), the cache device 2 stores the secondary storage device 4 connected to the device, the network, or the like. Access data stored in a data supply source such as the external information system 3 connected via the PC. The data accessed is based on the accuracy specified by Client 1 and
With reference to the parameters etc. set in the cache device,
The compression method is determined, compressed by the compression means 24, and returned to the client. Further, the compressed data is stored in the cache device 2 together with a data identifier, compression information, and the like.

To determine the compression method, the available compression method, the standard compression method, the standard compression ratio, and the cacheable data are set so as to satisfy the accuracy specified by the client and are restricted by the system variable of the cache device. Decide to satisfy the maximum length. If a compression method that satisfies these conditions is not found (for example, if the client requests high accuracy and exceeds the maximum length of cacheable data), the compression process is not performed and the data obtained from the information source is Return to client without compression.

In the example described with reference to FIG. 1, the data is compressed in the cache device. In the present invention, it is of course possible to compress the data at the data source.

FIG. 2 shows an embodiment of the present invention in the case of performing compression processing at a data supply source. The data access request with accuracy 101 issued from the client 1 is sent to the cache device 2. The cache device 2 searches the corresponding data identifier array 21 from the data identifier of the access request, and checks whether the corresponding data exists in the cache. When the access request from the client 1 does not exist in the cache held by the cache device 2 (cache miss), the cache device 2 accesses data stored in the data supply source. Here, the data supply source is a data processing device 31 and a data storage device 32.
, The cache device 2 issues a data access request with accuracy to the data supply source. Then, the data supply source compresses the data stored in the data storage device 32 by the compression means 33 provided in the data processing device 31 so as to satisfy the required data accuracy, and stores the compressed data in the cache device 2. And return the compression information.

When the cache processor 2 issues a data access request with accuracy to the data source, the accuracy of the data satisfies the accuracy requested by the client 1 and is limited by the system variables of the cache device 2. Is determined so as not to exceed the maximum length of the cacheable data. If both the limitation of the cache device and the accuracy required by the client 1 cannot be satisfied, the accuracy required by the client 1 is directly requested from the data supply source. In order to reduce the processing load of the cache processing device 2 and the latency of data access, the data accuracy required for the data source is set to be the same as that of the client 1, and the data is returned from the source. When the data arrives, the data may be returned to the client 1 as it is, and whether or not the data can be taken into the cache device 2 may be determined at that time. In that case, the cache device 2
If it is determined that the compressed data cannot be stored in the cache device, the caching of the compressed data may be given up, or further compression may be performed in the cache device 2 and re-processed into a cacheable form before caching. good.

In the example described above with reference to FIG. 2, data compression is performed when an access request is issued from the cache device 2 to the data supply source. Therefore, if the processing capability of the data supply source is a bottleneck in the entire system rather than the network, it may take time to retrieve the data. Therefore, a method for compressing large-scale data frequently accessed in advance by a data supply source will be described below.

FIG. 3 is a diagram showing an example in which the present invention is applied to a case where a data supply source has data compressed in advance separately from original data. Data access request with accuracy 10 issued from client 1
1 is sent to the cache device 2. Cache device 2
Then, the corresponding data identifier array 21 is searched from the data identifier of the access request, and it is checked whether the corresponding data exists in the cache. When the access request from the client 1 does not exist in the cache held by the cache device 2 (cache miss), the cache device 2 accesses data stored in the data supply source. At this time, similarly to the above-described method, the cache device 2
Issues an accurate data access request to the data source. Upon receiving the access request with accuracy, the data source determines whether there is compressed data, and if so, determines whether the accuracy of the data by the compression satisfies the access request accuracy, If satisfied, the compressed data is transferred to the cache device.

FIG. 4 is a diagram showing an embodiment of the internal configuration of the cache device according to the present invention. The access request 11 with accuracy issued by the client includes the data identifier and the data accuracy in the cache device. In the cache device,
First, the entry of the data identifier array 21 is searched using the data identifier information, and the associated compression attribute and compressed data are extracted. The extracted data identifier is collated with the data identifier of the access request (2).
6). On the other hand, the extracted compression attribute is compared with the required accuracy of the data to determine whether the required accuracy is satisfied (27). In this example, the required accuracy is JPEG1 /
100 compression, cache compression attribute is JPEG 1/5
Since the compression is 0, the required accuracy is satisfied. These two determination results are sent to the hit / miss determination processing unit (28),
If both are satisfied, a cache hit signal (29) is generated. The extracted compressed data is returned to the client together with the compression attribute (1).
2).

