CN1294516C - Modeling memory and access method of real time database - Google Patents

Abstract

The data modeling method of real-time database and the database storage and access method based on this modeling method. The data modeling method is: Divide the information contained in all real-time measurement and control data items into static information and dynamic information. The static information is represented by the traditional ER model, and the dynamic information is represented by the clock face model. The database storage method is: convert the ER model representing static information into a relational model and a two-dimensional table structure and store it in a disk record file, and store the clock face model representing dynamic information in a disk stream file, and the name of the disk stream file is the same as the real-time The names of the measurement and control data items are consistent. The access method is: calculate the time offset from 0 o'clock according to the time parameter of the access data, and then calculate the physical address offset, and then directly locate the physical address of the disk to perform the read or write operation. The invention uses the clock face to describe the temporal information in the real-time measurement and control data flow within a certain sampling period, without establishing an index, and reduces the time and space resources required for data management.

Description

Real-time database modeling, storage and access methods

technical field

The invention relates to the field of real-time databases, in particular to a real-time database modeling method and a data storage and database access method based on the modeling method.

Background technique

Traditional database systems generally use the entity-relationship model (E-R model) or the extended entity-relationship model (EE-R model) to model the actual application data (conceptual model), and the most widely used relational database currently uses this concept The model is converted into a relational model and then converted into a two-dimensional table structure to store data. This mode is easy to understand, but it cannot meet the needs of real-time system data processing applications. Its limitations are mainly reflected in the following two aspects: the first is the limitation of the data model, the two-dimensional tabular data model adopted by the relational database cannot effectively deal with the three-dimensional or multi-dimensional data with temporal information, and the result is often to establish a large number of Tables, handled in a complex way, still have difficulty imitating the real-world relationships of the data, and worse, the connections between these tables are often hidden in the application, rather than in a more manageable database; The second is performance limitations. Relational database management systems designed for static applications such as report generation are not optimized for efficient data processing. However, its performance in the real world of real-time data processing is not satisfactory.

That is to say, because in real-time systems, real-time measurement and control information contains important temporal information, traditional database systems treat temporal information as ordinary parameters (attributes), resulting in a rapid expansion of the number of records in the databases of these application systems For example, if the sampling period of a data item is 1s, 3600 records will be generated in 1 hour. In a continuous measurement and control system with multiple data items and their sampling periods are different, the growth rate of the number of records can be imagined ; However, the time and space resources of real-time systems are generally not sufficient, so the efficiency of applying traditional database technology to real-time systems has not been ideal.

In short, when traditional database technology is applied in a real-time system, the expressiveness of the model is not enough, and the storage and access efficiency is low, especially in the case of a large number of records. In practical applications, indexing technology is often needed to improve efficiency. However, the establishment of Indexing costs time and storage space.

Object-relational database and other technologies in the new database technology try to change this situation, but its essence is to integrate some characteristics of object-oriented technology and traditional database on the basis of relational database, as well as a new programming tool environment to adapt to the new Internet-based environment. Web-based applications do not make fundamental recommendations for real-time database design methods.

Chinese patent No. 00811092 application document discloses "collection of statistics of real-time database objects" (International Application No.: PCT/US00/17491), which is a method for collecting real-time statistics of database objects by means of a database management system. The idea is Data processing based on the existing relational model is not a fundamental solution to the efficiency of real-time database storage and access.

Chinese Patent No. 00818943 application document discloses "Information Modeling Method and Database Search Method Using Modeling Method" (International Application No.: PCT/KR00/00505), which is a method of parsing data into semantic elements and receiving various data The information modeling method, including time information data and spatial information data indicating location and other types of data, is not a method for real-time database system and cannot solve the efficiency problem.

The application document of U.S. Patent No. US 6644549 discloses "Portable terminal with real-time database access, printing and display", which is a terminal device with real-time database. The real-time database technology it uses is still traditional database technology and has not Techniques that involve improving or transforming the real-time database management system itself.

U.S. Patent No. US6556882 application document discloses "Method and apparatus for generating real-time data from static files", which includes a method of storing real-time data in a static file database. Streaming real-time data to run manufacturing systems in order to analyze the performance of a network of manufacturing systems is not the solution to the efficiency problem of real-time databases themselves.

Contents of the invention

The purpose of the present invention is to overcome the technical problem of low storage and access efficiency of the current real-time database.

Furthermore, the technical problem to be solved by the present invention is to overcome the technical problem that the current database modeling method ignores temporal information and disk storage structure.

Correspondingly, the object of the present invention also includes: providing a data storage method and a database access method based on the above modeling method.

The technical scheme that realizes the object of the present invention and takes is summarized as follows:

A modeling method for practical application data of a real-time database is provided, including the following steps: first, separate static information (such as data item names, sampling periods) in all real-time measurement and control data items, and express them with an E-R model; The dynamic information (such as measured value) in the real-time measurement and control data is represented by the clock face model. In the clock face model, a clock face represents a data item, the name of the data item and the sampling period are marked on the clock face, and 0 o'clock on the clock face represents the sampling start time. If there are data items with the same sampling period, they can be combined on the same clock face.

Based on the above modeling method, a real-time data storage method is provided, which includes the following steps: converting the E-R model representing static information into a relational model and a two-dimensional table structure and storing it in a disk record file; converting the clock face representing dynamic information The model is stored in the disk stream file, and the name of the disk stream file is consistent with the name of the real-time measurement and control data item. The 0 o'clock on the clock face represents the sampling start time and corresponds to the initial storage address of the disk.

Based on the above-mentioned data modeling method and storage method, a real-time database access method is provided, including the following steps: searching for the name of the disk stream file according to the name of the access data item, and calculating the time offset from 0 o'clock according to the time parameter of the access data The amount, and then calculate the physical address offset, and then directly locate the physical address of the disk, so as to perform read or write operations without indexing. Here, due to the inherent characteristics of real-time measurement and control data, it is stipulated that the write operation can only be modified, and cannot be deleted or inserted. If the access data requires static information, open the disk record file. Since the disk record file only stores static information at this time, the amount of data is small, and the number of records is small, it is accessed according to the traditional access method.

The characteristics of the present invention are: combining the static information in the database data items with the modeling of the dynamic information in each real-time measurement and control data, paying special attention to the dynamic information and using the most reasonable disk storage method, which can solve the existing database storage and access efficiency low problem.

The present invention uses the clock face to describe the temporal information in the real-time measurement and control data flow within a certain sampling period, and stores the static information and dynamic information in the real-time measurement and control information separately in recording files and streaming files, and the access to the streaming files can be Calculate the physical address offset according to the time offset to locate the data, no need to build an index, and reduce the time and space resources required to manage the data, which is of great significance to improve the efficiency and performance of the real-time system. By adopting the technical solution of the present invention, a data modeling method of a real-time database and a data storage and database access method based on this modeling method are provided. Engineers can easily establish a conceptual model and a database system by using this modeling method. The software adopts this storage method and access method to speed up the efficiency of data storage and database access, and greatly improve the performance of the real-time system.

Description of drawings

Figure 1 Schematic diagram of the structure of the clock face model

Figure 2 Logical Address Space Diagram

Detailed ways

Below in conjunction with accompanying drawing, illustrate the mode of an implementation of the present invention:

First, analyze the requirements of the real-time system, determine what measurement and control data are available, and compile the static information in all real-time measurement and control data items (including data item ID, data item name, measurement unit, analog or digital, sampling period, alarm upper limit, Alarm lower limit, etc.) are separated and expressed by the traditional E-R model, with the data item ID or data item name as the primary key, because the data item name is named by the engineer, so it can be ensured that there is no duplicate name; and then each real-time measurement and control The dynamic information in the data (including time and measurement value) is represented by the clock face model. In the clock face model, a clock face represents a data item. The clock face is visually represented by a circle. The name of the data item is marked with text below the center of the circle. An acute angle is drawn above the center of the circle. The sampling period is marked inside the acute angle. The 0 point on the clock face represents Sampling start time. If there are data items with the same sampling period, they can be combined on the same clock face, that is, the names of the data items with the same sampling period are written on the same clock face and separated by commas, as shown in Figure 1.

Based on the above modeling method, a real-time data storage method is provided, which includes the following steps: converting the E-R model representing static information into a relational model and a two-dimensional table structure and storing it in a disk record file; converting the clock face representing dynamic information Models are streamed to disk as files. A streaming file means that the data in the file does not constitute a record form, but only a series of information collections in sequence. The name of the disk streaming file is consistent with the name of the real-time measurement and control data item. The 0 o’clock on the clock face represents the sampling start time and corresponds to the initial storage address of the disk. The logical address space distribution of the stream file is shown in Figure 2. The storage address number of a data consists of three parts: cylinder number, read/write head number and sector number. When storing dynamic data, the streaming files corresponding to different data items can be stored on different cylinders, that is, the real-time measurement and control data of the same data item are all located on the same cylinder. The amount of data that can be stored on the same cylinder is determined by the hardware. For example, a disk set with a single drive has 6 disks, 10 heads, and the number of cylinders is 1200. The track of the selected cylinder has 32 sectors. Each sector can store 512B, so the cylinder can store 10×32×512=163840B data. Assuming that storing a measurement and control data item requires 4 bytes of space, the cylinder can store more than 40,000 sampled data. Then the disk can store more than 40,000 sampling results of 1,200 data items.

