9.1 Network Security Foundations: Confidentiality, Integrity & Availability

Back to Computer Networks Notes

📚Module 09: Network Security & CryptographyTopic 9.1⏱️15 min read

🎯High-Yield For:Semester Exams (All Universities) • Placement Technical Rounds • Engineering Foundations

💡 Core Intuition: The Royal Diplomatic Pouch & Wax Seal

🍳 The Everyday Analogy: Secure Diplomatic Courier

Imagine an ancient king dispatching a vital peace treaty across hostile territory to a neighboring monarch:

1. Confidentiality (The Locked Pouch): The treaty is written in a secret scrambled dialect and locked inside an opaque iron box so enemy highwaymen intercepting the pouch cannot read its terms.
2. Integrity (The Unbroken Wax Seal): The box is sealed with molten wax stamped with the king's signet ring. If a spy tampers with the lock or alters the wording, the wax seal fractures, alerting the receiving monarch instantly.
3. Availability (The Escorted Road): The courier is protected by armed cavalry along fortified roads. If enemy forces blow up the mountain pass to prevent the message from arriving, the treaty cannot be ratified (Denial of Service).
4. Authenticity & Non-Repudiation (The Royal Signature): The treaty is signed by the king's own hand. He cannot later claim: "I never authorized this peace treaty; an imposter sent it."

💻 Bridging to Computer Science

In computer networks:

• The Internet was originally engineered for academic cooperation, with zero built-in security. Routers blindly forward IP packets without verifying who created them or whether their contents were altered.
• Network Security introduces mathematical and architectural controls across the protocol stack to ensure that data remains private (Confidentiality), uncorrupted (Integrity), accessible to authorized users (Availability), and verifiably linked to legitimate senders (Authenticity & Non-Repudiation).

---

📑Table of Contents

• 🛡️ The Pillars of Network Security: The CIA Triad
• ⚔️ Threat Taxonomy: Passive Attacks vs Active Attacks
• 🌐 The Adversary Model: The Dolev-Yao Threat Model
  • Assumptions of the Dolev-Yao Model
  • The Cryptographic Boundary
• 📜 Kerckhoffs's Principle: Why Obscurity is Not Security
• 🔑 Cryptographic Services Taxonomy
• 🏭 Real-World Production Context: Zero Trust Architecture (BeyondCorp) & Enterprise Threat Models
  • 1. The Death of the Perimeter: Google BeyondCorp & Zero Trust
  • 2. Modeling the Dolev-Yao Adversary in Global Banking (SWIFT)
  • 3. High-Scale Availability Engineering (DDoS Scrubbing & CDN Shields)
• 📐 Numerical Applications & Worked Examples
  • Example 1: Exhaustive Brute-Force Key Search Complexity
  • Example 2: Availability Calculation & Service Level Agreement (SLA) Downtime
  • Example 3: IPsec ESP Tunnel Mode Encapsulation Overhead
• 🎯 Exam & Interview Pitfall Check

---

🛡️ The Pillars of Network Security: The CIA Triad

The cornerstone of information security is defined by the CIA Triad, augmented by Authenticity and Non-Repudiation.

The Core Security Objectives (CIA Triad & Supporting Services)

The essential security guarantees required for trustworthy communication over untrusted networks

🔒

Confidentiality

Secrecy & Privacy

• Guarantees that transmitted data is unintelligible to unauthorized interceptors.
• Defends against packet sniffing, eavesdropping, and wiretapping.
• Primary Mechanisms: Symmetric encryption (AES) and Asymmetric encryption (RSA/ECC).

🛡️

Integrity

Tamper Detection

• Ensures that data cannot be modified, deleted, inserted, or reordered in transit without detection.
• Defends against man-in-the-middle bit-flipping and packet injection.
• Primary Mechanisms: Cryptographic hash functions (SHA-256) and HMACs.

⚡

Availability

Operational Uptime

• Ensures that network services, bandwidth, and resources remain accessible to authorized entities.
• Defends against Denial of Service (DoS) and Distributed DoS (DDoS) botnet floods.
• Primary Mechanisms: Anycast routing, rate limiting, firewalls, and redundant clusters.

🔏

Authentication & Non-Repudiation

Identity & Accountability

• Authentication confirms the true identity of the communicating endpoints (PKI/X.509).
• Non-repudiation prevents a sender from falsely denying they authored a transmitted message.
• Primary Mechanisms: Asymmetric Digital Signatures and Public Key Infrastructure.

---

⚔️ Threat Taxonomy: Passive Attacks vs Active Attacks

Network vulnerabilities are categorized based on whether the adversary merely inspects packets or actively interferes with transit traffic.

