Disk Management

Disk management involves the organization, allocation, and scheduling of disk storage to optimize performance and reliability.

Disk Structure

Physical Structure

                    ┌─────── Spindle
                    │
            ┌───────┴───────┐
            │    Platter    │ ◄── Magnetic disk surface
            │   ┌───────┐   │
            │   │       │   │
            │   │   •   │   │ ◄── Track (concentric circle)
            │   │       │   │
            │   └───────┘   │
            └───────────────┘
                    │
                    ▼
              Read/Write Head
                    │
              Disk Arm (Actuator)

Key Components

Component	Description
Platter	Circular magnetic disk surface
Track	Concentric circle on platter surface
Sector	Smallest unit of storage on a track
Cylinder	Same track on all platters
Head	Read/write mechanism
Arm	Moves heads across platters

Disk Access Time

Total time to read/write data:

Access Time = Seek Time + Rotational Latency + Transfer Time

Component	Description	Typical Value
Seek Time	Time to move head to correct track	3-15 ms
Rotational Latency	Time for sector to rotate under head	2-6 ms
Transfer Time	Time to transfer data	< 1 ms

Formulas

Average Rotational Latency = 1 / (2 × RPM) × 60 seconds

Transfer Time = Data Size / Transfer Rate

Example:

Disk speed: 7200 RPM
Average rotational latency = (1/2) × (60/7200) = 4.17 ms

Disk Scheduling Algorithms

Disk scheduling determines the order in which disk I/O requests are serviced.

Goals

Minimize seek time
Maximize throughput
Ensure fairness
Minimize response time variance

1. FCFS (First Come First Serve)

Concept: Service requests in arrival order.

Example:

Initial head position: 53
Request queue: 98, 183, 37, 122, 14, 124, 65, 67

Head movement: 53 → 98 → 183 → 37 → 122 → 14 → 124 → 65 → 67

Total seek = |53-98| + |98-183| + |183-37| + |37-122| + 
             |122-14| + |14-124| + |124-65| + |65-67|
           = 45 + 85 + 146 + 85 + 108 + 110 + 59 + 2
           = 640 cylinders

Pros: Simple, fair
Cons: Poor performance, high seek time

2. SSTF (Shortest Seek Time First)

Concept: Service request closest to current head position.

Example:

Initial head position: 53
Request queue: 98, 183, 37, 122, 14, 124, 65, 67

Order: 53 → 65 → 67 → 37 → 14 → 98 → 122 → 124 → 183

Total seek = 12 + 2 + 30 + 23 + 84 + 24 + 2 + 59 = 236 cylinders

Pros: Better than FCFS
Cons: Starvation possible for far requests

3. SCAN (Elevator Algorithm)

Concept: Head moves in one direction servicing requests, then reverses.

Example:

Initial head position: 53, moving towards 0
Request queue: 98, 183, 37, 122, 14, 124, 65, 67

Direction: Towards 0 first
Order: 53 → 37 → 14 → 0 → 65 → 67 → 98 → 122 → 124 → 183

Total seek = 16 + 23 + 14 + 65 + 2 + 31 + 24 + 2 + 59 = 236 cylinders

Pros: No starvation, uniform wait time
Cons: Requests at edges wait longer

4. C-SCAN (Circular SCAN)

Concept: Head moves in one direction only, then jumps back to start.

Direction: Towards 199 (disk end)
Order: 53 → 65 → 67 → 98 → 122 → 124 → 183 → 199 → 0 → 14 → 37

Total seek = 12 + 2 + 31 + 24 + 2 + 59 + 16 + 199 + 14 + 23 = 382 cylinders
(Jump from 199 to 0 not counted in some versions)

Pros: More uniform wait time than SCAN
Cons: More head movement

5. LOOK

Concept: Like SCAN but head only goes as far as last request in each direction.

Direction: Towards 0 first
Order: 53 → 37 → 14 → 65 → 67 → 98 → 122 → 124 → 183

(Doesn't go to 0, stops at 14)

Pros: Less head movement than SCAN

6. C-LOOK (Circular LOOK)

Concept: Like C-SCAN but only goes to last request, not disk end.

Direction: Towards high end
Order: 53 → 65 → 67 → 98 → 122 → 124 → 183 → 14 → 37

(Jumps from 183 to 14, skipping 0 and 199)

Pros: Most efficient for uniform access patterns

Algorithm Comparison

Algorithm	Seek Time	Starvation	Variance
FCFS	High	No	High
SSTF	Low	Yes	Medium
SCAN	Medium	No	Medium
C-SCAN	Medium	No	Low
LOOK	Lower	No	Medium
C-LOOK	Lowest	No	Low

RAID (Redundant Array of Independent Disks)

RAID combines multiple physical disks into a logical unit for performance and/or redundancy.

