JPS63148345A - Data managing system - Google Patents

Abstract

PURPOSE:To efficiently take out data with a small capacity by taking out time series data at a specified time with a time index data storage means and an index data storage means in case of data retrieval. CONSTITUTION:In case of the retrieval of data at a time 15, the record pointer p1 of a time index data storage means 22 is obtained and is used to take out the contents of a record, and the record pointer p3 of a time data storage means 23 is obtained. This record pointer is used to take out the contents of the record of a data storage means 23, and the address of the record having data to be obtained in a time series data storage means 24 is obtained. That is, the time 10 of time index data having a value just preceding the time of obtained time data is subtracted from a given time to detect the position of the time 15, namely, a fifth place from the left in the record indicated by the pointer p3, and '1' in this position means that data at the time 15 exists in the time series data storage means 24.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a data management method in a computer system having files, and particularly to a time-series data management method.

[Conventional technology]

Conventionally, sequential files are used to manage time-series data using files. At this time, the file is sorted and stored in ascending order of time, but time data is stored inside the data, and in order to retrieve data at a specified time, check the data at this time from the first record. has been explored. However, with conventional data management methods, there was a problem in that it took time to search for data because it required reading up to the necessary parts in this order. To solve this problem, we construct a file with a time index for time series data, and use this index file to search for data at a specified time. The B-tree method was used to store it in a wooden structure.

[Problem that the invention seeks to solve]

The conventional data management method described above has a problem in that the capacity of the index file increases because each record has a time and a pointer. Furthermore, there is a problem in that an increase in the index file capacity causes an increase in the number of disk accesses required for access.

An object of the present invention is to provide a data management system that solves these problems.

[Means for solving problems]

The present invention provides a data management method for an information processing device that accesses a file holding time-series data, which is data related to time. It has a time series data storage means stored in a record and a flag indicating whether or not the time of the stored time series data corresponds on a time axis carved at a preset minimum time, or it can be processed using an appropriate compression algorithm. time data storage means having compressed data stored in ascending order of time; time index data storage means storing index data for each record of the time data storage means; and storage in the time index data storage means. and an index data storage means consisting of an index record storing index data for a time index record or an index record to be described later, and when searching for data, the time series data at a specified time is transferred to the time index data storage means and the index data. It is characterized in that it is carried out using a storage means.

[Effect]

According to the present invention, when searching for data at a specified time, a record in the time index data storage means corresponding to the data to be obtained from the index data storage means is obtained without storing (time and record pointer) for each record. Then, find the pointer to the time data storage means in the record, check the position corresponding to the data to be sought in the time data storage means, and determine the record containing the data to be found in the time series data storage means from this position. Furthermore, by reading this record, it is possible to extract the data at the specified time that should be obtained.

Furthermore, it is possible to reduce the disk access required to retrieve data.

〔Example〕

Embodiments of the present invention will be described below with reference to the drawings.

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

That is, for each of the time series data storage means 24, time data storage means 23, time index data storage means 22, and index data storage means 21, the access means (time series data access means 14, time data access means 13, time index data access means 12 and index data access means 11) are provided to search index data, time index data, time data, and time series data for a given time.

In order to explain the block diagram of FIG. 1, regarding the time series data shown in the following table, the time series data storage means 2 shown in FIG.
4. The contents of the time data storage means 23, time index data storage means 22, and index data storage means 21 are taken as an example.

Time series data Here, let us consider the case where the time data storage means 23 does not compress the data. In addition, for time index data and index data, since it is sufficient to have an index, for convenience of explanation, the last time of the record to be indexed, the number of time series data in the record to be indexed, and the pointer to the record to be indexed. shall have.

Now, when searching for data at time 15, the index data access means 11 obtains the record pointer of the time index data storage means 22. This first compares the time in the index data with the given time 15, finds the index data corresponding to the minimum time exceeding the given time, and obtains its record pointer.

In FIG. 2, since time 20 satisfies the above condition, the record pointer p is obtained.

After that, the time index data access means 12 retrieves the contents of the record in the time index data storage means using the record pointer to the time index data storage means 22 obtained by the index data access means 11, and Find the record pointer of 23. This first compares the time in the time index data with the given time 15, finds the time index data corresponding to the minimum time exceeding the given time, and obtains its record pointer.

