JPH0981439A - Access frequency threshold controlling method of distributed data base - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce the influence imparted to capacity by setting the present number of times of access of arbitrary registered threshold excess data as new threshold when a save area is filled up with data names. SOLUTION: When the control from a threshold excess information storage part 110 is delivered to a dynamic threshold changing part 111, the part 111 performs a decision as to whether the registration area of a using threshold excess data name save area is filled up or hot. When the registration area is filled up, an arbitrary threshold excess data name registered in the threshold excess data name save area is extracted and the present number of times of access of the extracted data name is set as new threshold to the threshold of threshold measured information. Next, threshold excess identifier is updated, further, save area identifier is changed, and finally, the new threshold and '0' are set to the threshold save of the the threshold excess data name save area corresponding to the changed save area identifier and the number of sets of registration, respectively.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is applied to a distributed database system which collects access frequency to data which is basic information for data rearrangement or data multiple arrangement, and is affected by the amount of data. The present invention relates to an access frequency threshold value control method for quickly identifying data with high access frequency without giving a threshold value in advance.

[0002]

2. Description of the Related Art An advanced communication service function and a related communication network control function are hierarchically separated from an information transmission network such as a telephone network, and are composed of three layers of a service operation layer, a service control layer and a service management layer. In the intelligent network, subscriber information such as what kind of network services subscribers receive under what conditions is managed in a database of service control layers, and these database of service control layers are distributed database systems. It is generally configured.

In a distributed database system which constitutes a service control layer of an intelligent network, in deciding which database system to allocate each distributed subscriber information, the network provided to each subscriber is decided. Data is placed in consideration of the load balance so that the load is not concentrated on a specific distributed database system by predicting the usage status of the service.

However, in the subscriber information allocation that anticipates the usage status of the network service, when an unexpected usage status occurs, one or a plurality of specific information allocated to a specific distributed database system is specified. Service requests to subscribers will be concentrated at the same time, the load will be concentrated on the specific distributed database system, and the processing capacity will be squeezed, and delays in data reference will prevent normal network services from being received. There is a possibility that a serious case will occur.

To address this problem, Japanese Patent Laid-Open No. 12338/1993 discloses a known example in which the number of data references and the average reference time for each database are collected during operation and the data arrangement of the database is changed from the collected information. Has been done.

Further, Japanese Laid-Open Patent Publication No. 6-119383 discloses a known example in which frequently used data are multiply arranged in a plurality of frequently used databases to minimize reference delay time.

As a publicly known example of a method for identifying data having a high access frequency with respect to the access frequency to the data such as the number of times of data reference or the frequency of use in these publicly known examples, there is a method disclosed in Japanese Unexamined Patent Publication No. 1-177673.

FIG. 19 is a diagram showing a record format of a frequency table for realizing the method disclosed in Japanese Patent Laid-Open No. 1-177673, in which the first field 1101 is a node id and each data in the database is The second field 1102, the third field 11
03, 4th field 1104, 5th field 110
5 is the number of accesses within a certain period, for example, 1
It is configured such that the second field becomes this week, the third field becomes last week, and so on, and the sixth field 1106 makes the sixth field 1106 the number of accesses of the second to fifth fields 1102 to 1105. Represents the total of.

As a method of identifying a node id having a high access frequency from the frequency table constructed as described above, each record of the frequency table is sorted in the order of total number of accesses (sixth field 1106), and the sorted records are sorted. , A method of selecting a node id having a high access frequency with a certain threshold value is disclosed.

[0010]

However, in the method of identifying the node id having a high access frequency disclosed in Japanese Patent Laid-Open No. 1-177673, the records in the frequency table are sorted in the order of the total access count 1106. Therefore, if the number of records to be sorted, that is, the number of node ids 1101 increases, the processing required for sorting increases.

The number of subscribers who receive network services using intelligent networks is enormous,
The number of subscribers accommodated in one distributed database system is tens to hundreds of thousands.