RAID Levels

RAID 0 (Striping)

Disk 1: A1 | A3 | A5 | A7
Disk 2: A2 | A4 | A6 | A8

Data: Striped across disks
Redundancy: None
Performance: Highest (parallel read/write)
Capacity: 100% (N disks)
Risk: Any disk failure = all data lost

RAID 1 (Mirroring)

Disk 1: A1 | A2 | A3 | A4
Disk 2: A1 | A2 | A3 | A4  (copy)

Data: Mirrored on two disks
Redundancy: Full copy
Performance: Read improved, write same
Capacity: 50% (N/2 usable)
Fault Tolerance: Survives 1 disk failure

RAID 5 (Striping with Distributed Parity)

Disk 1: A1 | A4 | A7 | Pb
Disk 2: A2 | A5 | Pa | A10
Disk 3: A3 | Pc | A8 | A11

Data: Striped with parity distributed
Redundancy: Parity information
Performance: Good read, slower write
Capacity: (N-1) disks
Fault Tolerance: Survives 1 disk failure
Minimum Disks: 3

RAID 6 (Striping with Double Parity)

Disk 1: A1 | A4 | Pa | Qb
Disk 2: A2 | Pb | A7 | A10
Disk 3: A3 | Qa | A8 | A11
Disk 4: Pa | A5 | A9 | Pc

Fault Tolerance: Survives 2 disk failures
Capacity: (N-2) disks
Minimum Disks: 4

RAID 10 (1+0, Mirrored Stripes)

       ┌─────────────┐
       │   RAID 0    │
       │  (Stripe)   │
       └──────┬──────┘
              │
    ┌─────────┴─────────┐
    │                   │
┌───┴───┐           ┌───┴───┐
│RAID 1 │           │RAID 1 │
│Mirror │           │Mirror │
└───┬───┘           └───┬───┘
    │                   │
  ┌─┴─┐               ┌─┴─┐
  D1  D2              D3  D4

Combines: RAID 0 + RAID 1
Performance: Best for read/write
Capacity: 50%
Fault Tolerance: Can lose 1 disk per mirror

RAID Comparison

RAID	Min Disks	Capacity	Fault Tolerance	Performance
0	2	100%	None	Highest
1	2	50%	1 disk	Good read
5	3	(N-1)/N	1 disk	Good
6	4	(N-2)/N	2 disks	Moderate
10	4	50%	1 per mirror	Best

Disk Formatting

Low-Level (Physical) Formatting

Divides disk into sectors
Creates sector headers and trailers
Done at factory
Sets up error-correcting codes (ECC)

Partitioning

Divides disk into logical regions
Each partition treated as separate disk
Partition table stored in first sector

High-Level (Logical) Formatting

Creates file system structures
Sets up root directory
Initializes free space management

Boot Process

BIOS/UEFI executes from ROM
Reads MBR/GPT from first disk sector
Boot loader (GRUB, Windows Boot Manager) loads
Boot loader loads OS kernel
Kernel initializes and starts services

MBR vs GPT

Feature	MBR	GPT
Max disk size	2 TB	9.4 ZB
Max partitions	4 primary	128
Redundancy	None	Backup at disk end
Boot mode	BIOS	UEFI

Swap Space Management

Swap space is disk area used as virtual memory extension.

Uses

Store inactive pages
Enable larger programs than physical RAM
Handle memory pressure

Allocation Methods

Swap partition: Dedicated disk partition
Swap file: File on regular file system

Linux Commands

# View swap status
swapon --show
free -h

# Create swap file
dd if=/dev/zero of=/swapfile bs=1G count=4
chmod 600 /swapfile
mkswap /swapfile
swapon /swapfile

Important Interview Questions

Q1: What is the difference between SCAN and C-SCAN?

A: SCAN reverses direction at disk ends, servicing requests both ways. C-SCAN only services in one direction, jumping back to start without servicing (provides more uniform wait times).

Q2: Why is SSTF not always used?

A: SSTF can cause starvation for requests far from current head position if new nearby requests keep arriving.

Q3: What is the purpose of RAID?

A: RAID provides data redundancy (protection against disk failure) and/or performance improvement by combining multiple disks.

Q4: Difference between RAID 0 and RAID 1?

A: RAID 0 stripes data for performance but has no redundancy. RAID 1 mirrors data for redundancy with 50% capacity.

Q5: What is seek time?

A: Seek time is the time required to move the disk head to the correct track/cylinder.

Quick Reference

Disk Access Time = Seek + Rotational Latency + Transfer

Algorithm preference:
├── General: LOOK or C-LOOK
├── Random access: SSTF
└── Uniform workload: C-SCAN

RAID selection:
├── Performance only: RAID 0
├── Redundancy only: RAID 1
├── Balance: RAID 5
├── Critical data: RAID 6 or RAID 10