- Structure of a hard disk
- Hard disk consists of a number of
platters
The platters rotate at a very high speed (5400 RPM to 10,000 RPM)
- Disk (read/write) heads move over the platter surface
to read and write (magnetize) data bits
The disk head can read or write data only when the desired disk surface area is under the disk head.
Internal operation of a hard disk:
- Hard disk consists of a number of
platters
- Data access time
of data on disk consists of:
- Seek time (get to the right track)
- Latency (wait for the right sector to rotate under the disk head)
- Transfer time (actual reading of the data)
- Each platter is
logically
divided into a number of
tracks
- Each track is
logically
divided into a number of
sectors:
- Each sector is uniquely identified by its sector number
Disk sectors are usually 512 bytes in size.
- To facilitate data access (larger chuncks)
a number of sectors are
logically
groups into a
disk block
- Each disk block
contains a number of
sectors
- Each disk block
is uniquely identified by its
block number
- Therefore, the disk has a mapping defined between
disk block numbers and
sector numbers:
If 1 disk block contains N sectors, then disk block k = sector number k*N. k*N+1, ..., k*N + (N-1)Disk blocks are usually 4K or 8 K bytes in size.
- Sectors of a disk block need not be located "contiguously"
on the platter (for speedier access,
sectors are usually interspersed)
E.g., the sectors (4, 5, 6, 7) of this block is interspersed:
The reason to intersperse the sectors is:
- Sectors of a block will
be read
consecutively (one after another)
- Furthermore, the disk drive
will perform
integrity checks on
the data read on each sector
(This correctness check will take up some time; in the mean time, the disk drive will keep rotating !!!).
- If the sectors are
located contiguously,
then the disk head may have
passed the
start of the
next sector when the drive
finishes checking....
In that case, the drive will have to wait one whole revolution of the platter to read the next sector.
- By interspersing the sectors, the disk drive can achive better access time for the consecutive sectors read/write
- Sectors of a block will
be read
consecutively (one after another)
- Each disk block
contains a number of
sectors
- Each computer system has a (built-in)
file system
- File systems used by Microsoft Windows are: FAT
(File Access Table), FAT32 and NTFS
- File systems used by Solaris UNIX are: UFS (UFS File System) and ZFS (ZFS File System)
- File systems used by Microsoft Windows are: FAT
(File Access Table), FAT32 and NTFS
- The file system always
reserve
a number of
disk blocks
(called: directory disk blocks)
to store information about the files on the disk:
- The
directory disk blocks
contains a large number of
directory entries
- The structure of a directory entry looks something like this
(and each file system will have it's own directory entry format,
so I can't be specific):
- A file name
- Zero or more block number (the number of blocks can be zero, because the file can be empty)
+---------------------------+----+----+----+----+----+----+........+ | file name | b1 | b2 | .. | .. | .. | .. | ...... | +---------------------------+----+----+----+----+----+----+........+
The use of the information stored in a directory entry is: given a file name of a file, we can use the directory entry with the given file name to find the blocks that store the data in the file !
I will illustrate the use of a directory entry in the next section...
-
(I will keep the discussion simple by discussing a flat file system.
Most modern fiel systems are hierachical.
A file is actually identified by a path, e.g.:
C:\users\cheung\myFolder\myFile.
In this kind of file system, the directory tracks are uses to store
the root directory (C:\) and
each sub-folder (sub-directory) is an ordinary file that stores
directory entries.)
- A file is uniquely identified by its
file name
- A (non-empty) file takes up one or more
disk blocks
The content of the file are stored in the disk blocks of the file
- Users can access a file by providing its
file name
in an
"open file"
operation.
This operation locates the directory entry of the file and obtain the disk blocks of the file.
The directory entry is stored in the Operating System area.
The Operating System manages the resourses of the computer.
Examples of Operating Systems:
- Microsoft Windows
- UNIX
- The
disk blocks
in the directory entry
identify the sectors of the disk where the data of the file
are stored.
- Subsequent
file read/write
operations can find the exact location to access the data
store in the file:
