L28_Ch11c

Chapter 11: I/O Management and Disk Scheduling

Overview

• This chapter discusses I/O management and disk scheduling, focusing on RAID (Redundant Array of Independent Disks) and disk cache management.

11.6 RAID

Definition and Characteristics

• RAID consists of seven levels (0 through 6) that combine multiple disk drives into a single logical unit.

• Key characteristics include:

  • Data distributed across the physical drives in a method called striping.

  • Redundant disk capacity used for parity information ensures recoverability if a disk fails.

Historical Context

• The term RAID was introduced by researchers from UC Berkeley.

• Initially defined configurations for improving data access and redundancy.

Benefits

• Simultaneous access to data improves I/O performance.

• Provides redundancy through stored parity, enabling recovery from disk failures.

RAID Levels

Level 0: Non-redundant

• Data is distributed across all disks without redundancy.

• Achieves high performance, but no data protection.

Level 1: Mirrored

• Duplicates all data across two drives.

• Eliminates write penalties; second drive provides data access during a failure.

• High cost due to using double the storage capacity.

Level 2: Redundancy through Hamming Code

• Implements parallel access and data striping.

• Utilizes Hamming code for error correction; effective in high-error environments.

Level 3: Bit-interleaved Parity

• Requires one redundant disk for the entire array.

• Provides parallel access, enabling high data transfer rates.

Level 4: Block-level Parity

• Constructs a parity strip across each disk, affecting small write requests.

• Involves a write penalty due to parity bit calculation.

Level 5: Distributed Block Parity

• Distributes parity across all disks, enhancing data availability.

• Loss of one disk does not lead to data loss.

Level 6: Dual Redundancy

• Utilizes two parity calculations for extremely high data availability.

• Each write impacts two parity blocks, resulting in a write penalty.

RAID Levels Overview

11.7 Disk Cache

Definition

• Cache memory is a smaller, faster memory location between main memory and the processor, reducing average access time.

• Disk cache serves as a buffer for disk sectors, containing copies of sectors for quick access.

Functionality

• Upon an I/O request, the system checks the disk cache first.

• If the requested sector is found in the cache, it is accessed directly; otherwise, it is read from the disk into the cache.

Replacement Strategies

• Least Recently Used (LRU): Replaces the block that has been in the cache the longest without being referenced.

• Least Frequently Used (LFU): Replaces the block with the fewest accesses; maintains a counter for each block.

Performance Evaluation

• LRU strategy is often effective, and its performance can be assessed through various cache sizes and miss rates, as illustrated in figures 11.10 and 11.11.

Summary of Disk Cache Management

• Disk cache optimizes I/O performance through effective memory storage solutions.

• LRU and LFU are critical algorithms for managing cache blocks efficiently.