Therefore, when the above-mentioned known example is applied to the distributed database system of the intelligent network to identify the data with high access frequency, it is necessary to sort the tens of thousands to hundreds of thousands of data. As a result, there is a problem that the processing capacity of the distributed database system is greatly affected and the original database access processing is squeezed.

In this case, as a method of extracting data with high access frequency without performing sorting, there is a method of registering only data whose access frequency exceeds a certain threshold value. Table becomes full immediately and it becomes necessary to reset the threshold value, or if a large threshold value is given, it takes time to generate data that exceeds the threshold value. When the load is concentrated on the distributed database system, even if you try to relocate the frequently accessed data, which is the cause of the concentrated load, to the distributed database system that has the least load, the frequently accessed data to be relocated is The problem is that it takes time to identify.

An object of the present invention is to identify data with high access frequency in a distributed database system that collects access frequency to data that is basic information for data rearrangement or data multiple allocation. To
It is an object of the present invention to provide an access frequency threshold value control method that does not require sorting processing and that can quickly identify data with high access frequency without setting a threshold value in advance.

[0015]

In order to achieve the above object, according to the present invention, in each distributed database system, an access count adding section for calculating the access count for each data, and the access count having an initial value "0". When the threshold value that dynamically changes from the above is exceeded, the threshold value excess information storage unit that stores the corresponding data name in the save area and the area that saves the threshold value excess data name are filled A dynamic threshold changing unit that sets the current access count of one data name exceeding the threshold as a new threshold,
A threshold-exceeding information periodical transmission unit that periodically transmits the threshold-exceeding data name to the distributed database system management unit and resets the threshold to the initial value “0” is provided.

When the access frequency collection is started, the number of accesses to the data in the database is measured in data name units, and the measured number of accesses is the initial value "0".
If it exceeds the threshold, it is determined whether or not it exceeds the threshold that changes dynamically, and if it exceeds, the data name is stored in the save area, and then it is determined whether or not the save area is full. It is characterized in that the process of setting the current access count of any one of the threshold-exceeding data names registered in (1) as a new threshold is repeated.

[0017]

According to the above means, at the start of collection,
Since the threshold value is the initial value "0", the corresponding data name is stored in the save area every time the first access request is generated for each data, and the save area registers the data name for which the first access request is made. When it becomes full, the current number of times of access to any registered threshold exceeding data is set as a new threshold.

For this reason, at least the value of "1" which is the first access is newly set as a threshold value, and if there is a plurality of accesses to the data arbitrarily extracted before the save area is full in the registration by the first access. , Its value (greater than “1”) is set as a new threshold, and registration in the save area is restarted.

Each time the save area becomes full, the threshold is dynamically updated by the above means while collecting frequently accessed data, and after a certain period of time, the over-threshold data names registered in the save area are distributed. It is sent to the database system management unit, the threshold value is returned to the initial value "0", and access frequency collection is restarted.

By adopting such a method, it is not necessary to externally provide a threshold value for identifying data with high access frequency, and the threshold value is dynamically updated. It will be narrowed down to the save area with the progress of. Therefore, when specifying data with high access frequency, it is not necessary to search all databases and sort them in order of access frequency.

As a result, it is possible to reduce the influence on the processing capacity of the distributed database system when the frequently accessed data is specified from the data stored in the distributed database system in which the number of data is very large.

[0022]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described in detail below with reference to illustrated embodiments.

FIG. 1 is a block diagram showing an embodiment of the present invention.

In FIG. 1, a distributed database system management unit 101 corresponds to a service management layer 190 in an intelligent network, and is connected to a plurality of distributed database systems 102 via a communication network 103.

The distributed database system 102 is a service control layer 1 in an intelligent network.
Database 1 corresponding to 91 and accommodated in its own system
Database access unit 1 for controlling access to 06
05, an access frequency collection unit 108 that collects the access frequency to the database 106, a communication control unit 104 for connecting the distributed database system management unit 101 or another distributed database system 102 with the communication network 103, and communication control The processing unit 104, the database access unit 105, and the CPU usage rate measurement unit 107 that collects the usage rate of a central processing unit (hereinafter abbreviated as CPU) that executes the processing of the access frequency collection unit 108.

