The Implementation Research On The Index Structure Of Database System

This paper discusses the design of UNIX-based index structure of database system and implement of index structure that based on Hash table. Hash table-based index structure is one of index structures used by current several main relational database systems, which is very suitable for equivalence query.The main purpose of index structure of database system is to improve the query speed of records, but at the same time, it is at the cost of increase in index maintenance, so tradeoff should be made in order to decide whether adopt index or not. Moreover, a special kind of index structure is suitable for a special query condition, for example, Hash table-based index structure is suitable for equivalence query, while B tree-based index structure is more suitable for range query, each of which has its merit. The paper gives an implementation solution of index structure in relational database system, the main idea of which is to combine Hash table and B tree so as to meet different query conditions. In general, B Tree is used for index of page level, while Hash table is used for index of content within page. Detail description about implementation of Hash table-based index structure is provided.The index structure in the database is running under very dynamic environment, it will be modified very frequently and randomly, how to improve the concurrent ability of the index structure become one of the keys to improve the DBMS's performance. PHILIP L.LEHMAN and S.BING YAO give the classic data structure B-Link tree which perfectly solve the concurrent problem between reader process and the inserter process, but it cause the waste of time when frequently operations are on the same page, and its delete algorithm cause the waste of disk space. This paper gives efficient solutions to solve the two problems above.