In FIG. 2, since time 20 satisfies the above condition, the pointer p of that record is obtained.

Next, the time data access means 13 uses the time data storage means 23 obtained by the time index data access means 12.
The content of the record in the time data storage means is retrieved using the record pointer to the time series data storage means 24.
Find the address of the record that has the data you want. First, the time of the time index data having the value immediately before the time of the obtained time data is extracted. In the present case, this is time 10. By subtracting this value from the given time, the position of time 15 in the record indicated by pointer p3 can be found.

That is, in the uncompressed time data storage means 23 of FIG. 2, this corresponds to the fifth value from the left. If this is 1, it means that data at the time given to the time series data storage means 24 exists, and if it is 0, it means that data does not exist. Now, since the value is 1, time series data exists.

The record has the data number 6 at time 10 of the time index data and time 15 in the p3 record of time data.
The minimum integer value 2 that exceeds or is equal to the value obtained by dividing the sum of 2 and the number of flags whose value is 1 at the following times by 4, which is stored in the code of the time series data storage means 24, is the time series data to be found. This is the record that contains.

Furthermore, when the given time is 23, in order to find the record in the time data storage means 23, the number of data 9 corresponding to the time immediately before the record in the index file that indexes the record at time 30 in the time index file is calculated. More needs to be added. This is because time 23 is at the beginning of the time index file record and not at the beginning of the index file record.

Finally, the time series data access means 14 uses the time series data storage means 2 obtained by the time data access means 13.
The record is read from the address of the record having the data to be obtained in step 4, and the contents are notified as the output result.

Next, a case in which the data in the time data storage means 23 is compressed will be described by taking as an example a case in which the time data is compressed as shown in FIG. In the time data storage means 23 of FIG. 3, the order from the left is an integer value string for the flags of 1 in the previous embodiment.

When the given time described in the previous embodiment is 15, the index data access means 11 uses the same method as in the previous embodiment to access the record pointer p of the time index data storage means 22,
get. Thereafter, the time index data access means 12 obtains a pointer p to the record in the time data storage means 23 in the same manner as in the previous embodiment. The element of the integer value string in the record pointed to by pointer p3 is compared with 7, which is obtained by subtracting the final time 8 of the immediately preceding data in the time index data from time 15.

Since there is a matching value, time series data for time 15 exists, and the number of data in the record 5 of the immediately preceding data in the time index data and the left of the value 7 in the integer value string in the record indicated by pointer p3 The minimum integer value 2 that is equal to or exceeds the value divided by the number 4 stored in the code of the time series data storage means is
This is a record containing data to be obtained in the time-series data storage means.

〔Effect of the invention〕

As described above, according to the present invention, data at a specified time can be retrieved in logarithmic order time without storing pointers for each record in the time-series data storage means. Therefore, it has become possible to efficiently retrieve data with less capacity than conventional data management methods.

[Brief explanation of the drawing]

Fig. 1 is a block diagram showing an embodiment of the present invention, Fig. 2 is a diagram showing a storage method used in the present invention for time series data, and Fig. 3 is a block diagram showing the storage method used in the present invention for time series data. FIG. 6 is a diagram showing a storage method when data is compressed in a data storage means. 11... Index data access means 12...
Time index data access means 13... Time data access means 14... Time series data access means 21... Index data storage means 22... Time index data storage means 23... ...
Time data storage means 24... Time series data storage means Time Fig. 1 Fig. 3

Claims (1)

[Claims] (1) In a data management method for an information processing device that accesses a file that holds time-related data (time-related data), when data is stored, a predetermined number of records are recorded of time-series data in ascending order of time. It has a time series data storage means stored in the memory and a flag indicating whether or not the times of the stored time series data correspond to each other on a time axis carved at a preset minimum time, or this is stored using an appropriate compression algorithm. a time data storage means having compressed data and stored in ascending order of time; a time index data storage means storing index data for each record of the time data storage means; and index data storage means consisting of an index record storing index data for a time index record or an index record to be described later, and when searching for data, the time series data at a specified time is stored in the time index data storage means and the index data storage means. A data management method characterized by using means.