Passive Attacks vs Active Attacks

Contrasting silent observation against intrusive traffic manipulation

Passive Attack

Passive Attacks (Eavesdropping)

👂

Dominant Architecture / DomainConfidentiality Violation (Hard to Detect, Easy to Prevent)

• •The attacker monitors, sniffs, or inspects packets without modifying their payload.
• •Packet Sniffing: Capturing plaintext credentials over unencrypted Wi-Fi using Wireshark.
• •Traffic Analysis: Observing packet timing, source/destination IPs, and message lengths to deduce operational patterns.
• •Detection Strategy: Almost impossible to detect because network transit is unaltered.
• •Defense Strategy: Strong end-to-end encryption (TLS, IPsec).

"Prevention is the primary defense; detection is virtually impossible."

Active Attack

Active Attacks (Tampering & Disruption)

⚔️

Dominant Architecture / DomainIntegrity & Availability Violation (Easy to Detect, Hard to Prevent)

• •The attacker actively injects, alters, delays, replays, or deletes data in transit.
• •Masquerade / Spoofing: Forging IP or MAC source addresses to impersonate legitimate hosts.
• •Message Modification: Altering financial transaction amounts or routing headers.
• •Replay Attack: Capturing a valid authentication packet and re-transmitting it later.
• •Denial of Service (DoS): Flooding server sockets to exhaust memory or bandwidth.
• •Defense Strategy: Cryptographic integrity (HMAC), sequence nonces, and stateful firewalls.

"Detection and containment are the primary defense strategies."

---

🌐 The Adversary Model: The Dolev-Yao Threat Model

When designing cryptographic protocols, computer scientists rely on the Dolev-Yao Threat Model (formulated by Danny Dolev and Andrew Yao in 1983).

Assumptions of the Dolev-Yao Model

Under this model, the network is assumed to be completely controlled by an active adversary:

1. Eavesdrop: The attacker can read every packet traversing the wire.
2. Intercept & Drop: The attacker can block legitimate packets from reaching their destination.
3. Modify: The attacker can alter arbitrary bits inside captured packets before forwarding them.
4. Inject: The attacker can craft and inject arbitrary packets with fake source and destination addresses.
5. Reorder & Delay: The attacker can hold packets and deliver them out of order at arbitrary future times.

The Cryptographic Boundary

Crucially, the Dolev-Yao model assumes that cryptographic primitives are computationally unbreakable:

• The attacker cannot decrypt ciphertext without the secret key.
• The attacker cannot invert a one-way hash function.
• The attacker cannot forge a valid digital signature without the private signing key.

Key Architectural Lesson: If a protocol relies on the network "behaving nicely" or assumes packets cannot be intercepted, it is fundamentally insecure. A protocol must prove mathematically that it remains secure even when an adversary controls every intermediate router!

---

📜 Kerckhoffs's Principle: Why Obscurity is Not Security

In 1883, military cryptographer Auguste Kerckhoffs stated the foundational axiom of modern cryptography:

Kerckhoffs's Principle: "A cryptosystem should be secure even if everything about the system, except the key, is public knowledge."

In 1949, Claude Shannon restated this axiom as Shannon's Maxim: "The enemy knows the system."

Dimension	Security Through Obscurity	Open Cryptographic Design (Kerckhoffs)
Premise	Security relies on keeping the algorithm, source code, or hardware design secret.	Security relies exclusively on the secrecy and mathematical entropy of the Key.
Vulnerability	If a reverse-engineer decompiles the binary or an employee leaks the algorithm, the entire system collapses.	The algorithm (AES, RSA, SHA-256) is completely open-source and scrutinized by thousands of mathematicians.
Real-World Examples	Proprietary DVD encryption (CSS cracked in 1999); GSM A5/1 cipher (broken).	TLS 1.3, AES-256, RSA-4096 (mathematically vetted over decades).

---

🔑 Cryptographic Services Taxonomy

Network protocols implement security by orchestrating four mathematical primitives:

Security Goal	Mathematical Primitive Used	Protocol Examples
Confidentiality	Symmetric Encryption (AES, ChaCha20)	TLS Application Data, IPsec ESP, SSH
Integrity & Origin Authentication	Hash-based Message Authentication Codes (HMAC)	TLS Record Layer, IPsec AH
Endpoint Authentication & Non-Repudiation	Asymmetric Digital Signatures (RSA, ECDSA)	TLS Handshake, DNSSEC, S/MIME, PGP
Key Agreement	Ephemeral Diffie-Hellman Key Exchange (ECDHE)	TLS 1.3 Handshake, SSH Key Exchange

---

🏭 Real-World Production Context: Zero Trust Architecture (BeyondCorp) & Enterprise Threat Models

1. The Death of the Perimeter: Google BeyondCorp & Zero Trust

For decades, corporate IT relied on the "Castle-and-Moat" security paradigm: once an employee connected to the office VPN or plugged into a wall jack, the internal network was completely trusted.