Here, the CPU usage rate is the rate of CPU time allocated for communication control per unit time, database access, and access frequency collection.

The distributed database system management unit 101 of the present embodiment has the same structure as the conventional structure, and a detailed description of its structure and operation will be omitted. System 10
The CPU usage rate of 2 and the data name having a high access frequency at that time are received from each distributed database system 102, and based on the information, the CPU usage rate is lower than that of the distributed database system 102 having a low CPU usage rate. A method of relocating frequently accessed data placed in another highly distributed database system 102 is adopted.

The CPU usage rate necessary for this purpose and the collection of data names with high access frequency are collected by the access frequency collection unit 108 in each distributed database system 102.

The access frequency collection unit 108 counts the number of times of access to each data, the access number addition unit 109, and the threshold excess information storage unit 110 that stores the data exceeding the threshold set by the dynamic threshold change unit 111. ,
Dynamic threshold changing unit 11 for changing the threshold again when the save data area exceeding the threshold is full and determining a new threshold for collecting frequently accessed data again
1. Distributed database system management unit 101 at regular intervals
And a threshold-exceeding information regular transmission unit 112 that performs processing for notifying the CPU usage rate and the name of the frequently accessed data at that time, and collecting again the frequently accessed data after the notification.

The threshold-exceeding information storage unit 110 is an area for saving various information necessary for storing threshold information or threshold-exceeding data names that change from moment to moment, and the threshold measurement information 201 stored here is As shown in FIG. 2, it comprises an access frequency collection identifier 202, a threshold 203, a threshold excess identifier 204, and a save area identifier 205.

Here, the access frequency collection identifier 202
Is for distinguishing the previous collection from the current collection when the access frequency collection for a certain period of time is completed, the notification to the distributed database system management unit 101 is finished, and the new access frequency collection is started. Information.

The threshold value 203 is a current threshold value updated by the dynamic threshold value changing unit 111 every time the over-threshold data name save area 401 of FIG. 4 is full.

The threshold excess identifier 204 is information for identifying whether or not the data is already saved in the threshold excess data name save area 401 shown in FIG. 4 when the access count of each data exceeds the threshold 203.

The save area identifier 205 is shown in FIG.
This is information for identifying which of the over-threshold data name save areas 401 consisting of A-side and B-side indicated by is currently used.

FIG. 3 shows the structure of each data arranged in the database 106. The data name 30
1, access frequency collection identifier 20 of threshold measurement information 201
Access frequency collection identifier save 302 for saving 2;
The threshold measurement information 201 is composed of a threshold excess identifier save 303 for saving the threshold excess identifier 204, an access count 304 for saving the access count, and actual data 305.

In FIG. 4, the access count 304 is the threshold 203.
When the number exceeds the threshold, an area for storing the data name is shown, and the threshold excess data name save area 401 has a two-sided structure of the A side and the B side.
Threshold 402 for saving the threshold 203, the registered number 403 indicating the number registered in the over-threshold data name save area 401, and the over-threshold value for storing a limited number (n in the case of FIG. 4) of data names And a data name 404.

An outline of the processing of the access count adding unit 109 of the access frequency collecting unit 108 will be described with reference to FIG.

When an access request to the database 106 occurs, the database access unit 105 transfers control to the access number addition unit 109. In the access count adding unit 109, the access frequency collection identifier save 302 of the corresponding data is saved.
Is extracted from the database 106 (step 50).
1).

Saved access frequency collection identifier save 3
It is determined whether the value of 02 and the access frequency collection identifier 202 of the threshold measurement information 201 match (step 50).
2). If they do not match, it is the first database access request since the access frequency collection was newly started.
In the access frequency collection identifier save 302 of the relevant data,
The access frequency collection identifier 202 of the threshold measurement information 201 is set (step 504), the access count 304 of the corresponding data is set to "0" (step 505), and the access count 304 of the corresponding data is incremented by "1" ( Step 506).

