Mass Storage Structure

Mass Storage Structure

1. Disk's Physical Structure

   • Transfer rate
     • Rate of data flow from disk to computer
   • Positioning Time
     • Time to move disk arm to cylinder a.k.a seek time
   • Av. Rotational Latency
     • 1 / (2 x rotational speed)
   • Head crash
     • Disk head making contact with the disk surface
   • Effective Access Time = seek time + rotational latency + transfer time
   • Logical blocks
     • 1-D array of blocks mapped onto disk sectors (tracks) sequentially
       • Mapped from outermost cylinder inwards
   • Disk bandwidth
     • Total number of bytes transferred / total time between first request and completion

2. Disk Requests' Scheduling Algorithms

   • Disk Scheduling
     • Order pending disk requests with respect to head and rotational POSITION - is it close or far?
       • Minimise seek and latency times
         • Allow OS to switch algorithms
   • Types
     • First Come-First Serve
       • Fair
       • Poor performance for heavy loads
     • Shortest Seek Time First
       • Issue of starvation
       • Discriminates inner and outermost tracks - Unfair
     • SCAN
       • Disk arm starts at one end, and moves to the other, servicing requests
         • Long delays for requests at outer locations
       • C-Scan
         • When end reached, skips back to beginning
         • C-Look
           • Reaches last request then goes to start without going to end of disk
           • This and SSTF good default algorithms
         • Good for heavy loads

3. Disk Management Issues

   • Management
     • Low-level formatting
     • Physical formatting
     • Dividing disk into sectors for disk controller
     • Disk Sector Info
       • Preamble
         • e.g. sector number
       • Data
       • ECC - Error Correction Code
   • Redundant Array of Inexpensive Disks (RAID)
     • 1. Uses array of physical disk drives
       • Appear as single virtual disk drive
     • 2. Store data from arrays
       • Allows parallel operation (disk striping)
         • Spreading blocks across multiple storage devices
     • 3. Redundant disk capacity used to respond to disk failure
       • Reliability via redundancy
     • Mirroring/Shadowing
       • Keeps duplicate of each disk
     • Parity Strip
       • Stored in each disk with striped data
         • Rebuilds data in event of data loss
       • Fast + fault tolerant
       • Used in RAID 2+
         • RAID 5 - Incorporates into storage disks

4. OS Services for Mass Storage

Images and files in the map