TW201812592A - Method for determining data in cache memory of cloud storage architecture and cloud storage system using the same - Google Patents

Abstract

A method for determining data in cache memory of a cloud storage architecture and a cloud storage system using the method are disclosed. The method includes the steps of: A. recording transactions from cache memory of a cloud storage during a period of time in the past; B. assigning a specific time in the future; C. calculating a time-associated confidence for every cached data from the transactions based on a reference time; D. ranking the time-associated confidences; and E. providing the cached data with higher time-associated confidence in the catch memory, and removing the cached data in the cache memory with lower time-associated confidence when the cache memory is full before the specific time in the future.

Description

 Method for determining data in cache memory of cloud storage device architecture and cloud storage device system using same  

The invention relates to a method for determining a cloud storage device architecture cache data and a cloud storage device system using the method, in particular to a method and a method for determining data in a cache memory of a cloud storage device architecture. The cloud storage device system of the method.

For cloud service systems, it is often tried to provide their services to customers as quickly as possible in response to customer requests. This target is easy to reach when the number of customers is small. However, if the number of customers increases greatly, subject to the hardware architecture of the cloud service system and network traffic, the response time will inevitably have a speed, but it should be within a reasonable range. On the other hand, if a cloud service is commercially competitive with other cloud services, regardless of what is limited, the cloud service system should technically respond to the client's request with the limited resources in the shortest time. This is a common issue faced by many cloud system developers, and everyone is looking forward to having the right solution.

In the traditional working environment, see Figure 1, there are many client computers 1 connected to a server 4 via the Internet 3. Server 4 is a device that primarily handles client requests and may perform complex calculations or perform only access to stored data. For the latter, the stored material can be retained in a cache 5 or a secondary memory 6. The number of cache 5 or auxiliary memory 6 may not be limited to one, but any number required for the cloud service. The server 4, the cache 5 and the auxiliary memory 6 form the architecture of the cloud service system. The cache 5 may refer to a dynamic random access memory (DRAM) or a static random access memory (SRAM). The auxiliary memory 6 may be a solid state drive (SSD), a hard disk drive (HDD), a writeable digital Versatile Disc (DVD), or even a magnetic tape. The physical difference between the cache 5 and the auxiliary memory 6 is the data storage property at the time of power failure. For cache 5, the data is temporarily stored when needed and disappears when power is lost. However, the auxiliary memory 6 can store data for a long time regardless of whether or not it is powered. Cache 5 has the advantage of fast access to data, but has the disadvantage of being volatile (volatile), high price and small storage space.

As mentioned above, it is clear that the hot data (more access) required to achieve most requests can be accessed quickly and provide cold data (less access) at a slower rate that can be tolerated, determining the appropriate It is important to store the data in the cache 5 and improve the performance of the cloud service. On average, the time to respond to all requests from client computers falls within acceptable limits. Recently, there are many traditional algorithms that can be used to determine which data should be cached (stored in cache 5). For example, Least Recently Used (LRU) algorithm, Most Recently Used (MRU) algorithm, Pseudo-LRU (PLRU) algorithm, Segmented LRU (SLRU) algorithm, 2-way set associative algorithm, Least-Frequently Used (LFU) algorithm, Low Inter-reference Recent Set (LIRS) algorithm, and so on. These algorithms are performed by the characteristics of the proximity and frequency of the data being analyzed, and the results are independent of other data (no data-related characteristics). There are also many conventional techniques, such as Chinese Patent No. CN101777081A and DOI: 10.1109/SKG.2005.136, which disclose other types of cache algorithms. They are classified as "data-related algorithms", using the original cached data (from the results of the aforementioned traditional cache algorithm) as the target data to obtain "data-related" data and cache storage. This means that the new cached data is somewhat related to the original cached data (new cached data has a higher chance of appearing alongside the original cached data). The above algorithms are perceived to be effective on certain mode workloads. However, because they both calculate the data that appears in the relative time period, rather than the data in the absolute time period, this leads to a phenomenon: the data selected by all algorithms to be cached for storage in a first time period (for example, the first 8 hours) may Not necessarily will be accessed during a second time period (for example, 8 hours after the first 8 hours). This is easy to understand because almost all data accesses are absolutely time-correlated or frequency-dependent, for example, starting at 8:55AM to 9:05AM every morning, meeting at 2:00PM every Wednesday, two weeks. The settlement of the salary, the inventory on the last day of the month, and so on. Therefore, the timestamp itself is an important and independent factor in considering cached data. However, no solution has been proposed to consider this.