FIG. 5 is a diagram showing an example of the operation when a cache miss occurs according to the present invention. When a request from the client misses the cache, the miss notification (4
The data access processing means 41 that has received 0) issues a data access request to the data supply source. The precision determination processing unit 42 determines a compression method in consideration of the attribute information of the data obtained from the data supply source, the required accuracy of the client, and system information indicating the characteristics of the cache device, and the compression unit 24 performs data compression. The compressed data is returned to the client, and further stored in the data identifier array, compression attribute array, and cache area in the cache device 2 together with the data identifier and the compression attribute.

In this embodiment, a detailed diagram corresponding to the embodiment in which the compression process in the case of a cache miss is performed in the cache device described in FIG. 1 is shown, but is shown in FIGS. 2 and 3. The same applies to the detailed drawings corresponding to the embodiments. In that case, the compression processing unit 24 and the accuracy determination processing unit 42 are provided on the data supply source side,
The data access request issued from the data access means is an access request with accuracy.

FIG. 6 is a diagram showing a data access procedure according to the present invention. The client 1 accesses data in the data source 3 via the cache device 2. Here, it is assumed that the client 1 is not requesting the original data in the data supply source 3 but requesting data whose accuracy is maintained at a certain level or more. For example, image quality,
There are sound quality, sentence summarization, DB sampling rate, and accuracy of statistical information. The cache device 2 holds a cache (original cache) 61 of the original data and a cache (compressed cache) 62 obtained by compressing the original data. The data stored in the compression cache is mainly data having a large size and data to which the accuracy of data is allowed to be reduced to some extent. The extent to which the accuracy of data is stored can be determined by storing the access history of the data.
In other words, by taking a history of the data identifier issued from the client and the required accuracy, it is possible to determine which data can be reduced in accuracy. Of course, for example, a still image of 50 KB or more is 1 /
It is also possible to adopt a method of deciding whether to perform unitary compression according to the data size or data type, such as 100 JPEG compression.

FIG. 6A is a diagram showing a data access procedure when an access request with accuracy from a client hits the compression cache of the cache device. Data access request with accuracy 6 issued from client
01 is sent to the cache device. Since the cache device is an access request with accuracy, the cache device accesses both the original cache and the compressed cache. At this time, if only the original cache is hit, the data is directly returned to the client. on the other hand,
As shown in the figure, if only the compression cache is hit, it is determined whether the data compression method satisfies the required accuracy (63). If the accuracy is satisfied, the compressed data is sent to the client together with the compression attribute. Forward.

When both the compressed cache and the original cache are hit and the required accuracy of the data is satisfied, the data in the compressed cache is given priority in principle in order to reduce the data transfer fee. To return. However, if the cost for data transfer between the cache device and the client is small, the original data may be transferred to the client.

In the conventional method shown in FIG. 8, when the compression cache is hit, a procedure for confirming the accuracy of the data is required. However, the accuracy of the requested data itself is sent to the cache device together with the access request. The procedure for checking the accuracy of data can be simplified. In addition, since the data accuracy determination process is performed in the cache device, it is not necessary to make a significant change to the client system. Further, since the required accuracy of the data access is transmitted to the cache device every time, the history of the required accuracy is held in the cache device, so that the data can be compressed and cached with appropriate accuracy.

FIG. 6B shows an access procedure when an access request from a client misses both the original cache and the compressed cache of the cache device. When an access request with accuracy issued from the client misses the original cache and the compression cache of the cache device (if the cache cache misses, the data of the compression cache is hit but the data accuracy satisfies the required accuracy). (Including the case where it is not performed), an access request with accuracy from the client is sent to the data supply source 5 (603). The data source searches for data that satisfies the received access request with accuracy from the original data and the compressed data held by the data source. Here, when the access request hits the compressed data and it is determined that the access request accuracy is satisfied (64), the compressed data is sent to the cache device together with the compression attribute (60).
4). The cache device transfers the received compressed data to the client (602) and registers the data in the compression cache (65).

According to the present invention, an access request with accuracy is issued to the data supply source even when a mistake is made in the cache device. Therefore, when the data supply source has compressed data satisfying the access request accuracy, By transferring only the compressed data, the access request of the client can be processed. According to this method, since it is not necessary to transfer the original data, the load on the network can be reduced, and the load for transmitting data from the data supply source is also reduced. The management of the compressed data of the data supply source and the hit determination method for the access request with accuracy can be realized by the same method as the cache device.

