The birth of data management
The history of database can be traced back to fifty years ago, when data management was very simple. Classify, compare and tabulate through a large number of machines, run millions of punched cards to process data, and print the running results on paper or make new punched cards. Data management is the physical storage and processing of all these punched cards. However, in 1950, a computer named Univac I of Remington Rand Company introduced a tape drive that could input hundreds of records in one second, which triggered a revolution in data management. 1956 IBM produced the first disk drive-305 RAMAC. This drive has 50 disks, each with a diameter of 2 feet, and can store 5MB of data. The biggest advantage of using disk is that it can access data randomly, while punched cards and magnetic tapes can only access data sequentially.
195 1: Univac system uses magnetic tape and punched cards as data storage.
The germination of database system appeared in the 1960s. At that time, computers began to be widely used in data management, which put forward higher and higher requirements for data enjoyment. The traditional file system can no longer meet people's needs, so the Database Management System (DBMS) which can manage and share data in a unified way came into being. Data model is the core and foundation of database system, and all kinds of DBMS software are based on some data model. Therefore, according to the characteristics of data model, traditional database systems are usually divided into three categories: mesh database, hierarchical database and relational database.
The earliest mesh DBMS is IDS (Integrated Data Storage) developed by Bachman and others of General Electric Company of the United States in 196 1 year. 1964, Charles Bachman of American General Electric Company successfully developed the world's first mesh DBMS, that is, the first database management system-Integrated Data Storage IDS, which laid the foundation of mesh database and was widely distributed and applied at that time. IDS has the characteristics of data mode and log, but it can only run on GE host. The database has only one file, and all tables in the database must be generated by manual coding. Later, BF goodrich Chemical Company, a customer of GE, finally had to rewrite the whole system and named the rewritten system Integrated Data Management System (IDMS).
The mesh database model can naturally simulate hierarchical and non-hierarchical things. Before the emergence of relational database, mesh DBMS was more widely used than hierarchical DBMS. In the history of database development, mesh database occupies an important position.
Hierarchical DBMS appears after the network database. The most famous and typical hierarchical database system is IMS (Information Management System) developed by IBM in 1968, which is a hierarchical database suitable for its mainframe. This is the earliest large-scale database system program product developed by IBM. It came into being in the late 1960s, and now it has developed to IMSV6, providing support for advanced features such as clustering, N-way data sharing and message queue sharing. This 30-year-old database product plays a new role in today's WWW application connection and business intelligence application.
1973, Cullinane Company (later Cullinet Software Company) began to sell the improved version of IDMS of Goodrich Company, and gradually became the largest software company in the world at that time. Mesh database and hierarchical database solve the problem of data set sharing well, but they still lack data independence and abstraction. When users access these two databases, they still need to make clear the storage structure of data and point out the access path. The relational database that appeared later solved these problems well.
1970, Dr. E.F.Codd, a researcher at IBM, published a paper entitled "Relational Model of Data in Large Shared Database" in ACM's Journal of Communication, and put forward the concept of relational model, laying a theoretical foundation for relational model. Although Childs proposed a set-oriented model in 1968, this paper is generally regarded as an epoch-making milestone in the history of database systems. Codd's wish is to build a beautiful data model for the database. Later, Codd published many articles one after another, discussing the paradigm theory of measuring relational system and 12 standard, and laid the foundation of relational database with mathematical theory. The relational model has a strict mathematical foundation, a high degree of abstraction, simplicity, and easy understanding and use. But at that time, some people thought that relational model was an idealized data model, and it was unrealistic to use it to realize DBMS, especially worried that the performance of relational database was unacceptable, and some even regarded it as a serious threat to the ongoing mesh database normalization. In order to promote the understanding of the problem, 1974 ACM took the lead in organizing a seminar, at which there was a debate between the two factions for and against relational databases, led by Codd and Bachman respectively. This famous debate promoted the development of relational database, and finally made it the mainstream of modern database products.
1969 Edgar ·CODD invented the relational database.
After the relational model of 1970 was established, IBM added more researchers to the San Jose laboratory to study this project, which is the famous system R. Its goal is to demonstrate the feasibility of a fully functional relational DBMS. The project ended in 1979, and the first DBMS to realize SQL was completed. However, IBM's commitment to IMS prevented System R from being put into production. It was not until 1980 that System R was officially put into the market as a product. There are three reasons for the slow pace of IBM's productization: IBM attaches importance to reputation and quality, and minimizes failures; IBM is a big company with a huge bureaucracy. IBM already has a hierarchical database product, and the relevant personnel are not active or even opposed.
At the same time, however, in 1973, michael stonebraker and Eugene Wong of the University of California, Berkeley began to develop their own relational database system, Ingres, using the data published by System R. The Ingres project they developed was finally commercialized by manufacturers such as Oracle and Ingres in Silicon Valley. Later, both System R and Ingres won the "Software System Award" of 1988 issued by ACM.
From 65438 to 0976, Honeywell developed the first commercial relational database system-Multics relational data storage. Relational database system is based on relational algebra. After decades of development and practical application, this technology is becoming more and more mature and perfect. Its representative products are Oracle, DB2 of IBM, MS SQL Server of Microsoft, Informix, ADABAS D and so on. 1974, Ray Boyce and Don Chamberlin of IBM expressed the mathematical definition of the Codd relational database 12 criterion with simple keyword grammar, and proposed SQL (Structured Query Language) as a milestone. The functions of SQL language include query, operation, definition and control. It is a comprehensive and universal relational database language, and it is also a highly non-procedural language, which only needs users to point out what to do instead of how to do it. SQL integration realizes all operations in the database life cycle. SQL provides a way to interact with relational databases, which can work with standard programming languages. Since its birth, SQL language has become the touchstone for testing relational databases, and every change of SQL language standard guides the development direction of relational database products. However, it was not until the mid-1970s that relational theory was applied to commercial databases Oracle and DB2 through SQL.