This paragraph of text extracts and compiles certain features of the present invention. Other features will be revealed in subsequent paragraphs. The intention is to cover various modifications and similar arrangements in the spirit and scope of the appended claims.

It is an object of the present invention to provide a method for determining data in a cache memory of a cloud storage device architecture and a cloud storage device system using the same, which method is time-dependent to be accessed in a past period of time Data to analyze which data should be cached for storage. The method comprises the steps of: A. recording the processing content of the cache memory of the cloud storage device system in a past period of time, wherein each processing content includes a recording time, or a recording time and the past time period are accessed. Caching data; B. specifying a specific time in the future; C. calculating a time-dependent confidence for each cached data from the processed content based on a reference period; D. sorting the time-related Confidence; and E. providing cache data having a higher time-dependent confidence in the cache memory, and removing the cache memory when the cache memory is exhausted before the specific time in the future Cache data with lower time-dependent confidence in memory. Step E may be replaced by step E': E' provides cache data with a higher time-dependent confidence and data calculated from at least one other cache algorithm into the cache memory for future The use of cache memory is exhausted before a certain time, where there is a fixed ratio between cache data with higher time-dependent confidence and data calculated from other cache algorithms.

According to the present invention, the fixed ratio can be calculated based on the amount of data or the space occupied by the material. The specific time may be a specific minute of one hour, a specific hour of the day, a specific day of the week, a specific day of the month, a specific day of the season, a specific day of the year, a specific day of the year, A specific week in January, a specific week in a season, a specific week in a year, or a specific month in a year. The processing contents are periodically recorded in such a manner that the processing contents of two consecutive records are separated by a time span. The reference period may be within a particular minute of an hour, within a particular hour of the day, or within a particular day of the year.

The time-dependent confidence can be calculated by the following steps: C1. Calculating a first quantity, the first quantity is the quantity that occurs in the reference period in the past period of time; C2. calculating a second quantity, the second quantity The number of reference periods for accessing a target cache; and C3. dividing the second number by the first number.

Preferably, the cache algorithm may be a Least Recently Used (LRU) algorithm, a Most Recently Used (MRU) algorithm, a Pseudo-LRU (PLRU) algorithm, a Random Replacement (RR) algorithm, and a Segmented LRU (SLRU). Algorithm, 2-way set associative algorithm, Least-Frequently Used (LFU) algorithm, Low Inter-reference Recent Set (LIRS) algorithm, Adaptive Replacement Cache (ARC) algorithm, Clock with Adaptive Replacement (CAR) algorithm Method, Multi Queue (MQ) algorithm, or data-related algorithm with the result from step D as the target data. The type of the material can be an object, a block, or an archive.

The invention also discloses a cloud storage device system. The cloud storage device system comprises: a host for accessing data; a cache memory connected to the host for temporarily storing cached data for quick access; and a processing content recorder configured to be installed or installed The cache memory is connected to the host to record the processing content of the cache memory in the past period of time, wherein each processing content includes a recording time, or a recording time and a fast access time in the past period of time Taking data, receiving the host to specify a specific time in the future, calculating a time-related confidence for each cached data from the processed content based on a reference period, sorting the time-related confidences, and providing Cache data having a higher time-dependent confidence in the cache memory, and removing the cache memory when the cache memory is exhausted before the specific time in the future Time-dependent confidence cache data; and a plurality of auxiliary memories connected to the host for distributing stored data for access.

The cloud storage device system may also include: a host for accessing data; a cache memory connected to the host for temporarily storing cached data for quick access; and a content recorder, configured to or Installed in the cache memory, connected to the host to record the processing content of the cache memory in the past period of time, wherein each processing content includes a recording time, or a recording time and the past time period is accessed The cache data, the receiving host specifies a specific time in the future, calculates a time-dependent confidence for each cached data from the processed content, and sorts the time-related confidences based on a reference time period. And providing cache data having a higher time-dependent confidence and data calculated from at least one other cache algorithm into the cache memory to deplete the use of the cache memory before the specific time in the future , wherein there is a fixed ratio between the cache data having a higher time-dependent confidence and the data calculated from other cache algorithms; and a plurality of supplements A memory that is connected to the host to distribute the stored data for access. This fixed ratio can be calculated based on the amount of data or the space occupied by the data.