Thus, as the first database access request since the access frequency collection was newly started,
“1” is set in the access count 304.

On the other hand, when it is determined in step 502 that they coincide with each other, the collection of access frequencies is newly started, and then
Since the corresponding data has already been accessed at least once, the latest access count of 3 can be obtained by adding “1” to the access count 304 of the corresponding data (step 506).
Set 04.

Finally, from the access count adding unit 109,
Control is passed to the threshold excess information storage unit 110 (step 50).
8).

An outline of processing of the threshold excess information storage unit 110 of the access frequency collection unit 108 will be described with reference to FIG.

When the control is passed from the access count addition unit 109, the threshold excess information storage unit 110 receives the access count 30 of the corresponding data calculated by the access count addition unit 109.
4 is extracted (step 601), and the extracted access count 304 and the threshold 203 of the threshold measurement information 201 are compared (step 602).

When the access count 304 exceeds the threshold 203, the threshold excess identifier save 303 of the corresponding data is set in order to determine whether or not it has already been registered in the threshold excess data name save area 401 due to the current threshold excess. It is extracted (step 604) and it is determined whether or not the extracted threshold excess identifier save 303 and the threshold excess identifier 204 of the threshold measurement information 201 match (step 605).

If the threshold excess identifier save 303 and the threshold excess identifier 204 of the threshold measurement information 201 do not match, it is the first database access request since the threshold was changed to a new threshold, so the threshold excess data corresponding to the save area identifier 205. The data name 301 of the corresponding data is registered in the name save area 401, the registered number 403 is incremented by "1" (step 607), and the threshold excess identifier save 304 of the relevant data is added to the threshold excess identifier 204 of the threshold measurement information 201. Is set (step 608).

Then, control is passed to the dynamic threshold value changing unit 111 (step 609).

On the other hand, if it is determined in step 602 that the threshold is not exceeded, and if it is determined in step 605 that they have already been registered in the threshold-exceeded data name save area 401 and are in agreement, the processing of the access frequency collection unit 108 is executed. The process ends (step 612).

An outline of processing of the dynamic threshold value changing unit 111 of the access frequency collecting unit 108 will be described with reference to FIG.

When the control is passed from the threshold excess information storage unit 110, the dynamic threshold changing unit 111 determines whether or not the currently used threshold excess data name save area 401 is full (step). 701).

When the registration area is full, one arbitrary threshold value excess data name 404 registered in the threshold value excess data name save area 401 is extracted (step 703), and the current access count 304 of the extracted data name is extracted. Is set to the threshold 203 of the threshold measurement information 201 as a new threshold (step 704).

The method of arbitrarily selecting in step 703 is as follows.
The over-threshold data name (1) 404 registered at the beginning of the save area 401 may be either the over-threshold data name (1) 404 or the i-th registered over-threshold data name (i) 404.

Data arbitrarily selected until the data area for over-threshold data name saving area 401 becomes full and the threshold value is updated by the dynamic threshold value changing section 111, no matter how many times the data is requested to be accessed, the data is again requested. Only the access count 304 is updated without being registered. Therefore, the access count 304 of the corresponding data at the time point selected in step 703 is a value larger than the current threshold value.

Therefore, if the data arbitrarily selected in step 703 is data with high access frequency,
The updated threshold value is set to a value larger than the current threshold value, and even if the data is accessed infrequently, at least a value “1” larger than the current threshold value 203 is set, and a new threshold value is used. All frequently accessed data are narrowed down.

Next, the threshold excess identifier 204 is updated (step 705), the save area identifier 205 is further changed (step 706), and finally the threshold excess data name save area corresponding to the changed save area identifier 205 is updated. A new threshold 203 is added to the threshold save 402 of 401,
"0" is set to the registered number 403 (step 707).

On the other hand, when the registration area is not full in step 701, it is not necessary to change the threshold value and the dynamic threshold value changing process is terminated.

