To better understand what are the DBMS, we will review its evolution from the sixties until today.
DBMS to the sixties and seventies (IBM IMS, IDS Bull, Univac DMS, etc.) Were totally centralized systems, as befits those years operating systems, and hardware for which they were made: a large enterprise-wide computer and a network of dumb terminals and memory.
The first DBMS in the sixties, yet they were called and were aimed at facilitating the use of large data sets in which the interrelationships are complex.The archetype of implementation was the Bill of materials or explosion Parts, typical in the automotive, construction of spacecraft and related fields. These systems worked only in batches (batch).
Appearing keypad terminals, connected to the central computer via a telephone line, they start to build great applications on-line transaction processing (OLTP). The DBMS software were closely linked to communications and transaction management.
Although to write application programs using high level languages such as Cobol or PL / I, were also available instructions and subroutines specialized to treat BD requiring that the programmer knew many details of physical design, and that made the was very complex programming.
Since the programs were related to the physical level, it should change continuously when changes were made in the design and organization of the database.The basic concern was to maximize performance: response time and transactions per second.
Computers minis, first, and then micro computers, computer spread to virtually all businesses and institutions.
This required the development of applications would be easier.The DBMS of the seventies were too complex and inflexible, and could only use highly qualified personnel.
The emergence of relational DBMS * marks a significant step to facilitate the programming of applications with BD and to ensure that programs are independent of the physical aspects of the database.
* Oracle appears in 1980.
All these factors make greater use of the DBMS.Standardization, in 1986, the SQL language was a veritable explosion of relational DBMS.
During the eighties appear and spread very quickly on personal computers. It also appears these teams single-user software (eg, dBase and its derivatives, Access), with which it is very easy to create and use data sets, and that personal data are called bases. Notice that the fact referred to these early systems DBMS PC is a bit forced, as it does not accept complex structures or relationships, or could be used in a network that simultaneously serve many different users.But some, over time, have been turning into real DBMS.
At the end of the eighties, and relational DBMS is used in virtually all businesses. Nevertheless, until the mid-nineties, when needed a high performance have continued to use the DBMS prerelational.
In the late eighties and early nineties, companies have found that their departments have been buying departmental and personal computers, and applications have been making BD. The result has been that within the company there are numerous DBMS BD and several different types or suppliers.This phenomenon of multiplication of the BD and the DBMS has been increased by the fever of mergers.
The need to have an overview of the company and of linking different applications using different BD, together with the ease that give the networks for communication between computers, has led to the current DBMS, which allow a program to work with different BD as if it were a single. This is what is known as distributed database.
This ideal distribution is achieved when the various BD are supported by one brand of DBMS, ie when there is homogeneity.However, this is not so simple if the DBMS are heterogeneous. Today, thanks largely to the standardization of SQL language, DBMS of different brands can be serviced at each other and work together to provide service to an application program. However, in general, in cases of heterogeneity can not be reached to give the program that uses the appearance that it is a single BD.
In addition to this distribution "imposed", wanting to separate the integrated treatment of pre-existing BD, you can also make a distribution "desired" BD designing a physically distributed and replicated with parts in different systems.The basic reasons for which are interested in this distribution are:
1) Availability. The availability of a distributed system with a BD can be higher, because if it goes down one of the systems, others still work. If the data residing in the system is not available are replicated on another system, continue to be available. Otherwise, only available data from other systems.
2) Cost. A BD can reduce the cost distributed. In the case of a centralized system, all users computers that can be distributed to different and distant geographical areas are connected to the central system via communication lines.The total cost of communications can be reduced by a user to close the data used most often, eg on a computer in your office or even on your personal computer.
The technology is commonly used to distribute data is known as environment (or architecture) Client / Server (C / S). All of the relational DBMS market have been adapted to this environment.
The idea of C / S is simple. Two different processes running on one system or separate systems, they act so that one has the role of client or a service requester, and the other server or service provider.
For example, a program that a user application running on your PC (which is connected to a network) requests some data from a DB that resides on a UNIX computer which, in turn, runs the relational DBMS that manages it.The application program is the client and the DBMS is the server. A client process can request services to multiple servers. A server can receive requests from many customers. In general, a process that makes customer requesting a service to another process B can also do a service server which prompted another process C (or B, that this request would be the client).Even the client and server can reside on one system.
The ease of distribution of data available is not the only reason, not even the basic, the great success of the environments C / S in the nineties. Perhaps the main reason was the flexibility to build and grow the global computer configuration of the company, as well as making modifications to it, using very standard hardware and software and cheap.
The success of BD, including personal computers, has led to the emergence of the Fourth Generation Languages (4GL), very easy and powerful languages, specializing in application development based on BD.They provide many facilities at the time to define, usually visually, talks to enter, modify, and query data to the C / S.
Today, relational DBMS are undergoing transformation to accommodate three recent successful technologies, closely related: multimedia, object-oriented (OO) and Internet and web.
The types of data that can be defined in relational DBMS of the eighties and nineties are very limited. The incorporation of multimedia technologies, image and sound-in makes it necessary to accept relational DBMS attributes of these types.
However, some applications do not have enough with the addition of specialized media types.Need complex types that the developer can define as the application. In short, we need to abstract data types: TAD. The latest DBMS already incorporated this possibility, and a wide open market or TAD predefined class libraries.
This brings us to the object-oriented (OO). The success of the OO at the end of the eighties, the development of basic software applications in industrial engineering and construction of graphical interfaces with users, has made during the nineties is widespread in virtually all the fields of computing.
In the SI is also initiated the adoption, shy of the moment, the OO.The use of languages like C + + or Java requires relational DBMS fit them with appropriate interfaces.
The rapid adoption of the SI web makes the DBMS server resources to be incorporated into websites, such as SQL scripts including HTML, Java embedded SQL, etc.. Notice that in the world of the web are common OO data and multimedia.
In recent years it has begun rolling out an application type of the BD called Data Warehouse, or data warehouse, which also produces some changes in the relational DBMS market.
Over the years I have worked with BD in different applications, companies have accumulated large amounts of data of all kinds.If these data are analyzed appropriately can provide valuable information *.
Therefore, it is a great BD keeping with information from all kinds of enterprise applications (and even outside). The data in this big warehouse, Data Warehouse, you get a more or less elaborate replication of which is in the BD used in the daily work of the company. These data warehouses are used exclusivelymind to make inquiries, most especially to carry out studies * financial analysts, market analysts, etc..
Currently, the DBMS is adapted to this type of application, including, for example, tools such as:
a) The creation and maintenance of aftershocks, with some data processing.
b) Consolidation of data from different sources.
c) The creation of physical structures that efficiently support multidimensional analysis.
* For example, market developments in relation to pricing policy.
* These are often multi-dimensional statistics.