According to the present invention, the specific time in the future may be a specific minute of one hour, a specific hour of the day, a specific day of the week, a specific day of the month, a specific day of the season, and a year of the season. A specific day in, a specific week in a month, a specific week in a season, a specific week in a year, or a specific month in a year. The processing contents are periodically recorded in such a manner that the processing contents of two consecutive records are separated by a time span. The reference period may be within a particular minute of an hour, within a particular hour of the day, or within a particular day of the year.

The time-dependent confidence can be calculated by the following steps: C1. Calculating a first quantity, the first quantity is the quantity that occurs in the reference period in the past period of time; C2. calculating a second quantity, the second quantity The number of reference periods for accessing a target cache; and C3. dividing the second number by the first number.

Preferably, the cache algorithm may be an LRU algorithm, an MRU algorithm, a PLRU algorithm, an RR algorithm, a SLRU algorithm, a 2-way set associative algorithm, an LFU algorithm, a LIRS algorithm, an ARC algorithm, CAR algorithm, MQ algorithm, or data-related algorithm for processing data generated by a content recorder as a target data. The type of the material can be an object, a block, or an archive.

Cached stored data is time dependent. Thus, when the next relevant time comes, these materials are most likely to be accessed. Prior to this relevant time, the data can be stored in cache memory to improve the performance of the cloud storage device system. This is beyond the reach of traditional cache algorithms.

1‧‧‧Client computer

3‧‧‧Internet

4‧‧‧Server

5‧‧‧Cache

6‧‧‧Auxiliary memory

10‧‧‧Cloud Storage Device System

100‧‧‧Server

101‧‧‧Host

102‧‧‧Cache memory

103‧‧‧Processing content recorder

104‧‧‧Assistive memory

200‧‧‧Internet

301‧‧‧ PC

302‧‧‧ Tablet PC

303‧‧‧Smart mobile phone

Figure 1 is a schematic diagram of a conventional data access architecture.

2 is a schematic diagram of a cloud storage device system in accordance with the present invention.

Figure 3 shows the form for processing content records.

Figure 4 is a flow chart of the method proposed by the present invention.

Figure 5 and Figure 6 show the time-dependent confidence in the calculation of all cached data.

The invention will be more specifically described by reference to the following embodiments.

Figure 2 shows an ideal architecture for practicing the invention. A cloud storage device system 10 includes a host 101, a cache memory 102, a processing content recorder 103, and a plurality of auxiliary memories 104. The cloud storage device system 10 supports data storage for cloud services, which may be partially installed in a server 100, as shown in FIG. The server 100 is used to receive hardware requests from client devices such as a personal computer 301, a tablet 302, and a smart phone 303, or other remote devices connected via the Internet 200. After running the requests, the server 100 will reverse the corresponding response to the client device. Each component will be described in detail below.

The job function of the host 101 is mainly to perform data access in response to a request from a client device. In fact, host 101 may be a controller in server 100. In other embodiments, if the central processor of the server 100 has the same functionality as the controller described above, the host 101 refers to the central processor, or even the server 100 itself. The definition of the host 100 is not determined by its form but by its function. In addition, host 101 may have other functions, such as obtaining a hot data cache for storage into cache memory 102, but this is not within the scope of the present invention.

The cache memory 102 is connected to the host 101, and the cache data can be temporarily stored for quick access. In practice, the cache 102 can be any hardware that provides high speed access to data. For example, the cache memory 102 can be an SRAM. The cache memory 102 can be a standalone module for a large cloud storage device system, and some architectures can embed the independent module into the host 101 (CPU). As with caches in other cloud storage device systems, there is a preset cache algorithm that determines which data should be cached for storage in cache memory 102. The present invention provides a parallel mechanism that operates with existing cache algorithms for a particular purpose or timing. In fact, the cache mechanism can also be used to replace the cached data determined by the original cache algorithm.

