Methods for parallel execution of complex database queries

摘要

During the last decade, all commercial database systems have included features for parallel processing into their products. This development has been driven by the fact that databases grow in size at considerable rates. According to the results of the 1998 ‘very large database contest' the world's largest databases, which have reached a size of over 10TB, double in size every year. At that speed, they outgrow the increase in processor speed and memory size, so additional measures are required to accommodate the effects of rapidly growing volumes of data. Parallelism is one of those options. It helps to keep processing times constant, even if the size of the database increases. That effect, which is often referred to as ‘scaleup' is important for loading, index creation. all kinds of administrative operations on the database, and of course for long batch-type applications. Parallelism is also employed to speed-up queries that otherwise would take days or weeks to process and thus would be useless for the application. This type of requirement f fast results of complex queries on large data sets is characteristic of decision support applications. In this overview we will explain how parallelism in databases can help to solve such problems.