Now everything in society saves binary data in the calculation, the computer is just a place where data is stored. It does not understand these 0 and 1 representatives, people also don't understand. So this data requires people to process, which will change their information that people can understand. The processing of data can now be divided into two types: operational processing and analytical processing. Operating processing is mainly referred to as a machine transaction (OLTP), which is ordinary daily operation, such as the ATM machine withdrawal, check balance, aircraft booking, etc. Analytical treatment is mainly online analysis processing (OLAP), mainly used to analyze data and make decisions. For example, banks have evaluated customer credit.
1.1 from the database to the data warehouse
The database is mainly used for OLTP, and the current rise of DSS needs to be analyzed. The OLTP database cannot support DSS very well. The reason for the OLTP database is not suitable for DSS:
1. Different performance characteristics of transaction and analytical processing. The user behavior in the transaction environment is that the data has a high access operating frequency and a short time per operation process. In the analysis processing environment, the user's behavior mode is completely different. A DSS application may run a few hours in this period of time, you need to consume a lot of system resources. So the curve of their resource consumption is completely different and cannot coexist in a database.
2, data integration problem. The primary premise of analysis and decision is effective data. The application of real-world transaction processing system may be more dispersed, the data is inconsistent, and there is a spider web problem, external data, and unstructured data. So you need to integrate existing data in one
(In the data warehouse), it is easy to analyze and utilize.
3. Dynamic integration of data. The integration of data is not once, it requires periodicity to extract data from other data sources, and refresh the data in the data warehouse.
4, historical data. The data in the OLTP is generally only stored for a while. A large number of previous historical data is stored in a data warehouse, which is convenient for certain analysis and decision making.
5, data comprehensive problem. The data in the database is detailed, while DSS needs to analyze a large amount of data, it is very efficient, so
Data can be synthesized, such as a summary of summary information, which is also stored in a data warehouse.
Due to the above reasons, the transaction processing is separated from the analysis process, and the data they rely on is also separated, so the data warehouse is also from the database.
Independent, becoming a matter of completely different database properties.