The processing content recorder 103 is a requirement in the cloud storage device system 10. In the present embodiment, it is a hardware module and is configured in the cache memory 102. In other embodiments, the processing content recorder 103 can be software installed in the cache memory 102 or the controller of the host 101. In the present embodiment, the processing content recorder 103 is connected to the host 101, and its many functions are a feature of the present invention: the processing contents of the cache memory 102 are recorded in the past period of time, wherein each processing content includes a recording time. Or a record time and the cache data accessed in the past time period, the receiving host 101 specifies a specific time in the future, based on a reference time period, and calculate a time and time for each cached data from the processed content. Relevant confidence, sorting the time-related confidences, and providing cache data having a higher time-dependent confidence in the cache memory 102, and when the cache memory 102 is in the future When the specific time is exhausted before, the cache data with lower time-dependent confidence in the cache memory 102 is removed (or the cache data with higher time-dependent confidence is provided and at least one is The data calculated by other cache algorithms is transferred to the cache memory 102 to deplete the use of the cache memory 102 before the specific time in the future. These functions will be related to the proposed method. To be described later. It is emphasized that the term "time-dependent confidence" as used in the present invention is similar to the confidence defined in the association rule. The time-dependent confidence further extends to a confidence value calculated by taking a particular time or period of time as the target to obtain the probability that one or more of the materials were accessed in the past history.

The auxiliary memory 104 is also coupled to the host 101, which is capable of distributing stored data for access by customers. Unlike the cache memory 102, the output/input speed of the auxiliary memory 104 is slow, so that any data stored therein has a slower access speed in response to an access request. Frequently accessed material in the auxiliary memory 104 will be copied and stored in the cache memory 102 for caching. In practice, the auxiliary memory 104 can be an SSD, an HDD, a writable DVD, or even a magnetic tape. The configuration of the auxiliary memory 104 is determined in accordance with the purpose of the cloud storage device system 10 or the workload running thereon. In this example, there are three auxiliary memories 104. In fact, in a cloud storage device system, the number of auxiliary memories may be hundreds to thousands or even more.

Before the further explanation, some of the definitions used in the present invention are set forth first. See Figure 3, which is a form of processing content records for monitoring how data in the cache 102 has been accessed in the past. The form has a TID (Processing Content ID, from 0001 to 0024) column, a cache data (from D01 to D18) column, a reference time period (from H00 to H08) column, and a record time. H00 means that the recording time falls between 00:00 and 01:00, H01 means that the recording time falls between 01:00 and 02:00, and so on. The "1" in the TID and cache data fields means that the cached data has been accessed before the "current" record time and the "last" record time. The "1" in the TID and the reference period field means that the processing content is quantized for the recording time in different periods. The processing content is a record in which the cached data was accessed during the past period. In this example, the records (processing contents) in the past 8 hours are used for analysis. For better illustration, each processing content has a corresponding TID for identification. The processing content recorder 103 periodically records the processing contents in such a manner that the processing contents of the two consecutive records are separated by a time span. In this example, each processing content is recorded 20 minutes after the previous processing of the content record, with a time span of 20 minutes. In practice, the recording time does not necessarily need to be accurately placed on the scheduled timetable. For example, the recording time may fall at 00:30:18, 00:50:17, etc., not exactly at the 15th second but with a range. This is because there may be some large data being accessed or processed by the content recorder 103 waiting for a response from the remotely connected cache 102. A more positive way that is acceptable is that the time span is randomly selected, which is also within the scope of the invention.

It should be noted that in practice, the amount of processing content is very large, and may be thousands or more. For example, a 10-month time span of three minutes is recorded, and 24 processing contents are for illustrative purposes only. example. The processing content recorder 103 has more processing content, and the data demand can be predicted more accurately in a specific time in the future. Of course, not all cached data stored in the cache memory 102 will be accessed for a period of time. As shown in Figure 3, the processing content 0015 has no record of accessed data, it only has a recording time, 04:50:05.

Before the data in the cache memory 102 is revealed by the method of the present invention by the cloud storage device system 10, the cache data is first viewed. Although there are 18 caches, depending on the capacity of the cache 102, the number of cached data may be greater than 18. The 18 cache data is obtained at 07:50:05 by the method of the present invention and/or other cache operations used by the cloud storage device system 10. Because if some of the material is accessed too often, the processing content recorder 103 can add new material from one of the auxiliary memories 104 to the cache memory 102, and the cached data for analysis may also change accordingly. There may be other information that was cached before 03:50:05 but was later removed because it was not requested or "expected to be accessed."

