Device Management - Chapter 7

Device Management - Chapter 7

• The chapter covers different types of devices, including dedicated, shared, and virtual devices, and how they compare.
• It also explains how blocking and buffering improve input/output (I/O) performance.
• Seek time, search time, and transfer time calculation methods are described for various device types.
• Discusses how different device access times vary and highlights the strengths and weaknesses of various seek strategies.
• Explores different RAID levels and how they vary from each other.

Types of Devices

• Dedicated Device:
  • Assigned to one job at a time (e.g., tape drives, printers, plotters).
  • Inefficient if not used constantly.
• Shared Device:
  • Assigned to multiple processes concurrently, as in the process of printing from multiple systems.
  • Processes share the device simultaneously.
  • Requests are interleaved by the device manager.
• Virtual Device:
  • Combination of dedicated and shared devices.
  • Can be converted to sharable devices using spooling techniques.
  • Example includes converting a dedicated printer into a shared one using a queuing system.

Sequential Access Storage Media

• Magnetic Tape:
  • Routine secondary storage in early computer systems.
  • Records are stored serially, with a determined length by the application program.
  • Record access is time-consuming as tape rotates until the head reaches the desired record for reading/writing.
• Sequential access storage media properties
  • Record access based on position on tape
  • Tape rotates passing under read/write head
  • Record length determined by application
  • Record identified by position on tape.
  • Time-consuming process.
  • Example of device: tapes

Sequential Access Storage Media - Continued

• Tape Density:
  • Characters or bytes recorded per inch of tape; impacted by storage method (individual or blocked records).
• Interrecord Gap (IRG):
  • Gap inserted between records, necessary for the tape to stop before receiving the next record.
  • Size is consistent regardless of record size.
• Interblock Gap (IBG):
  • Inserted between blocks of records; more efficient than having an IRG between individual records.

Blocking Advantages and Disadvantages

• Advantages:
  • Reduced I/O operations-one READ instruction moves an entire block
  • Less space used on tape
• Disadvantages:
  • Overhead software routines needed for blocking and record-keeping
  • Waste of buffer space

Direct Access Storage Devices

• Random access storage devices: read and/or write directly to specific areas.
  • E.g., magnetic disks, optical discs, solid-state memory
• Access time variance:
  • Access time depends on where a data record is located on the device
• Magnetic Disk Storage:
  • Concentric, number tracks from outside-in
  • Read/write heads move in unison
• Accessing records: - Needs cylinder number, surface number, and sector number.

Access Times

• File Access Time Factors:
  • Seek time (moving the read/write head)
  • Rotational delay (waiting for the desired data to rotate to the read/write head)
  • Search time involved in determining where the record is located is critical to speed and efficiency
  • Transfer time (actually reading/writing the data)

Fixed-Head Magnetic Drives

• Record access with track number and record number
• Total access time search time + transfer time
• Three basic positions for requested records relative to read/write head position
• Blocking minimizes access time.

Movable-Head Magnetic Disk Drives

• Access time seek + search + transfer time
• Strategies use calculations for seek and transfer time.

Device Handler Seek Strategies

• Methods (e.g., first-come, first-served (FCFS), shortest seek time first (SSTF), SCAN, C-SCAN, LOOK) that define seek order.
• Goal minimizes seek time, arm movement, mean response time.
• Methods include:
  • FCFS
  • SSTF
  • SCAN
  • LOOK
  • N-Step SCAN
  • C-SCAN
  • C-LOOK

Optical Disc Storage

• Design Features:
  • Single spiral track, concentric tracks
  • Same sized sectors
  • Constant linear velocity (CLV)
• Performance Measures:
  • Data transfer rate (Mbps): speed of reading large data amounts
  • Average access time: time needed to locate and access a random data item.
• Additional features:
  • Data recorded/retrieved consistently
  • Data storage in sectors
  • Two-sided/single-sided storage

CD and DVD Technology

• Data recording: zeros and ones, pits, and lands, light striking land reflects to photodetector
• Reads with low-power laser.
• Light strikes land: reflected light to photodetector
• Light strikes pit: scattered and absorbed light
• Photodetector: converts light intensity into digital signal.
• CD-R: write-once technology, requires expensive controller
• DVD: several discs with greater capacity
• Rewritable disks use phase-change technology.

Solid State Storage

• Implements Fowler-Nordheim tunneling phenomenon.
• Stores electrons in floating gate transistor.
• Electrons remain after power off.
• Typical in flash memory storage.

Flash Memory Storage

• Electrically erasable, programmable, read-only memory (EEPROM).
• Data written as electric charge sent through floating gate.
• Erasing data uses a strong electrical field.

Solid State Drives

• Fast storage devices (expensive)
• Typical device functions in smaller physical space than magnetic drives.
• Work electronically; no moving parts.
• Require less power—silent, and relatively lightweight.
• Disadvantages: disastrous crashes, data degradation over time.

Components of the I/O Subsystem

• Input/output channels
• Programmable units
• Located between CPU and control unit
• Synchronizing device speeds (CPU with I/O device)
• Managing concurrent CPU and I/O device requests for faster processing; handles overlap.
• Input/output control unit
• Input/output channel program
• Disk controller
• Input/output subsystems

Communication Among Devices

• Resolves problems in communication among devices.
• Hardware flag tested by CPU, channel status word (CSW) contains flag and three bits in flag represent I/O system component.
• Communication links between channels, control units, and devices
• Speed disparities between CPU and I/O, handled by buffers.
• Direct memory access (DMA) allows control units to access main memory directly without CPU intervention.
• Double buffering improves data flow between slow I/O devices and fast CPUs.
• Multiple paths increase flexibility and reliability.

RAID

• RAID levels provide different degrees of redundancy.
• RAID 0: data striping, non-critical, devices appear as one logical unit.
• RAID 1: mirroring (backup), expensive.
• RAID 2: small strips, expensive, Hamming code.
• RAID 3: striping, parity bit, one disk for redundancy.
• RAID 4: simple scheme, computes parity, parity and data stored in different strips.
• RAID 5: data striping and parity are distributed across multiple disks.
• RAID 6: provides redundancy, requires two parity bits.
• Nested RAID Levels: combine different levels.