lec14-hash index精选.pdfVIP

Hash-Based Indexes Chapter 11 11/2/2011 EECS 484 1 Index Design Space Organization Structure for k* • Hash-based + Equality search Static hashing Data Entry (k*) Contents Extensible hashing 1. Actual Data record Linear hashing - index = file • Tree -based 2. k, rid + Range, equality search - actual records in a different file 3. k, list of rids 11/2/2011 EECS 484 2 Static Hashing • # primary bucket pages fixed, allocated sequentially, never de-allocated; overflow pages if needed. • h(key) mod N = bucket to which data entry with key belongs. (N = # of buckets) 0 h(key) mod N 1 key h What are important N-1 characteristics Primary bucket pages Overflow pages of h()? 11/2/2011 EECS 484 3 Static Hashing (Contd.) • Buckets contain data entries • Number of buckets (N) is fixed ahead of time • Static structure can be problematic • Consider many insertions • Long overflow chains can develop (and degrade performance!) • Might consider periodically doubling N and “rehashing” file • Entire file has to be read and written • Index unavailable while rehashing • Extensible and Linear Hashing : Dynamic techniques to fix this problem. 11/2/2011 EECS 484 4 Extensible Hashing • Main Idea: Use a directory of pointers to buckets • On overflow,