1986, ANSI adopted SQL as the American standard of relational database language, and published the standard SQL text in the same year. There are three versions of SQL standard. The basic definition of SQL is ANS ⅸ 3 135-89 Database Language-SQL with Enhanced Integrity [ANS 89], commonly known as SQL-89. SQL-89 defines schema definition, data manipulation and transaction processing. SQL- 89 and the following ANS ⅸ 3 168- 1989, "Database Language-Embedded SQL" constitute the first generation of SQL standards. Ans ⅸ 3135-1992 [ans92] describes an enhanced SQL called SQL-92 standard. SQL-92 includes enhanced features, such as schema operation, dynamic creation and execution of SQL statements, and network environment support. After completing the SQL-92 standard, ANSI and ISO began to cooperate to develop the SQL3 standard. The main feature of SQL3 is that it supports abstract data types, which provides a standard for a new generation of object-relational databases.
1976, IBM E.F.Codd published a landmark paper "r system: database relational theory", which introduced relational database theory and query language SQL. Ellison, the founder of Oracle Bone Inscriptions, read this article very carefully and was shocked by its contents. This is the first time that someone has used a comprehensive and consistent scheme to manage data information. Author E.F.Codd 1966 published the theory of relational database, and developed the prototype in IBM research institute. This project is an R system, and the language for accessing data tables is SQL. After reading it, Ellison was keenly aware that a software system could be developed on the basis of this research. At that time, most people thought that relational databases would not have commercial value. Ellison thought this was their chance: they decided to develop the general commercial database system Oracle, which was named after the project they had done for the CIA. A few months later, they developed Oracle 1.0. But this is just a toy. They can do nothing but complete simple relationship queries. It took them a long time to make Oracle available. Maintaining the company's operation mainly depends on undertaking some database management projects and consulting work. However, IBM has no plans to develop it. There are many reasons why Big Blue gave up this product worth tens of billions: IBM researchers are mostly from academic backgrounds, and they are most interested in theory, not products put on the market. From an academic point of view, research results should be published. Papers and speeches can make them famous. Why not? Another main reason is that IBM has a hierarchical database product IMS, which sold well at that time. It was not until 1985 that IBM released the relational database DB2, and Ellision had already become a multimillionaire. Ellison once compared IBM's choice of Microsoft's MS-DOS as the operating system of IBM-PC: "The most serious mistake in the history of world business, worth more than hundreds of billions of dollars." The mistake of IBM publishing R system papers without launching relational database products soon may be second only to this. The market value of Oracle Bone Inscriptions reached $28 billion in 1996. With the development of information technology and market, people find that although the technology of relational database system is mature, its limitations are obvious: it can handle the so-called "tabular data" well, but it can do nothing about the increasingly complex data types in the technical field. Since 1990s, the technical field has been researching and seeking new database systems. However, the industry was once quite confused about the development direction of the new database system. Influenced by the technical trend at that time, people spent a lot of energy on the research of "object-oriented database system" or "object-oriented database system" for short. It is worth mentioning that the object-oriented relational database theory put forward by Professor Stonebraker in the United States was once favored by the industry. At that time, Stonebraker himself was hired by Informix as the chief technology officer at a large price.
However, the development of several years shows that the market development of object-oriented relational database system products is not ideal. Theoretical perfection has not brought enthusiastic response from the market. The main reason for its failure is that the main design idea of this database product is to try to replace the existing database system with a new database system. For many customers who have used the database system for many years and accumulated a lot of work data, especially big customers, the huge workload and huge expenses brought by the conversion of old and new data are unbearable. In addition, the object-oriented relational database system makes the query language extremely complex, which makes both database developers and application customers regard its complex application technology as a daunting road. At the end of 1960s, a new database software, Decision Support System (DSS), appeared, aiming at making managers use data information more effectively in the decision-making process. So at 1970, the first on-line analytical processing tool-Express was born. Other decision support systems followed, many of which were developed by the company's IT department.
1985, the first business intelligence system was developed by Metaphor Computer System Co., Ltd. as Procter &; Developed by Gamble Company, it is mainly used to connect sales information and retail scanner data. In the same year, Pilot Software Company began to sell the first commercial client/server execution information system-Command Center. Also in this year, the Ingres project of the University of California, Berkeley evolved into Postgres, with the goal of developing an object-oriented database. The following year, Graphael Company developed the first commercial object database system-GBASE.
1988, IBM researchers Barry Devlin and Paul Murphy invented a new term-information warehouse. After that, IT vendors began to build experimental data warehouses. 199 1 year, W. H. bill inmon published the book "How to Build a Data Warehouse", which really started the application of data warehouses.
199 1: W.H. "Bill "published" Building a Data Warehouse "on mon.
In 1990s, with the widespread adoption of client/server computing mode based on PC and enterprise software packages, the reform of data management was basically completed. Data management is no longer just to store and manage data, but to transform into various data management methods that users need. The sudden emergence of Internet and the appearance of XML language have opened up a new world for the development of database system.