The access request with accuracy issued from the cache device to the data supply source may be the same as the access request with accuracy from the client, or the data stored in the cache device. Due to access history and restrictions on cache devices,
It may be issued after being modified within a range that satisfies the accuracy of the data request from the client.

FIG. 7 shows a data access procedure when a client access request makes a mistake in the cache device and also makes a mistake in the compressed data of the data source (or the compressed data does not exist in the data source from the beginning). FIG. In this case, it is necessary to access the original data stored in the data source.

FIG. 7A shows an example in which original data is transferred from a data source to a cache device.
Access request 70 with accuracy issued from client
1 is sent to the cache device, where a cache hit determination is made. Here, if the cache misses, the access request 703 with accuracy to the data supply source is made.
Is issued. The data source searches for data based on the received access request. If there is no appropriate compressed data, the original data is transferred to the cache device as it is (704). The cache device compresses the received data according to the access request accuracy of the client (71), and returns the result to the client (702). Further, the cache device registers the compressed data in the cache (72).

In this example, the original data is transferred from the data supply source to the cache device. Therefore, only the transfer load of the original data is applied to the data supply source, and the load of the compression process is applied to the cache device.
Therefore, it is particularly effective when the processing performance of the data supply source is low.

However, when the processing capability of the data supply source is sufficiently large, it is preferable to perform the data compression processing on the data supply source side as shown in FIG. in this case,
The original data is compressed according to the data access request with accuracy 703 received from the cache device (7
3) The result is transferred to the cache device together with the compression attribute (705). By doing so, although the load on the data supply source is increased, there is an advantage that the amount of data flowing through the network can be greatly reduced, and the load on the cache device can also be reduced. Further, by storing the data compressed on the data supply source side in the compressed data storage device, there is an advantage that the data can be reused at the next request for data access with accuracy to the data supply source.

Which method of FIG. 7 is adopted can be determined by a trade-off between the load on the server and the load on the network. In any case, on the cache device side,
Issue an access request with accuracy to the data supply source, and if the received data is already compressed, pass the data as it is to the client side and register it in the cache. If there is, you can do the same procedure after compressing it,
Both of these schemes can be mixed.

In the example described above, the data supply source is directly accessed from the cache device. However, a cache device having the same function may be interposed between the cache device and the data supply source. Absent.

FIG. 9 is a diagram showing a case where a plurality of cache devices (201 to 103) exist between a client and a data supply source. FIG. 9A shows a data access procedure according to the conventional method. In this case, a data access request, confirmation of data accuracy,
The data precision rejection protocol is repeatedly executed (9
01-904) Data access performance is degraded. In the conventional method, when the cache of the first cache device 201 is missed, the original data is accessed. Therefore, there is compressed cache data in the subsequent cache devices (202 to 203), and the accuracy thereof satisfies the client's request. The compressed data is not used.

FIG. 9B is a diagram showing an outline of an access procedure when the access method of the compressed cache according to the present invention is applied. According to the present invention, since the data request accuracy is described in the access request from the client, it is not necessary to confirm the data accuracy to the client when the data exists in the compression cache (911 to 914). Further, in this method, in order to issue an access request including data precision, if the data supply source 5 holds the compressed data or if the data supply source 5 performs a compression process at the time of the access request, the compressed data is transmitted. Can be returned to the client (915-918). For this reason, the amount of data flowing through the network can be significantly reduced as compared with the conventional method.

Further, when a mistake is made in the cache device, an access request with data request accuracy is sent to the subsequent cache device (202 to 203) or the data supply source 5.
If the compressed data of the intermediate cache device satisfies the required accuracy of the data when the first cache device misses, the compressed data can be used to respond to the client's access request. .

FIG. 9C is a diagram showing the flow of processing when an access request missed by the cache device 201 hits the compressed cache of the cache device 202. Access request missed to cache device 201 (921)
Is sent to the cache device 202 as an access request with data precision (922). Therefore, if the cache device 202 holds data that satisfies the required accuracy as a cache, the data can be returned to the client (923 to 924).

[0053]

As a method of caching and accessing a data object in an information system, large-sized data is compressed and stored in a cache device, and at the time of a data access request, the required accuracy of the data is also notified to the cache device and a data supply source. As a result, the compressed data can be effectively used as a cache, and the data access time, the network load, the server load, and the like can be reduced.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram showing an outline of a data cache and an access method according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram showing a flow of processing when performing data compression processing at a data supply source.

FIG. 3 is an explanatory diagram showing a processing flow when compressed data is held in a data supply source in advance.

