内容文本讲义数据库chapter8indexing.pptxVIP

Chapter 8. Indexing ;8.1 The Concept of Indexing;X/OPEN Syntax, Fig. 8.1, pg. 467, is extremely basic (not much to standard). CREATE [UNIQUE] INDEX indexname ON tablename (colname [ASC | DESC] {,colname [ASC | DESC] . . .}); Index key created is a concatenation of column values. Each index entry looks like (keyvalue, recordpointer) for a row in customers table. An index entry looks like a small row of a table of its own. If created concatenated index: create index agnamcit on agents (aname, city); had (aname, city) for two rows eual to (SMITH, EATON) and (SMITHE, ATON), Which once comes earlier alphabetically?;After being created, index is sorted and placed on disk. Sort is by column value asc or desc, as in SORT BY description of Select statement. NOTE: LATER CHANGES to a table are immediately reflected in the index, dont have to create a new index. Ex 8.1.1. Create an index on the city column of the customers table. create index citiesx on customers (city); Note that a single city value can correspond to many rows of customers table. Therefore term index key is quite different from relational concept of primary key or candidate key. LOTS OF CONFUSION ABOUT THIS.;With no index if you give the query: select * from customers where city = Boston and discnt between 12 and 14; Need to look at every row of customers table on disk, check if predicates hold (TABLE SCAN or TABLESPACE SCAN). Probably not such a bad thing with only a few rows as in our CAP database. But bad in DB2 from standpoint of concurrency. If we have a million rows, tremendous task to look at all rows. With index on city value, can winnow down to (maybe 10,000) customers in Boston, like searching a small fraction of the volumes.;Query Optimizer figures out an index exists to help, creates a Query plan (Access plan) to take advantage of it an Access plan is like a PROGRAM that extracts needed data to answer the Select. Recall X/OPEN form of Create Index: CREATE [UNIQUE] INDEX indexname ON