The threshold value excess information regular transmission unit 112 of the access frequency collection unit 108 uses the threshold value excess data name 404 registered in the threshold value excess data name save area 401 collected in the threshold value excess information storage unit 110 in the system. A function of notifying the distributed database system management unit 101 at a predetermined time interval arbitrarily set is realized.

An outline of the process of the threshold excess information regular transmission unit 112 will be described with reference to FIGS. 8 and 9.

Data transmission information 801 exceeding threshold value data name in FIG.
Shows the configuration of information transmitted to the distributed database system management unit 101, and the average CP within a fixed time measured by the CPU usage rate measurement unit 107 using the conventional technique.
CPU usage rate 802, which is an area for storing the U usage rate
And a threshold-exceeded data name 804 and an access count 805 that stores the access count of the corresponding data at the time of transmission.

In the threshold value excess information regular transmission unit 112, the CP
The CPU usage rate is collected from the U usage rate measuring unit 107 and set in the CPU usage rate 802 of the threshold value excess data name transmission information 801 (step 901), and the threshold value excess data name save area (A side) 401 and the threshold value excess data are set. The current access count 304 of the data name 404 registered in the name save area (side B) 401 is collected, and the data name 804, the access count 805, and the transmission count 803 are set (steps 902 and 903).

Next, the threshold value excess data name save area 40
The registered number 403 on both sides of 1 is set to "0" (step 904), the access frequency collection identifier 202 of the threshold measurement information 201 is updated (step 905), and the threshold 203 of the threshold measurement information 201 is set to "0". Initialization is performed (step 906), and the process ends.

As a result, the collection of frequently accessed data at a fixed time interval is restarted.

A specific example in which the threshold value is dynamically changed will be described with reference to FIG.

In the specific example of FIG. 10, the number of data items that can be registered in the over-threshold data name save area 401 is set to 5, and the data has already been registered in the save area, and one more data item is used as the current save area. This shows a case where the registration area 401 on the side A is full and the threshold value is updated.

When a database access request for the data name A5 is generated, the processing of the access count adding section 109 is executed.

Then, since the access frequency collection identifier save 1004 of the data name A5 and the access frequency collection identifier 1001 of the threshold measurement information 201 match, the access count 1 of the data name A5 is 1 by the processing of step 506 in FIG.
006 is added.

Since the access count 1017 of the data name A5 is larger than the threshold 1002 of the threshold measurement information 201 in the threshold excess information storage unit 110, the processing of step 604 of FIG. 6 is executed. Further, since the threshold excess identifier save 1005 of the data name A5 does not match the threshold excess identifier 1003 of the threshold measurement information 201, the process of step 607 of FIG. 6 is executed, and the threshold value of the A side currently used as the save area. The data name A5 is registered in the over threshold data name (5) 1012 of the excess data name save area 401, and the registered number 1010 is updated.

Next, the threshold excess identifier 1 of the threshold measurement information 201 is stored in the threshold excess identifier save 1016 of the corresponding data A5.
003 is set, and control is passed to the dynamic threshold value changing unit 111.

Since the registration area is full, the dynamic threshold value changing unit 111 executes the processing from step 703 onward in FIG. 7 to set a new threshold value. As a method of extracting an arbitrary threshold value excess data name in step 703 of FIG. 7, FIG. 10 shows an example of extracting the threshold value excess data name registered at the head of the registration area.

In the example of FIG. 10, the data name A1 is registered at the top 1011 of the A side of the currently used threshold exceeding data name save area 401, and the current access count 1015 of the data A1 is set to the new threshold value. Is set to the threshold value 1007 of the threshold value measurement information 201.

Next, the threshold excess identifier 1008 is updated,
The save area identifier 1009 is changed to the B side.

Finally, a new threshold value 1007 is added to the threshold value save 1013 in the data area save area 401 for exceeding the threshold value of the B side.
Is set, the registered number 1014 is further initialized to “0”, and the processing of the dynamic threshold value changing unit 111 is ended.

