RAID (cont.)

• increases mean time to failure
  • chance that a disk out of $N$ disks fail: much higher than one specific disk failing
  • if: mean time to failure of a single disk is 100,000 hours
    • mean time to failure in 100 disks: 100,000 / 100 = 1,000 hours = 41.66 days!
• data loss rate: unacceptable if only 1 copy of data is stored
  • solution: introducing redundancy
  • simply & most expensive approach: mirroring
    • duplicate every disk
  • every write: carried out on two physical disks
    • data: lost only if the second disk fails before first failed disk gets replaced
• mean time to repair

RAID performance improvement

• parallelism (disk system) via data striping: two main goals
  • increase: throughout of multiple small access by load balancing
  • reduce: response time of large access
• bit-level striping
  • array of 8 disks: can be treated as a single disk
    • w/ each sectors are 8 times the normal sector size
  • bit level striping
• block-level striping
  • blocks of a file: stripes across multiple disks
  • with $n$ disks: block $i$ of a file goes to $(i\mathrm{mod}n)+i$

RAID structure

• mirroring: providing high reliability, but expensive
• striping: providing high data-transfer rates, not providing reliability
• 

RAID (0+1) and (1+0)

• both striped mirrors (RAID 1+0) or mirrored stripes (RAID 0+1):
  • provides high performance (RAID 0)
  • and high reliability (RAID 1)
    • 👨‍🏫 can also rely on error correcting code!
• (RAID 0+1)
• (RAID 1+0): drives are mirrored in pairs, and
• 

Other features

• regardless of what RAID implemented:
  • useful features can be added at each level
• snapshot: view of file before the last update took place
  • for recovery
• replication
• hot spare: dis

File concept

• contiguous logical address space
• types
  • data
    • numeric
    • character
    • binary
  • program
• content defined by the file creator
  • many types: e.g. test, source, and executable file

File attributes

• name: information kept in human readable form
• identifier
• type: needed by systems that support different types
  • 👨‍🏫 in UNIX/Linux: a positive integer
• location: pointer to file location on device
  • 👨‍🏫 can be very complex
  • depends on: how disk space is allocated to the file!
• size: current file size
• protection
• time, date, and user identification: data for protection, security, and usage monitoring

• information about files: kept in directory structure
  • maintained on the disk
    • part of which: currently in use can be cached in main memory
  • many variations: extended file attributes like file checksum

File operations

• file: an abstract data type (ADT)
• create:
• write:
• read:
• reposition within file: seek
  • 👨‍🏫 not an issue unless you use magnetic tape
• delete:
• truncate:
• open:
• close:
• such operations: involve changes of various OS kernel data structures

Open files

• several data structures: needed to manage open files
  • open-file tables: tracks open files, system-wide open-file table
    • and per-process open-file table
  • file pointer: pointer to last read / write location
    • per process that has the file open
  • file-open count: counting the no. of processes that the file has been opened
    • allowing: removal of data from open-file table
      • when the last processes close it (when file-open count is 0)
  • disk locations of a file: cache of data access information
  • access rights: per-process access mode information

File types

Access methods

• sequential access: simplest access method

  read next
  write next
  reset 
  // no read after last write (rewrite)
  

• direct access: file is fixed length logical records

  read n
  write n
  position to n
      read next
      write next
  rewrite n
  

  • n: relative block number
• relative block numbers: allows OS to decide where file should be placed
  • more in disk block allocation problem

Other access methods

• other file access methods: can be build on direct-access
• generally: involve creation of an index for a file
• keep index in memory for fast location of data
• IBM
• 

Directory structure

• collection of nodes containing information about all files
• 
• doth directory structure & files: reside on disk

Disk structure

• disk: can be subdivided into partitions
• disks / partitions: can be RAID protected against failure

• partitions: aka minidisks / slices
• volume: entity on a disk containing a file system
  • each volume containing a file system: keeps track of the file system into
    • in device directory or volume table of contents
• other than general purpose file systems, many special purpose file systems exist
  • frequently within the same OS / computing system
  • 👨‍🏫 e.g. backup...

Typical file-system organization

Operations performed on directory

• search for a file
• create a file
• delete a file
• list a directory
• rename a file
• traverse the file system

Organize the directory (logically) to obtain..