FIG. 4 is an explanatory diagram showing a flow of processing when a hit occurs in a compression cache of the cache device.

FIG. 5 is an explanatory diagram showing a flow of processing when a mistake is made in a compression cache of the cache device.

FIG. 6 is an explanatory diagram of a data access procedure when a hit occurs in a compressed cache or compressed data.

FIG. 7 is an explanatory diagram of a data access procedure when a mistake is made in a compressed cache and compressed data.

FIG. 8 is an explanatory diagram of a data access procedure using a conventional compression cache.

FIG. 9 is an explanatory diagram of an access procedure when a plurality of cache devices are connected.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Client, 2 ... Cache apparatus, 3 ... External information system, 4 ... Secondary storage apparatus, 5 ... Data supply source, 6 ...
Network, 21: data identifier array, 22: compression attribute array, 23: cache area.

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Shigekazu Inohara 1-280 Higashi Koigakubo, Kokubunji-shi, Tokyo Inside Central Research Laboratory, Hitachi, Ltd.

Claims (12)

[Claims] 1. A cache device having an identifier array for uniquely identifying data and a cache area for storing data, wherein said cache area stores compressed data to which an irreversible compression method for reducing data accuracy is applied. Then, the compression attribute related to the compression of the data is stored together with the data in the cache device, and the client accessing the data issues a data access request specifying the precision of the data to the cache device. The data identifier array is searched to determine whether the data is present in the cache. If the data exists, the compression attribute corresponding to the data is compared with the required accuracy of the data. Compresses compressed data if the required data accuracy is satisfied Data cache wherein the returning with sex client. 2. An access request specifying data precision to the data cache device, wherein when the cache miss occurs, an access request specifying data precision that satisfies the precision specified by a client is sent to a data supply source or a subsequent cache. 2. The data cache method according to claim 1, wherein the data cache is issued to an apparatus. 3. The data obtained in response to a data access request issued to a subsequent data supply source or a cache device when an access request specifying data accuracy to the data cache device causes a cache miss is compressed. If the data is not original data, the compression is performed in the cache device within a range satisfying the data accuracy specified by the client, the compressed data is returned to the client, and the compression attribute is stored in the cache device. 2. The data caching method according to claim 1, wherein the data is registered together with a data identifier. 4. A data caching method according to claim 1, wherein when the data access request from the client does not specify the data precision, it is determined that the original data is requested and the compression cache is not accessed. 5. Compressed data to be stored in a compressed data cache for compressing and storing data, in addition to compressed data to which an irreversible compression method with reduced data precision is applied, reversible compression that does not reduce data precision. 2. The data cache method according to claim 1, wherein compressed data to which the method is applied is also stored, and the precision of the losslessly compressed data is treated to satisfy all access requests with precision. 6. In addition to a compressed data cache for compressing and storing data in a cache device, a normal data cache for storing data without compression is provided, and a client accessing the data specifies the precision of the data. When the data access request hits the compressed data cache and the data request accuracy is satisfied, the data access request is compressed regardless of whether the data access request hits the normal cache. A data caching method characterized by preferentially returning data to a client. 7. A data source for holding original data, wherein said data source holds compressed data obtained by compressing original data, and said data is supplied by said compression in response to an access request with data precision specified. A data access method comprising: determining whether a decrease in precision of a data item satisfies the accuracy specified in an access request; and returning the compressed data when the accuracy is satisfied. 8. In a data source holding original data, the data source does not hold compressed data, or the precision of the compressed data held by the data source is specified by an access request. When the accuracy is not satisfied, the data source compresses the original data so as to satisfy the accuracy specified in the access request, and returns the compressed data. 9. A process for compressing original data to generate compressed data according to claim 3, wherein a history of a data access request with accuracy from the client is retained in a cache device to thereby request the data from the client. 4. The data cache according to claim 3, wherein the data caching accuracy is determined and the compression method is determined within a range that satisfies the accuracy and a range that satisfies the required accuracy of the data for the past data as much as possible. Method. 10. A cache device for holding compressed data obtained by compressing original data, wherein the compression method employs irreversible compression for reducing the accuracy of data, and is adapted to a data access request issued by a client to the cache device. A data access method characterized in that data precision is specified. 11. A compression method for reducing the precision of data according to claim 1, wherein:
A data compression method characterized by performing data compression by holding sampled data. 12. The data compression method according to claim 1, wherein the compression is performed on the text data by holding a summary of the text data or a keyword list.