From Figure 3, the characteristics of the cached data can be seen. The cache data D01 is often accessed during the first 3 hours and the last hour. The cache data D02 is accessed on average every other 20 minutes. The cache data D03 is accessed on average every two 20 minutes. The cache data D04 is accessed on average from 00:10:05 to 00:30:05, 02:50:05 to 03:10:05, and 05:30:05 to 05:50:05. The cache data D05 is accessed at 00:30:05 to 00:50:05 and 06:10:05 to 06:30:05. The cache data D06 is only accessed from 05:30:05 to 05:50:05. The cache data D07 is accessed on average from 00:30:05 to 01:10:05, 03:10:05 to 03:50:05, and 06:10:05 to 06:50:05. The cache data D08 is only accessed from 07:10:05 to 07:30:05, it may be the latest information added after 07:10:05 due to expected demand. Almost every time except 04:30:05 to 04:50:05, the cache data D09 is most frequently accessed. The cache data D10 is randomly accessed. The cache data D11 has no access to the record. The cache data D12 is accessed on average every 40 minutes of 20 minutes. The cache data D13 is randomly accessed. The cache data D14 is densely accessed between 00:50:05 and 04:30:05. The cache data D15 is densely accessed from 02:50:05 to 06:50:05, except for 04:30:05 to 04:50:05. The cache data D16 and the cache data D01 have similar access requirements. The cache data D17 and D18 are both accessed on average, but the cache data D17 has more requests between 03:50:05 and 04:30:05, and the cache data D18 is at 01:50:05 and 03: There are more requests between 10:05.

The main object of the present invention is to predict the data requested at a specific time in the future according to historical information, and provide corresponding data to the cache memory 102 before the arrival of the specific time in the future. A method for determining data in the cache memory 102 of the cloud storage device system 10 has several steps. Please refer to Fig. 4, which is a flow chart of the method proposed by the present invention. As described above, the method is performed by the processing content recorder 103. First, the processing contents of the cache memory 102 of the cloud storage device system 10 over a past period of time are recorded (S01). Each processing content includes a recording time (processing content ID 0015), or a recording time and a cached data accessed during the past time (8 hours in the example). Next, it is specified at a specific time in the future (S02). The cache memory 102 receives this particular time from the host 101 in the future. In accordance with the present invention, this particular time in the future can be any time or period of time in the future. For example, it can be a specific minute in an hour (for each hour), a specific hour of the day (for each day), a specific day of the week (for each week) , a specific day of the month (for each month), a specific day of the season (for each season), a specific day of the year (for each year), mid-January a specific week (for each month), a specific week in a season (for each season), (for each year), or a specific month of the year (for each year) Word). In this example, the processing content is used to determine which data should be cached before 00:00:00 (H00) on other days.

The third step is to calculate a time-dependent confidence for each cached material from the processed content based on a reference period (S03). The reference period refers to the time of "in a specific minute of one hour" (H00, every 20 minutes of the first hour of the day). In other examples, the reference period may be "within a particular hour of the day" or "within a particular day of the year", which varies with the number of time span records. In a particular example, the reference period may be "within a particular sub-time unit in a primary time unit." For example, within 24 hours of the day. The time-dependent confidence can be calculated by the following steps: A. Calculating a first quantity, the first quantity being the quantity of the reference period that occurred in the past period of time; B. calculating a second quantity, the second quantity The number of reference periods for accessing a target cache; and C. dividing the second number by the first number. In this example, the calculated time-dependent confidence table for all data is listed in Figure 5. If the specific time in the future is the first minute of 8:00 AM, and the reference time period refers to all 20 minutes in the past 8 hours, the result is shown in Fig. 6. From Figures 5 and 6, based on different situations, each cached data has a different calculated time-dependent confidence relative to other cached data.

Next, the time-dependent confidences are sorted (S04). The results in the examples are also shown in Figures 5 and 6, respectively. Finally, a cache data having a higher time-dependent confidence is provided in the cache memory 102, and the cache memory is removed when the cache memory 102 is exhausted before the specific time in the future. The cache 102 has a lower time-dependent confidence cache data (S05). Take Figure 6 as an example. Before 00:00 on other days, perhaps at 12:59:59 PM, except for D11, all the data is stored as new cache data in the cache memory 102 for access after 00:00. need. The reason for the D11 removal is that the cache memory 102 has insufficient space for 18 data storage and D11 has a lower time-related confidence than other data. The 18 cache cache files were analyzed for reasons of low hit rate or other factors and new data (D08) added, and one or more cached data has been removed by the cloud storage device system 10. The number of cached data used is 18. The newly cached data in the cache 102 is the most likely to receive the requested data after 08:00, and they are all calculated based on the time-dependent confidence. It should be noted that the above mentioned data or cache data type can be an object, a block, or a file.