• efficiency: locating a file quickly
• naming: convenient to users
  • two users: might have same name for different files
  • same file: can have several different names
• grouping: logical grouping of files by properties
  • e.g. all Java programs, games, COMP 3511 ...

Single-level directory

• single directory for all users
• 

Two-level directory

• e.g. separate directory for each user
• 
• a path name is required
  • to identify files accurately (e.g. /user1/cat)
• same file name may exist under different users
• more efficient searching than single-level directory
  • yet no grouping capability

Tree-structured directory

• 
• efficient searching & grouping capability
• current directory (working directory)

  cd /spell/mail/prog
  type list
  

• path name: either absolute or relative
  • 👨‍🏫 absolute: always unique
• creating a new file: done in the current directory
• delete a file in the current directory
  • rm <file-name>
• creating a new subdirectory: done in the current directory
  • mkdir <file-name>
• 👨‍🏫 almost perfect... but it doesn't support shared file!

Acyclic-graph directories

• 👨‍🏫 cycle: must be avoided as we might fall into infinite loop when we traverse!
• now: we have shared subdirectories & files
  • more flexile and complex
• 
• new directory entry type
  • link: another name (pointer) to an existing file
  • resolve the link: follow pointer to locate the file
• alias: 2 different path names
  • ensure: traversing shared structure more than once
  • deletion of one: might lead to dangling pointers
    • pointing to empty files, or even wrong files
• also: difficulty ensuring there is no cycle in graph
  • either ban directory sharing (inconvenient), or run cycle detection per every new link (time consuming)
  • 👨‍🏫 no free lunch!
• 

File system mounting

File sharing

Protection

• file owner / creator: should be able to control
  • what can be done
  • by whom
• types of access
  • read
  • write
  • execute
  • as well as...
    • append
    • delete
    • list

Access lists and groups

• mode of access: read, write , execute
  • a 3-bit integer, or 0-7 in decimal
  • when bit=1: has permission
• three classes of users: on UNIX / Linux
  1. owner
  2. group
  3. general / public
  • thus, 3 3-bit integer to specify permission
• sample
  • starts w/ d if directory, - otherwise
  • -rw-rw-r--: file w/ 664 permission
  • drwxrwxrwx: directory w/ 777 permission

File system structure

• disk: provides most of second storage
  • on which: file systems are maintained
• two characteristics: make disk convenient for this usage
  1. can be rewritten in place
     • can be read from block, and write it back to same place after modification
  2. disk: can access directly any block of info. it contains
     • simple to access any file either sequentially / randomly
     • switching from one file to another: requires only moving read-write heads & waiting for disk to rotate
• to improve IO efficiency: IO transfers between memory & disk: done in units of blocks
  • each block: 1 or more sectors
  • sector size: varies from 32 bytes to 4KB
    • usually 512 bytes
• file structure
  • logical storage unit
  • collection of related information
• file system: resides on secondary storage (hard drive / disks)
  • provides: efficient & convenient access to disk
    • by allowing data to be: stored, located, and retried easily
  • provides: UI: file and file attributes, operations on files, directory for organizing filed
  • provides: data structure and algorithms for:
    • mapping logical file system onto physical secondary storage devices
• file systems: organized into different layers

Layered file system

• 👨‍🏫 multi layer file system: one (not only) way

application programs 
=> logical file system
=> file-organization module
=> basic file system
=> IO control
=> devices

File system layers

• IO control & device drivers: manage IO devices at the IO control layer
• basic file system: issues generic commands to the appropriate device driver to read & write physical blocks of dick
  • caches: hold frequently used file-system metadata to improve performance
• file organization module: knows files & their logical blocks
  • as well as physical blocks
  • translates logical block addr. to physical block addr.
    • pass this to basic file system for transfer
  • managers free disk space
    • disk block allocation
• logical file system: manages metadata information
  • metadata: includes all of the file-system structure
    • except the actual data (i.e. contents of file)
  • managing: directory structure to provide: info needed by the file-organization module
  • translates: file name into file no. / file handle / location
    • by maintaining file control blocks
  • file control block (FCB) (aka inode in UNIX)
    • contains all info about file, including:
      • ownership, permissions, location of the file contents (on the disk)
    • also responsible for file protection