Transcript Processing Data in External Storage

Processing Data
in External Storage
CS 302 - Data Structures
Mehmet H Gunes
Modified from authors’ slides
Contents
• A Look at External Storage
• Sorting Data in an External File
• External Dictionaries
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
A Look at External Storage
• External storage exists after program
execution
• Generally, there is more external storage than
internal memory
• Direct access files are essential for external
dictionaries
• A file contains records that are organized into
blocks
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
A Look at External Storage
• A file partitioned into blocks of records
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
A Look at External Storage
• Direct access input and output involves blocks
instead of records
• A buffer stores data temporarily
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
A Look at External Storage
• Updating a portion of a record within a block
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
A Look at External Storage
• Time required to read or write a block of data
– Longer than the time required to operate on the
block’s data
– Implication: reduce the number of required block
accesses
• File access time dominant factor in an
algorithm’s efficiency
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
Sorting Data in an External File
Example Problem
• Fille contains 1,600 employee records
• Sorted by Social Security number
• Each block contains 100 records
• Program can access only enough internal
memory to manipulate 3 blocks at a time
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
Sorting Data in an External File
• (a) Sixteen sorted runs, one block each, in file F1 ;
(b) Eight sorted runs, two blocks each, in file F2 ;
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
Sorting Data in an External File
• (c) Four sorted runs, four blocks each, in file F1 ;
(d) Two sorted runs, eight blocks each, in file F2
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
Sorting Data in an External File
• (a) Merging single blocks;
(b) merging long runs
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
Sorting Data in an External File
• Algorithm for merging arbitrary-sized sorted runs
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
External Dictionaries
• Records in order by search key
• Algorithm to traverse file in sorted order
• Retrieval could be done with binary search
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
External Dictionaries
• Shifting across block boundaries
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
Indexing an External File
Benefits of indexing with catalog
1.Index record will be much smaller than a data
record
2.Do not need to maintain the data file in any
particular order, insert new records in any
convenient location
3.Maintain several indexes simultaneously
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
Indexing an External File
• A data file with an index
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
Indexing an External File
• A data file with a sorted index file
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
External Hashing
• Similar to the internal scheme described in
Chapter 18
• Hash the index file instead of data file
• Hash table — contains a pointer to beginning
of chain of items that hash into that location
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
External Hashing
• A hashed index file
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
External Hashing
• A single block with a pointer
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
External Hashing
Insertion under external hashing
1.Insert the data record into the data file
2.Insert a corresponding index record into the
index file
Removal under external hashing
1.Search index file for corresponding index record
2.Remove data record from the data file
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
B-Trees
• (a) Blocks organized into a 2-3 tree;
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
B-Trees
• (b) a single node of the 2-3 tree
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
B-Trees
• (a) A node with two children; (b) a node with
three children; (c) a node with m children
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
B-Trees
• (a) A full tree whose internal nodes have
five children; (b) the format of a single node
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
B-Trees
• A B-tree of degree 5
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
B-Trees
Insertion into a B-tree
1.Insert the data record into the data file
2.Insert a corresponding index record into the
index file
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
B-Trees
• (a through d) The steps for inserting 55
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
B-Trees
• (a through d) The steps for inserting 55
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
B-Trees
• (e) splitting the root
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
B-Trees
Removal from a B-tree
1.Locate the index record in the index file
2.Remove the data record from the data file
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
B-Trees
• (a through e) The steps for removing 73 ;
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
B-Trees
• (a through e) The steps for removing 73 ;
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
B-Trees
• (a through e) The steps for removing 73 ;
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
B-Trees
• (a through e) The steps for removing 73 ;
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
B-Trees
• (f) removing the root;
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
Traversals
• Accessing only the search key of each record,
but not the data file
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
Traversals
• Accessing data file also
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013
Multiple Indexing
• Multiple index files
Data Structures and Problem Solving with C++: Walls and Mirrors, Carrano and Henry, © 2013