Please enable JavaScript.

Coggle requires JavaScript to display documents.

Mass-Storage Systems (RAID structure (Six levels (Block intervealed parity…

- - - - Failure
      - repair
      - Data loss
  - - - RAID 1
    - - RAID 1+0
    - - RAID 0+1
    - - RAID 4
      - RAID 5
      - RAID 6
  - - - view of system before set of changes
- - - - EIDE
      - ATA
      - SATA
      - USB
      - Fibre
- - - - 24-bit address space
        
        Basis of storage area networks
  - - - Provide features to attached hosts
        
        Ports to connect
        
        Memory control software
        
        RAID, hot swap, spares
        
        Shared storage
        
        File system features
  - - - multiple storage arrays
    - - connected to switchs
  - - - TCP/UDP on IP network
- - - - implies
        
        Fast access time
        
        Disk bandwith
        
        Total number of bytes transferred/
        total time between first request and completion
        
        Minimize seek time
        
        Seek time = seek distance
  - - - input
      - output
      - disk address
      - memory address
      - number of sectors
  - - - Shortest seek time first
        
        request with minimum seek time
        
        starvation
    - - The disk arm starts at one end
      - Moves toward the other end
        
        servicing requests
        
        gets to end
        
        reverses
      - elevator algorithm
    - - More unifrom time than SCAN
        
        Head moves from one end to other
        
        Servicing as it goes
        
        when the end is reached
        
        returns to beginning without servicing
    - - Version of C-SCAN
        
        Arm only goes as far as the last request
        
        each direction
        
        reverses inmediately
- - - - with independent failure modes
    - - ensure
        
        We can recover stable data
  - - - Sucessful competion
      - Partial failure
        
        Midst transfer
        
        only some sectors were written
        
        Corruption
      - Total failure
        
        Before disk write
  - - - 2 physical blocks per logical
        
        Write 1st physical
        
        when succesful
        
        Write 2nd physical
        
        declare complete only on second
- - - - each sector holds
        
        Header
        
        Data
        
        Error correction
        
        Sector sparring
        
        handles bad blocks
  - - - logical disk
      - Logical formatting
        
        File system
      - Blocks -> clusters
- - - - Carved out of normal file system
      - Separate disk partition
        
        raw
    - - Swap maps
        
        Track swap-space usage
        
        some systems allow multiple swap spaces
- - - - composed of logical blocks
        
        Smallest unit of transfer
        
        Low-level formatting creates logical blocks
      - Sectors
        
        sector 0
        
        First sector
        
        Outmost cylinder
        
        Mapping proceeds in order
        
        From outermost to innermost
        
        Mapping logical to physical
        
        Should be easy
        
        Bad sectors
        
        Non constant number
        
        angular velocity