Advanced Topics
Open in the Padhaiko app7.1 Introduction to software-defined networking (SDN), basic architecture, importance and applications
In traditional networks, every router and switch tightly bundles packet forwarding (data plane) and routing intelligence (control plane) inside proprietary vendor firmware. Software-Defined Networking (SDN) decouples these planes, centralizing control logic in programmable software controllers while leaving switches as fast, simple forwarding hardware.
Software-Defined Networking (SDN)
Software-Defined Networking (SDN) is a network architecture that separates the control plane (deciding where packets go) from the data plane (forwarding packets), exposing a centralized, software-programmable network view to applications via open APIs.
SDN Architecture (Three Planes and Two Interfaces)
- Data (Infrastructure) Plane: Physical and virtual switches (e.g., Open vSwitch) that maintain match-action flow tables (matching L2/L3/L4 header fields and executing actions such as forward, drop, modify, or send to controller).
- Control Plane (SDN Controller): Logically centralized software platform (e.g., ONOS, OpenDaylight, Ryu) that maintains a global topology graph and computes forwarding rules for all switches.
- Application Plane: Network apps (traffic engineering, load balancers, firewalls, IDS) that read network state and push policies dynamically without touching individual switch CLIs.
- Southbound Interface (SBI): Connects the Control Plane to the Data Plane (most famously OpenFlow, as well as NETCONF/gNMI).
- Northbound Interface (NBI): Connects the Application Plane to the Control Plane (typically RESTful APIs).
| Aspect | Traditional Distributed Networking | Software-Defined Networking (SDN) |
|---|---|---|
| Control & Data Planes | Tightly coupled inside each closed vendor switch/router | Physically decoupled; centralized software controller + simple switches |
| Configuration & Policy | Box-by-box manual CLI/SNMP configuration | Global programmatic automation via REST APIs |
| Network Visibility | Distributed local view per router | Unified global topology view at the controller |
| Innovation Speed | Slow; tied to vendor hardware/OS release cycles | Rapid; new routing or security behaviors deployed as software modules |
Importance and Applications of SDN
- Traffic Engineering & Dynamic QoS: Global visibility lets the controller route elephant flows around congested links immediately.
- Data Center & Cloud Multi-Tenancy: Rapidly provisions isolated virtual networks across thousands of servers.
- SD-WAN (Software-Defined WAN): Steers enterprise branch traffic dynamically across MPLS, broadband, and 5G links based on real-time application latency.
- Dynamic Security Enforcement: Automatically quarantines compromised hosts by pushing drop flow rules to edge switches in milliseconds.
7.2 Network security and its importance
Modern networks carry financial transactions, healthcare records, industrial telemetry, and cloud services, making network security essential to preserve service availability, data privacy, and organizational trust.
The CIA Security Triad
CIA Triad of Network Security
- Confidentiality: Ensures sensitive data is readable only by authorized parties (enforced via encryption such as AES, TLS, and IPsec).
- Integrity: Guarantees that messages have not been altered or tampered with in transit (enforced via cryptographic hashes SHA-256, HMAC, and AEAD tags).
- Availability: Ensures networks and services remain accessible and responsive to legitimate users even under fault or attack (enforced via redundancy, rate limiting, and DDoS mitigation).
(Modern security also emphasizes Authentication—verifying identity via digital certificates/MFA—and Non-repudiation—proving a sender cannot deny sending a signed message.)
Network Attacks: Passive vs. Active
- Passive Attacks: Monitor communications without altering packet contents or disrupting flows (hard to detect, prevented via encryption).
- Examples: Eavesdropping (packet sniffing) and Traffic Analysis (observing communication frequency and endpoints).
- Active Attacks: Modify, fabricate, replay, or block packets to disrupt operations or impersonate entities (hard to prevent completely, mitigated via integrity checks and authentication).
- Examples: Man-in-the-Middle (MitM), Spoofing/Masquerade, Replay Attack, Denial of Service (DoS / DDoS), and ARP/DNS Poisoning.