In the processing example of the dynamic threshold value updating unit 111 shown in FIG. 7, one arbitrary threshold value excess data name 404 is extracted (step 703), and the access count 304 of the corresponding data at that time is set to the new threshold value 203. However, a new threshold 203 may be selected from among a plurality of arbitrary threshold value excess data names 404 that have the maximum number of access times 304 at that time. According to this method, the new threshold value is updated based on the number of times the data that is accessed more frequently is updated, so that the data that is accessed frequently can be narrowed down more quickly and the data that is accessed frequently can be identified. It will be more effective if you hurry.

Further, in the processing example of the threshold value excess information regular transmission unit 112 shown in FIGS. 8 and 9, the threshold value excess data name 4 registered in the threshold value excess data name save area 401 is used.
04, the method of extracting the number of times of access 304 at that time and transmitting it to the distributed database system management unit 101 has been described, but the over-threshold data name save area 401 of two sides shown in FIG. 4 may be transmitted as it is. .

When the threshold value 203 is initialized (step 906), it is not "0" but the current threshold value 203.
If the initial value is set under a condition such as "1/2", then the data frequently accessed after the first regular transmission can be narrowed down more quickly.

Further, the distributed database system 102 does not autonomously specify frequently accessed data and periodically transmits the result, but the frequently accessed data is instructed by the distributed database system management unit 101. May be started, and the result may be reported by an instruction from the distributed database system management unit 101.

FIG. 11 is a diagram showing the configuration of a system by which a catalog seller realizes an online home shopping service, so-called virtual shopping.

The catalog distributor 1201 requests the detailed information of the products described in the catalog or requests the products from the clients 1205 via the network 1204 by the server 1202. Then, the product information stored in the database 1203 is provided to the client 1206, or the stock status associated with the order from the client 1206 is updated.

In virtual shopping, it is difficult to accurately predict the number of accesses by product, such as which product more detailed information a customer accesses, or the number of orders received by product. Therefore, if you can change the provided information appropriately, such as identifying the products that are frequently accessed by customers, making the information of the products that are frequently accessed more detailed, or reducing the prices of those products, you can place an order. It is possible to increase even more.

Further, if it is possible to detect the case where the orders are concentrated and the inventory is insufficient at an earlier stage, it is possible to prevent the order from being refused or the shipment of the product to be delayed, thereby further improving the serviceability to the customer. You can

However, as the number of target products increases, the number of frequently accessed products and orders increases, and if an attempt is made to identify products with low inventory in real time, the processing capacity of the server 1202 will be greatly affected. There is a problem that the response to the customer 1205 is delayed.

Therefore, by applying the present invention, it is possible to realize a system that identifies products with high access frequency in real time and identifies products with low inventory. FIG. 12 shows the configuration of the threshold value measurement information 201 in the case of realizing such a system.
FIG. 13 shows the structure of FIG. 13, and FIG. 14 shows the structure of the area for saving products with a small inventory.

In FIG. 12, 202 to 205 are the same information as described in FIG. In the present embodiment, in addition to this, an inventory threshold 1301, an inventory threshold drop identifier 1302,
An inventory threshold value drop save area identifier 1303 is newly provided, and these pieces of information correspond to the threshold value 203, the threshold value excess identifier 204, and the save area identifier 205, which are used when specifying data with high access frequency.

In FIG. 13, the inventory quantity 1402 is data that is subtracted from time to time when an order is received from the client 1206, and the inventory threshold value drop identifier save 1401 indicates that when the inventory quantity 1402 of each product is smaller than the inventory threshold 1301.
Inventory threshold drop data name save area 150 shown in FIG.
This is information for identifying whether or not it has already been saved in 1.

The identification of products with high access frequency can be realized by the processing described in FIGS. 5, 6, 7, and 9. However, the CPU usage rate information described in FIG. 9 is not necessary.

