Query Processing On Massive Data In Hierarchical Storage System

With the development of information technology, massive data with capacity over 1012 can be seen everywhere and the data capacity keep increasing. Taking all sorts of factor like cost and storage capability into account, people mostly use tertiary storage equipment like tape library or optical disk library as the primary storage medium to store massive data. At present, a three layer storage system composed of main memory, disk and tertiary storage has already been the primary storage structure of massive data. However, the actuality of massive data application based on three layer storage system is not optimism. One of the most important reasons is that most of the popular database products can not access data on tertiary storage online effectively. So far, there hasn't been a formal massive data management system based on three layer storage system. For making the best use of the data with abundance information which store in the three layer storage system, a massive data management system based on three layer storage system should be developed.As we known, there is a big performance gap of data access among each layer in three layer storage system. So in order to integrate them into a seamless storage system and support database system well, we must design a query processing method which fit the hardware characteristic of three layer storage system. Query processing is the key in developing massive information system based on hierarchical storage system. Resolving the problem of query processing will greatly accelerate the development of the management and application of massive data.Query processing on massive data stored in three layers storage system is studied in this Dissertation. Queries are divided to three kinds which are D-Query, T-Query and TD-Query base on where the data accessed is stored (on the disk or on the tertiary storage). Because the data accessed by D-Queries is stored on the disk, so we can use the method of conventional database to process them. We mainly study method for processing queries on tertiary storage (T-Query and TD-Query) later.First, a new way to process queries on tertiary storage is advanced after...