Point of Today Lecture: Disk Layout concern of Reliabilities and Performance

·	Final Exam: Tuesday 8:30 Same Room As Lecture (Double check the exam week schedule)
·	Continue Disk Layout

So far in this lecture we have learned three different kinds of Disk Layout
	+ Contiguous Allocation
	+ Linked Structure
	+ Indexed Structure

I/	Contiguous Allocation --
		+ Advantage:  like memory, fast--not indirection and simplifies directory access
		+ Disadvantage: inflexible, causes fragmentation over time--needs compaction
	Example: could be CD-w, write one time

II/	Linked Structure
		+ Properties: insert easy--just like link list structure, can't read ahead
		+ Advantage: good for sequential access, insert easy
		+ Disadvantage: can't read ahead, random access
	Very simple but poor performance-- file system consistency check

III/	Indexed Structure
		+ Properties: like indexed array, store pointers in array
		                   An "index block" contains pointers to many other blocks
		+ Advantage: better for random access
		+ Disadvantage: may need multiple index block (link together)
	Example: Floppy disk, the Unix Inode, DOS FAT

	DOS FAT
		+ One file allocation table describes the layout of the entire disk
		+ Each entry in the table refers to a specific cluster within a file.
			-- Zero says cluster not used
			-- Not zero says where the next FAT entry for the file is.
		+ A file's directory entry points to the first FAT entry for the files.
	However DOS FAT has limitation:
		Let's say FAT index is 16 bit -- the disk can have up to 64K clusters
		In order to have a disk larger than 32 MB you must use clusters that are larger than a single sector
			–- leads to fragmentation.

	UNIX INODE
		+ Designed to give fast access to small files, and eventual access to large files.
		+ Every file and directory has an Inode.
		+ An Inode has 15 entries.
			-- The first 12 entries point directly to the first 12 blocks of a file
			-- The 13th entry points to an array of 1024 pointers, each points to a block
			-- The 14th entry uses double indirection, an array of 1024 pointers, each pointing to an array of 1024 pointers, which then each point to a block.
			-- The 15th entry uses triple indirection...
		+ As UNIX evolved, they went from UFS -> FFS(Fast File System)
			-- Puts the inodes and data physically near each other.
			-- Keep space near the inode when initialized in anticipation of data coming in.
			-- This will keep it fast but the disk will be underutilized.
		+ Directories are blocks of data that contain a name and an inode.
			-- Directory != Inode, a directory structure contains an inode.
		+ The only difference between a directory and a file is interpretation of data.