• In 2009, advanced persistent threats (Operation Aurora) breached internal corporate networks, moving laterally without resistance.
• In response, Google pioneered Zero Trust Architecture (BeyondCorp / NIST SP 800-207):
  • Principle: "Never Trust, Always Verify".
  • The physical network is treated as hostile (untrusted public Internet).
  • Every device and service must authenticate and authorize every single request: microservices communicate using Mutual TLS (mTLS) with ephemeral short-lived certificates rotated every few hours.
  • Access to internal applications is gated through an Identity-Aware Proxy (IAP) inspecting user identity, multi-factor authentication (MFA), and real-time device security posture.

2. Modeling the Dolev-Yao Adversary in Global Banking (SWIFT)

In financial transaction networks like SWIFT (Society for Worldwide Interbank Financial Telecommunication):

• System architects operate strictly under the Dolev-Yao threat model: they assume state-sponsored adversaries physically control undersea cables, compromise BGP routes, and intercept every packet.
• Relying on network link security is unacceptable. Every payment instruction is cryptographically signed using hardware security modules (HSMs) with non-repudiation digital signatures and validated end-to-end at the application layer before ledger debiting occurs.

3. High-Scale Availability Engineering (DDoS Scrubbing & CDN Shields)

Securing Availability ($A$ in the CIA Triad) against multi-terabit volumetric DDoS attacks (e.g. Mirai botnets or DNS amplification floods):

• Enterprise organizations route their external public IP prefixes through global Cloud DDoS Scrubbing Centers (Cloudflare Magic Transit, AWS Shield Advanced).
• Incoming traffic passes through BGP Anycast scrubbing centers that filter out spoofed UDP floods and SYN packets using hardware eBPF/XDP filters in milliseconds, forwarding only sanitized HTTP traffic across private interconnects to origin application servers.

---

📐 Numerical Applications & Worked Examples

Example 1: Exhaustive Brute-Force Key Search Complexity

Problem Statement: A supercomputer cluster can test $10^{14}$ cryptographic keys per second ($100\text{ Trillion keys/sec}$).

1. Calculate the maximum time required to brute-force a legacy 56-bit DES key.
2. Calculate the maximum time required to brute-force a standard 128-bit AES key.
3. If an adversary harnesses 1 billion such supercomputers simultaneously, determine how many years are required to crack the 128-bit key, explaining the physical reality of modern key lengths.

Step-by-Step Solution:

1. 56-Bit DES Brute-Force Time:
   • Total key space: $K_{\text{DES}} = 2^{56} \approx 7.2057 \times 10^{16}\text{ keys}$
   • Time to test entire key space: $T_{\text{DES}} = \frac{7.2057 \times 10^{16}\text{ keys}}{10^{14}\text{ keys/s}} \approx 720.57\text{ seconds} \approx \mathbf{12.01\text{ minutes!}}$
   • On average, the key is found in half this time ($\approx 6\text{ minutes}$). This mathematical proof explains why 56-bit DES was completely retired in the late 1990s.
2. 128-Bit AES Brute-Force Time (Single Supercomputer):
   • Total key space: $K_{\text{AES-128}} = 2^{128} \approx 3.4028 \times 10^{38}\text{ keys}$
   • Time required: $T_{\text{AES-128}} = \frac{3.4028 \times 10^{38}\text{ keys}}{10^{14}\text{ keys/s}} \approx 3.4028 \times 10^{24}\text{ seconds}$
   • Converting to years ($1\text{ year} \approx 3.1536 \times 10^7\text{ seconds}$): $T_{\text{years}} = \frac{3.4028 \times 10^{24}}{3.1536 \times 10^7} \approx \mathbf{1.079 \times 10^{17}\text{ years}}$
   • (The universe is approximately $1.38 \times 10^{10}$ years old; brute-forcing AES-128 requires nearly 10 million times the age of the universe!)
3. Massive Distributed Attack (1 Billion Supercomputers):
   • Aggregated test rate: $10^9 \times 10^{14} = 10^{23}\text{ keys/second}$.
   • Time required: $T_{\text{years}} = \frac{1.079 \times 10^{17}\text{ years}}{10^9} = \mathbf{1.079 \times 10^{8}\text{ years}} \ (107.9\text{ Million years!})$
   • Even with 1 billion supercomputers, breaking AES-128 via brute force remains thermodynamically impossible.