In other embodiments, the last step (S05) may be different, which means that the processing content recorder 103 has a different function than in the previous embodiment. The changed step content is to provide cache data having a higher time-dependent confidence and data calculated from at least one other cache algorithm into the cache memory 102 to deplete faster before the specific time in the future. The use of the memory 102 is taken. There is a fixed ratio between the cache data having a higher time-dependent confidence and the data calculated from other cache algorithms, which is calculated based on the amount of data or the space occupied by the data. Go back to Figure 6. If the cache memory 102 is set to cache 20 data, when the ratio of the data for the cached data mentioned in the present invention is 60%, and the remaining data calculated by other cache algorithms accounts for 40%, then The obtained cache data are D01, D02, D03, D07, D09, D10, D12, D13, D14, D15, D16, and D18, a total of 12 pieces of data, and the rest of the data is proposed by the aforementioned cache algorithm. If some of the same cached data is jointly proposed by the two, the data with lower priority calculated by the present invention or other cache algorithms may be used in a supplemental manner, and the present invention is not limited thereto. Of course, in most cases, the cache memory 102 is designed to cache data according to its capacity, rather than the amount of data. From the above example, the capacity of the 60% cache memory 102 should be left to the data determined by the present invention, while the remaining 40% is given to at least one of the existing cache algorithms. The aforementioned cache algorithm includes, but is not limited to, Least Recently Used (LRU) algorithm, Most Recently Used (MRU) algorithm, Pseudo-LRU (PLRU) algorithm, Random Replacement (RR) algorithm, Segmented LRU (SLRU) Algorithm, 2-way set associative algorithm, Least-Frequently Used (LFU) algorithm, Low Inter-reference Recent Set (LIRS) algorithm, Adaptive Replacement Cache (ARC) algorithm, Clock with Adaptive Replacement (CAR) Algorithm, Multi Queue (MQ) algorithm, or data-related algorithm defined in the background of the invention. It should be noted that if the data-related algorithm is applied, the target data should use the result of the operation of the present invention, which means that the cached data with higher ordering is obtained from step S04 and then input into the data-related algorithm as the target data. To get the results of the data-related algorithm. In the cloud storage device system 10, this is generated by the processing content recorder 103 for use with the material related algorithms. The data related algorithm can also be executed using the processing content recorder 103.

Although the present invention has been disclosed in the above embodiments, it is not intended to limit the invention, and those skilled in the art can make some modifications and refinements without departing from the spirit and scope of the invention. The scope of the invention is defined by the scope of the appended claims.

Claims (18)