On the other hand, the inventory threshold value is dynamically decreased in order to specify the product whose inventory is low, but the inventory threshold value drop data name save area 1 shown in FIG. 14 is used.
When 501 becomes full, the current inventory quantity of data arbitrarily selected from the inventory threshold value drop data name save area 1501 is also displayed in the same flow as the processing flow of the dynamic threshold value changing unit 111 shown in FIG. In addition, the new inventory threshold 1301 is dynamically updated.

That is, although the threshold value is increased according to the access frequency in the embodiment of FIG. 1, the number of objects to be specified, the number of times
When specifying a measurable physical quantity such as a ratio or required time that is small, the threshold is dynamically updated to a smaller direction.

The initial value of the inventory threshold 1301 is the area of the inventory number 1402 of the database 1203 (byte length.
Use the maximum value that can be expressed in (bit length). Or, if the proper inventory quantity is known in advance for each product, instead of the inventory quantity, the inventory quantity is the ratio (%) to the appropriate inventory quantity.
Can be set as the inventory threshold, and in this case, the value “100 (%)” can be used as the initial value of the inventory threshold 1301.

In the system configuration of FIG. 11, the server 1202 executes the processing described with reference to FIGS. 5, 6, 7, and 9 in the server 1202 for virtual shopping.
It is possible to accurately measure the number of accesses for each product, such as which product more detailed information the customer accesses, or the number of orders for each product, without imposing a heavy burden on 02. As a result, it is possible to contribute to an increase in orders by making the information of the products that are frequently accessed more detailed or reducing the prices of those products.

Further, it is possible to identify products whose orders are concentrated and whose inventory is insufficient, at an early stage, and to prevent the order from being refused or the shipment of products being delayed, thereby further improving the serviceability to the customer. You can

FIG. 15 shows video-on-demand using a video server having a distributed processing configuration having a load distribution function when access is concentrated on specific video software (hereinafter, abbreviated as video stream). It is a diagram showing a configuration of a system that provides a service (hereinafter, abbreviated as VOD service).

The VOD service provider 1601 is the customer 1
The video server 1602 accepts the request from each client 1610 of 609 via the network 1608, and delivers the video stream to the client. In the example of FIG. 15, the video server 1602 has three load-balanced video stream transmission units 1604,
It is composed of one VOD management unit 1606, each has a video stream storage unit 1603, and an access frequency database 1605 for collecting access frequencies. The VOD management unit 1606 and the video stream transmission unit 1604 are networks. Connected by 1607.

Video server 160 having load balancing configuration
In No. 2, when there is a case where requests are concentrated on a specific video stream, the video stream having the concentrated requests is transmitted to the video stream transmitting unit 160 having the lightest load.
4 has a function to disperse the load on the video stream transmitting unit 1604 and to prevent the situation in which the request from the client 1610 cannot be met in advance.

In the case of the VOD service as well, the distribution of demand from customers is not uniform and changes from moment to moment, and there are cases in which social media cause the concentration of a particular video stream at a particular date and time. The greater the number of, the more difficult it is to identify the frequently accessed video streams in real time.

Therefore, the present invention is applied to a video server having such characteristics, and the structure of threshold value measurement information in the case of specifying a video stream having a high access frequency is shown in FIG.
Figure 6 shows the structure of the data area for saving data exceeding the threshold.
FIG. 18 shows the configuration of the access frequency database 1605.

The threshold measurement information 1701 shown in FIG. 16 is the video transmitted by the threshold 203, the threshold excess identifier 204, the save area identifier 205 and the video stream transmitting unit 1604 to each of the three video stream transmitting units 1604. It is composed of the total number of transmission streams 1703 for counting the total number of streams.

The threshold-exceeded data name save area 1801 composed of two surfaces A and B also has three video files.
The configuration corresponds to each of the stream transmission units 1604.

Further, the access frequency database 1605 managed by the VOD management unit 1606 stores the threshold excess identifier save 303 and the access count 304 in three video files.
The configuration corresponds to each of the stream transmission units 1604.

To specify a video stream having a high access frequency, the process described with reference to FIGS.
It is realized by implementing each video stream transmission unit of each unit.