---

Example 2: Availability Calculation & Service Level Agreement (SLA) Downtime

Problem Statement: An enterprise cloud banking service guarantees "Four Nines" of availability ($99.99\%$) over a calendar year ($365\text{ days}$).

1. Calculate the maximum permissible downtime (in hours, minutes, and seconds) per year under a $99.99\%$ SLA.
2. The banking architecture consists of three sequential components: an edge API Gateway ($99.95\%$ availability), an application server ($99.9\%$ availability), and a backend database cluster ($99.99\%$ availability). If any component fails, the service is unavailable. Calculate the overall end-to-end system availability and determine whether the architecture meets the $99.99\%$ SLA.

Step-by-Step Solution:

1. Permissible Downtime for $99.99\%$ Availability:
   • Total seconds in a 365-day year: $T_{\text{year}} = 365 \times 24 \times 3600 = 31,536,000\text{ seconds}$
   • Permissible downtime percentage: $100\% - 99.99\% = 0.01\% = 0.0001$.
   • Permissible downtime seconds: $\text{Downtime} = 31,536,000 \times 0.0001 = 3,153.6\text{ seconds}$
   • Converting to minutes and seconds: $\frac{3,153.6}{60} = \mathbf{52\text{ minutes and } 33.6\text{ seconds per year}}$
2. Series System Availability Calculation:
   • For sequential dependencies where any failure causes system failure: $A_{\text{total}} = A_{\text{gateway}} \times A_{\text{app}} \times A_{\text{database}}$ $A_{\text{total}} = 0.9995 \times 0.9990 \times 0.9999 = 0.9984004 \approx \mathbf{99.84\%}$
   • Expected annual downtime: $\text{Actual Downtime} = 31,536,000 \times (1 - 0.9984) \approx 50,457.6\text{ seconds} \approx \mathbf{14\text{ hours and } 1\text{ minute}}$
   • Conclusion: The system achieves only $99.84\%$ ("Two Nines"), severely violating the $99.99\%$ SLA contract. To achieve Four Nines, components must be engineered with redundant parallel failover ($A = 1 - (1 - A_1)(1 - A_2)$).

---

Example 3: IPsec ESP Tunnel Mode Encapsulation Overhead

Problem Statement: A company transmits sensitive patient medical records over a public internet VPN using IPsec in ESP (Encapsulating Security Payload) Tunnel Mode with AES-256-CBC and HMAC-SHA256.

• Original payload IP packet size: $L = 1000\text{ bytes}$ (including original 20-byte IPv4 header).
• New outer IPv4 header: $20\text{ bytes}$.
• ESP Header (SPI + Sequence Number): $8\text{ bytes}$.
• ESP IV (Initialization Vector for AES-CBC): $16\text{ bytes}$.
• AES-CBC block size: $16\text{ bytes}$.
• ESP Trailer: Padding (to align to 16 bytes) $+ \text{Pad Length (1 byte)} + \text{Next Header (1 byte)}$.
• ESP ICV (Integrity Check Value - HMAC-SHA256 truncated): $16\text{ bytes}$. Calculate the total encapsulated packet size and determine the bandwidth overhead percentage.

Step-by-Step Solution:

1. Determine Padding Needed for Block Alignment:
   • Data to be encrypted: Original packet ($1000\text{ bytes}$) $+$ Pad Length ($1\text{ byte}$) $+$ Next Header ($1\text{ byte}$) $= 1002\text{ bytes}$.
   • AES-CBC requires data to be a multiple of the 16-byte block size: $\frac{1002}{16} = 62.625 \implies \text{Next multiple is } 63 \times 16 = 1008\text{ bytes}$
   • Padding required: $\text{Padding} = 1008 - 1002 = \mathbf{6\text{ bytes}}$
   • Total encrypted payload (Data + Padding + PadLen + NextHeader) $= 1008\text{ bytes}$.
2. Calculate Total Encapsulated Packet Size: $\text{Outer IP Header} = 20\text{ bytes}$ $\text{ESP Header (SPI + Seq)} = 8\text{ bytes}$ $\text{ESP IV} = 16\text{ bytes}$ $\text{Encrypted Payload + Trailer} = 1008\text{ bytes}$ $\text{ESP ICV (Auth Tag)} = 16\text{ bytes}$ $\text{Total Packet Size} = 20 + 8 + 16 + 1008 + 16 = \mathbf{1,068\text{ bytes}}$
3. Bandwidth Overhead Percentage: $\text{Overhead \%} = \frac{1068 - 1000}{1000} \times 100\% = \frac{68}{1000} \times 100\% = \mathbf{6.8\%}$ IPsec tunnel encapsulation introduces an exact $6.8\%$ bandwidth overhead for confidentiality and integrity protection.