Based on the above data modeling method and storage method, a real-time database access method is provided, including the following steps: if the access data requires static information, then open the disk record file, because the disk record file only stores static information at this time , the amount of data is small, and the number of records is small, so it is accessed according to the traditional access method. For the access of real-time measurement and control data, first search the disk stream file name according to the access data item name, and then calculate the time offset from the sampling start time according to the time parameter of the access data and the sampling period, and then calculate the physical address offset of the data. Shift, and then directly locate the physical address of the disk, so as to perform read or write operations. For example, if the sampling period is 1S, the sampling start time is 8:10:10, a temperature sampling value occupies 4 bytes, and the starting sector number of the file is xxxx10, if you want to query the temperature sampling value at 9:00:00, Then the time offset is (9-8)×3600+(0-10)×60+(0-10)=2990, the address offset is 2990×4/512=23, then it is directly located in the xxxx33 sector to perform a read operation. Here, due to the inherent characteristics of real-time measurement and control data, it is stipulated that the write operation can only be modified, and cannot be deleted or inserted.

Furthermore, the storage method of sampling data in a sector can be determined according to the sampling period and the space occupied by each sampling data. For example, for a sector of 512 bytes, one sector can store 128 sampling data. If each sampling data It takes up 4 bytes. In order to correspond to the time, each sector can store 120 sampling data, so as to facilitate the precise positioning of the data. Flag bits can also be set in the data stream, that is, the sampling time and the sampling value are stored in pairs, so as to further determine the data value after obtaining the sector data.

It is worth noting that the operating unit of the disk is a sector. When using disk storage, during specific operations, according to the given disk address, the movable arm assembly performs mechanical movement to locate on the specified cylinder, and the system selects the specified read-write head to determine Finally, the read-write head tracks the rotating track and reads the address of each sector during rotation. When the address is consistent with the given disk address, the system reads the data in the address into the disk buffer in the memory, or reads the data from the The disk buffer writes data to the specified disk address to complete a disk read and write operation.

The disclosed real-time database modeling, storage and access methods have been specifically described above, those skilled in the art will be able to understand and implement, without departing from the scope of the present invention, the modeling, storage and access methods can be formally Various modifications of details and details, so the above suggested but not limited modifications are within the scope of the present invention.

Claims (3)

1, a kind of storage means of real-time data base is characterized in that adopting the following step: at first, the static information in all real-time measuring and control data items is separated, with entity one contact E-R model representation; Then the multidate information in each real-time measuring and control data is come out with the clock face model representation; Become relational model and bivariate table structure to deposit magnetic disc storage formula file in the E-R model conversation of expression static information; Deposit the clock face model of expression multidate information in the disk stream-oriented file; In the clock face model, a clock face is represented a data item, and clock face is represented with circle, indicate the title of data item below the center of circle with literal, an acute angle is drawn in top, the center of circle, indicates the sampling period in the acute angle, in the time of initial moment of 0 of clock face expression sampling, the initial memory address of corresponding disk; The memory address number of data is made up of three parts: right cylinder number, read/write is No.1 and sector number; Disk stream-oriented file name is with the measuring and control data key name is consistent in real time; During the storage dynamic data, the pairing stream-oriented file of different pieces of information item is stored on the different right cylinders, promptly the real-time measuring and control data of same data item all is positioned on the same right cylinder.

2, the storage means of real-time data base according to claim 1 is characterized in that having the data item in identical sampling period, then is incorporated on the same clock face.

3, the storage means of real-time data base according to claim 1, its feature are that also data store at disk sector, and the operating unit of disk is the sector; According to the disk address that provides, the lever arm subassembly is positioned specified cylinder do mechanical motion, the read/write head of selecting appointment simultaneously is to determine magnetic track, final read/write head is followed the tracks of the magnetic track of rotation, read the address of each sector of when rotation, when this address is consistent with giving disk address, the data in this address are read in disk buffer in the internal memory, or data are write the designated disk address from disk buffer, finish a disk read-write operation.

Images