The video stream multiplex arrangement is such that the total number of transmission streams 1703 is 1703 for the video stream transmission unit 1604 having the smallest total number of transmission streams 1703.
Can be realized by multiplexing the video stream having the highest access frequency registered in the over-threshold data name save area 1801 specified by the video stream transmitting unit 1604 having the largest number.

This makes it possible to promptly deal with the case where access to a specific video stream is concentrated at a specific date and time, and improve the service to the user.

The present invention is not limited to the above-mentioned embodiment, and it is possible to specify the number of objects to be specified, the number of times, the ratio, the required time, or the like, which has a small or large measurable physical quantity. It can be applied to all.

[0103]

As described above, according to the present invention, in the process of identifying the frequently accessed data in the distributed database system, the search process or sort process for all the data, which is required in the prior art, becomes unnecessary, Moreover, since it is not necessary to add a threshold value regarding the number of accesses from the outside, there is an effect that the specific target in the distributed database system can be efficiently and easily specified.

[Brief description of drawings]

FIG. 1 is a block diagram of a distributed database system showing an embodiment of the present invention.

FIG. 2 is a diagram showing a configuration of threshold value measurement information in FIG.

FIG. 3 is a diagram showing a configuration of a database in FIG.

FIG. 4 is a diagram showing a configuration of an area for saving an over-threshold data name in FIG.

5 is a flowchart showing a processing procedure of an access count adding unit in FIG. 1. FIG.

FIG. 6 is a flowchart showing a processing procedure of a threshold excess information storage unit in FIG.

FIG. 7 is a flowchart showing a processing procedure of a dynamic threshold value changing unit in FIG.

8 is a diagram showing a configuration of information notified to a distributed database system management unit in FIG.

9 is a flowchart showing a processing procedure of a threshold excess information regular transmission unit in FIG.

FIG. 10 is an explanatory diagram showing a specific example of dynamic threshold value change according to the present invention.

FIG. 11 is a block diagram showing an embodiment of an online shopping service system to which the present invention is applied.

12 is a diagram showing a configuration of threshold value measurement information in FIG.

13 is a diagram showing a configuration of a database in FIG.

FIG. 14 is a diagram showing the structure of an area for saving the over-threshold data names in FIG.

FIG. 15 is a block diagram showing an embodiment of a video on demand service system to which the present invention is applied.

16 is a diagram showing a configuration of threshold value measurement information in FIG.

FIG. 17 is a diagram showing a structure of an area for saving an over-threshold data name in FIG.

18 is a diagram showing a configuration of an access frequency database in FIG.

FIG. 19 is a diagram showing a configuration of a conventional access frequency table.

[Description of Reference Signs] 108 ... Access frequency collection unit, 109 ... Access frequency addition unit, 110 ... Threshold excess information storage unit, 111 ... Dynamic threshold change unit, 112 ... Threshold excess information regular transmission unit, 203 ... Threshold value, 304 ... Access count, 401 ... Data name save area over threshold.

Claims (3)

[Claims] 1. In a distributed database system in which a database is distributed among a plurality of database systems and data with high access frequency is specified for each database system, the number of times the database data is accessed is measured in data name units. , It is determined whether or not the measured access count exceeds a threshold value that dynamically changes from the initial value “0”, and if it exceeds, after storing the corresponding data name in the save area, whether or not the save area is full. If it is full, the access frequency threshold value control method for the distributed database is characterized in that the current access count of any one threshold value exceeding data name registered in the save area is set as a new threshold value. 2. The new threshold value is set to a value having the maximum number of times of current access among the arbitrary two or more threshold value excess data names registered in the save area. Method for controlling access frequency threshold of distributed database. 3. After setting the new threshold value, the threshold-exceeded data name is periodically transmitted to the distributed database management unit,
3. The access frequency threshold value control method for a distributed database according to claim 2, further comprising initializing the threshold value based on the threshold value at the time of transmission in the initialization of the threshold value after the completion of the regular transmission.