External Methods - Files

	Two type of files

		Sequential Access

			Records are read/written in order from beginning to end

		Direct (Random) Access
		
			Records are accessed directly, as in an array

	File Organization

		A file is a collection of BLOCKS
		A block is a collection of RECORDS
		A record is a collection of FIELDS

	File Access

		All input/output is done at block level, called BLOCK ACCESS

		Block size is determined by configuration of:
			hardware
		operating system

	Sorting Data

		A sort requires data to be in memory
		This is not possible for large data files
		Need to use a modified Merge Sort

	Modified Merge Sort

		Phase I:

			1. Read a block into an array from File1
			2. Sort the array
			3. Write the array to File2
			4. Repeat Steps 1-3 until EOF
			5. File2 now contains SORTED RUNS

		Phase II:

			Merge pairs of sorted runs to form larger sorted runs
			Repeat until only one sorted run remains

		Problem:

			Run size may eventually exceed memory capacity

		Solution:

		Input and Output must be BUFFERED for each pair of runs

		Buffering Input and Output for a Pair of Runs
		
			1. Create 3 internal arrays: Input1, Input2, Output

			2. Read a block from Run1 into Input1

			3. Read a block from Run2 into Input2

			4. Merge Sort Input1 and Input2 into Output

			5. If Input1 or Input2 become exhausted, read another block from the appropriate run
			
			6. If Output becomes full, write contents to the external sort file
			
External Methods - Tables

	Searching

		Simple binary search

			1. Recursively split file segment in half (initially, the entire file)
			
			2. Read the middle block of the file segment into an array
			
			3. Determine if the search key is in the block by comparing it to the first and last record key in the block

			4. Search the block for the search key
			
	Inserting and Deleting


		Problem:

			For large files, this could be problematic since it may require shifting large amounts of data
			
		Solution:
		
			Create an index file
			
			The index file is smaller, and therefore less costly to keep in sorted order
			
			Insertion into the data file can now be in any convenient location, since the data doesn't need to be sorted

		Layout of an Index File
			
			Each record in the file contains two parts:

				1. The key value matching the actual record in the data file
				
				2. A number "pointing to" block number in the data file containing the record
				
	Insertion and Deletion Revisited

		Deletion:
			
			Fill the record in the data file with spaces, and remove the pointer record from the index file
			
			Keep track of deleted records in another external file

		Insertion:

			Use one of the records marked as deleted, or place it at the end of the file
			
	Hashing

		An index file may become too large to manage insertions and deletions
		Use the hashing schemes from Chapter 12 on the index file