 A method for determining data in a cache memory of a cloud storage device architecture, comprising the steps of: A. recording a processing content of a cache memory system of a cloud storage device system in a past period of time, wherein each processing content includes a record time, or a record time and cached data accessed during the past period of time; B. specified at a specific time in the future; C. based on a reference time period, for each cached data from the processed content a time-dependent confidence; D. sorting the time-related confidences; and E. providing cached data with a higher time-dependent confidence in the cache, and when When the memory is exhausted before the specific time in the future, the cache data with lower time-dependent confidence in the cache memory is removed.    The method of claim 1, wherein the specific time is a specific minute of one hour, a specific hour of the day, a specific day of the week, a specific day of the month, and a season of the month. A specific day, a specific day of the year, a specific week of the month, a specific week of the season, a specific week of the year, or a specific month of the year.    The method of claim 1, wherein the processing contents are periodically recorded in a manner that the processing contents of the two consecutive records are separated by a time span.    The method of claim 1, wherein the reference period is within a specific minute of one hour, within a specific hour of the day, or within a specific day of the year.    The method of claim 1, wherein the time-dependent confidence is calculated by the following steps: C1. Calculating a first quantity, the first quantity being the quantity of the reference period that occurred in the past period of time C2. Calculating a second quantity, the number of the reference time period when a target cache data is accessed; and C3. dividing the second quantity by the first quantity.    The method of claim 1, wherein the type of the material is an object, a block, or an archive.    A method for determining data in a cache memory of a cloud storage device architecture, comprising the steps of: A. recording a processing content of a cache memory system of a cloud storage device system in a past period of time, wherein each processing content includes a record time, or a record time and cached data accessed during the past period of time; B. specified at a specific time in the future; C. based on a reference time period, for each cached data from the processed content a time-dependent confidence; D. sorting the time-dependent confidence; and E. providing cache data with a higher time-dependent confidence and computing from at least one other cache algorithm Data into the cache memory to deplete the use of cache memory before that particular time in the future, where cache data with higher time-dependent confidence and data computed from other cache algorithms There is a fixed ratio between them.    The method of claim 7, wherein the fixed ratio is calculated based on the amount of data or the space occupied by the data.    The method of claim 7, wherein the cache algorithm is a Least Recently Used (LRU) algorithm, a Most Recently Used (MRU) algorithm, a Pseudo-LRU (PLRU) algorithm, and a Random Replacement (RR). Algorithm, Segmented LRU (SLRU) algorithm, 2-way set associative algorithm, Least-Frequently Used (LFU) algorithm, Low Inter-reference Recent Set (LIRS) algorithm, Adaptive Replacement Cache (ARC) algorithm , Clock with Adaptive Replacement (CAR) algorithm, Multi Queue (MQ) algorithm, or data-related algorithm with the result from step D as the target data.    A cloud storage device system includes: a host for accessing data; a cache memory connected to the host for temporarily storing cached data for quick access; a processing content recorder, configured or installed The cache memory is connected to the host to record the processing capacity of the cache memory in the past period of time, wherein each processing content includes a recording time, or a recording time and the past time period are accessed. The cache data, the receiving host specifies a specific time in the future, calculates a time-dependent confidence for each cached data from the processed content, and sorts the time-related confidences based on a reference time period. And providing cache data having a higher time-dependent confidence in the cache memory, and removing the cache memory when the cache memory is exhausted before the specific time in the future A lower time-dependent confidence cache data; and a plurality of auxiliary memories connected to the host for distributing stored data for access.    The cloud storage device system of claim 10, wherein the fixed ratio is calculated based on the amount of data or the space occupied by the data.    The cloud storage device system of claim 10, wherein the specific time is a specific minute of one hour, a specific hour of the day, a specific day of the week, a specific day of the month, A specific day of the season, a specific day of the year, a specific week of the month, a specific week of the season, a specific week of the year, or a specific month of the year.    The cloud storage device system of claim 10, wherein the processing contents are periodically recorded in a manner of two consecutively recorded processing contents separated by a time span.    The cloud storage device system of claim 10, wherein the reference period is within a specific minute of one hour, within a specific hour of the day, or within a specific day of the year.    The cloud storage device system of claim 10, wherein the time-dependent confidence is calculated by the following steps: C1. calculating a first quantity, the first quantity being a reference period in the past period of time The number of occurrences; C2. A second quantity is calculated, the second quantity being the number of reference periods when a target cache data is accessed; and C3. dividing the second quantity by the first quantity.    A cloud storage device system includes: a host for accessing data; a cache memory connected to the host for temporarily storing cached data for quick access; a processing content recorder, configured or installed And the cache memory is connected to the host to record the processing content of the cache memory in the past period of time, wherein each processing content includes a recording time, or a recording time and the past time period is accessed. Cache data, the receiving host specifies a specific time in the future, calculates a time-dependent confidence for each cached data from the processed content, sorts the time-related confidence levels based on a reference time period, and Providing cache data having a higher time-dependent confidence and data computed from at least one other cache algorithm into the cache memory to deplete the use of the cache memory at a particular time in the future, There is a fixed ratio between the cache data having a higher time-dependent confidence and the data calculated from other cache algorithms; and a plurality of supplements A memory that is connected to the host to distribute the stored data for access.    The cloud storage device system of claim 16, wherein the cache algorithm is an LRU algorithm, an MRU algorithm, a PLRU algorithm, a RR algorithm, a SLRU algorithm, a 2-way set associative algorithm, LFU algorithm, LIRS algorithm, ARC algorithm, CAR algorithm, MQ algorithm, or data-related algorithm for processing data generated by the content recorder as a target data.    The cloud storage device system of claim 16, wherein the data type is an object, a block, or an archive.