---

🎯 Exam & Interview Pitfall Check

Question 1: Define the CIA triad. Contrast passive network attacks with active network attacks, providing two concrete examples of each.

Answer:

• The CIA Triad:
  • Confidentiality: Ensuring that sensitive data is accessible and readable solely by authorized entities, preventing unauthorized disclosure.
  • Integrity: Ensuring that data remains accurate and complete, guaranteeing that unauthorized alterations, insertions, deletions, or reorderings are detected.
  • Availability: Ensuring that network services and data remain accessible and operational for authorized users whenever requested.
• Passive vs Active Attacks:
  • Passive Attacks: Attacks where the adversary merely monitors or eavesdrops on communication channels without modifying or disrupting the data stream.
    • Example 1: Packet Sniffing: Using promiscuous network interfaces to capture unencrypted HTTP passwords or Telnet keystrokes.
    • Example 2: Traffic Analysis: Observing the frequency, packet sizes, and transmission timing between endpoints to deduce military operations or business mergers without decrypting the payload.
  • Active Attacks: Attacks where the adversary actively alters data, injects fabricated messages, or impairs system availability.
    • Example 1: Man-in-the-Middle (MitM) Tampering: Intercepting an unauthenticated IP packet and modifying the destination bank account number.
    • Example 2: Denial of Service (DoS): Flooding a web server with TCP SYN packets to exhaust memory connection tables and crash legitimate access.

---

Question 2: State Kerckhoffs's Principle and explain why "Security through Obscurity" fails in modern computer network engineering.

Answer:

• Kerckhoffs's Principle: A cryptosystem must be designed such that it remains completely secure even if the adversary knows all mathematical details, algorithmic steps, source code, and hardware specifications of the system, provided only that the specific secret Key remains hidden.
• Why Security Through Obscurity Fails:
  1. Inevitability of Discovery: Proprietary algorithms can be reverse-engineered from compiled binaries, extracted via hardware memory inspection, or leaked by disgruntled employees. Once a secret algorithm is leaked, every deployed system running that algorithm is permanently compromised and cannot be fixed without rewriting all software.
  2. Lack of Peer Review: Open algorithms (like AES and RSA) are subjected to public scrutiny by global cryptanalysts for years before standardization. Flaws are discovered and eliminated early. Proprietary secret algorithms rarely receive rigorous mathematical vetting, almost invariably containing catastrophic hidden mathematical flaws.
  3. Key Revocability: Under Kerckhoffs's model, if a secret key is compromised, the operator simply generates a new key in microseconds. The underlying cryptographic algorithm remains secure and reusable.

---

Core Conceptual Questions

• Why doesn't encryption alone guarantee message integrity?: Encryption guarantees confidentiality (the attacker cannot read the plaintext). However, in many symmetric block ciphers (such as CBC mode or stream ciphers), an active attacker can perform bit-flipping attacks: modifying specific ciphertext bits predictably alters corresponding plaintext bits upon decryption without knowing the key! True security requires combining encryption with an integrity tag (AEAD, HMAC, or Digital Signature).
• What is Non-Repudiation?: Non-repudiation guarantees that the sender of a message cannot falsely claim they did not author or transmit the message. Symmetric encryption cannot provide non-repudiation because both sender and receiver share the exact same key. Non-repudiation strictly requires Asymmetric Digital Signatures, where only the sender possesses the unique private key.

Common Interview Traps

• Trap: Confusing Authentication with Authorization: Authentication verifies who you are (verifying identity via password, certificate, or biometric). Authorization determines what you are allowed to do (verifying access permissions to read or write a resource).
• Trap: Believing Redundancy Guarantees Availability Against Application Attacks: While redundant servers and BGP Anycast protect against physical link outages and volumetric network floods, they do not protect against algorithmic complexity attacks (e.g. ReDoS or slow-loris attacks that consume all CPU cores regardless of server count).

---

🗺️View Complete Computer Networks Roadmap & Syllabus

← Previous Topic8.5 File Transfer Protocol (FTP) & Dynamic Host Configuration (DHCP)Next Topic →9.2 Symmetric Key Cryptography: DES & Advanced Encryption Standard (AES)