Core Defense Mechanisms
- Cryptography: Symmetric encryption (AES-GCM) for high-speed bulk confidentiality; asymmetric encryption (RSA, ECC) for key exchange and digital signatures.
- Firewalls & IDS/IPS: Stateful packet-filtering and Next-Generation Firewalls (NGFW) enforce zone policies; Intrusion Detection/Prevention Systems inspect payloads for attack signatures and anomalies.
- VPNs (IPsec & TLS VPNs): Create authenticated, encrypted tunnels over untrusted public networks.
- Zero Trust & Network Segmentation: Enforces least-privilege access, VLAN/micro-segmentation, and continuous verification regardless of whether a user is inside or outside the perimeter.
7.3 Overview of content delivery networks (CDNs), named data networking (NDN), intent based networking (IBN)
1. Content Delivery Networks (CDNs)
A CDN is a geographically distributed network of edge cache servers (Points of Presence, PoPs) placed close to end users (often inside ISP networks).
- Mechanism: When a user requests static or streaming content (images, video chunks, software updates), Anycast BGP or DNS-based redirection steers the request to the nearest edge server.
- Cache Hit: Edge serves the file locally with ultra-low RTT.
- Cache Miss: Edge fetches the object from the origin server, caches a copy for future local requests, and streams it to the user.
- Benefits: Slashes user latency, offloads origin servers, saves core ISP transit bandwidth, and absorbs volumetric DDoS attacks at the edge.
2. Named Data Networking (NDN)
NDN is a Future Internet architecture based on Information-Centric Networking (ICN) that shifts communication from where data is located (host IP addresses) to what data is requested (hierarchical content names, e.g., /edu/ioe/enct304/lecture1.pdf).
- Two Packet Types:
- Interest Packet: Sent by a consumer carrying the desired content name.
- Data Packet: Returned by a producer or any intermediate router cache, cryptographically signed by the publisher (securing the data itself rather than the channel).
- Three Router Data Structures:
- Content Store (CS): In-network packet cache fulfilling matching Interests immediately.
- Pending Interest Table (PIT): Records unsatisfied Interests and their incoming interfaces so returning Data follows the exact reverse breadcrumb path (and naturally aggregates duplicate concurrent requests).
- Forwarding Information Base (FIB): Routes Interest packets toward potential data sources by name prefix.
3. Intent-Based Networking (IBN)
IBN builds atop SDN and AI/automation to let operators declare high-level business intent (what the network should achieve, such as "Isolate guest Wi-Fi from lab servers and guarantee latency for video conferencing") rather than low-level device configurations.
- Closed-Loop Lifecycle:
- Translation & Validation: Converts natural/declarative intent into verified network policies.
- Automated Activation: Deploys configurations across switches, routers, and firewalls via SDN/NETCONF.
- Continuous Assurance: Uses streaming telemetry to verify whether live network behavior still satisfies the intent, automatically remediating drift or link degradation.
7.4 Quantum network, network virtualization and recent trends in networking
1. Quantum Networks
A quantum network transmits quantum bits (qubits)—such as polarized photons over optical fiber or free space—to connect quantum processors or enable provably secure communication.
- Core Physical Principles:
- Superposition & Measurement: Measuring an unknown quantum state irreversibly disturbs it.
- No-Cloning Theorem: An unknown quantum state cannot be copied perfectly.
- Quantum Entanglement: Correlated photon pairs allow states to link across distance (quantum teleportation of quantum information, aided by classical channels and quantum repeaters).
- Quantum Key Distribution (QKD — e.g., BB84 Protocol): Uses two non-orthogonal bases (rectilinear and diagonal ) to agree on a shared classical cryptographic key. Any eavesdropper intercepting and re-transmitting qubits introduces detectable statistical bit errors (Quantum Bit Error Rate, QBER), revealing eavesdropping by laws of physics rather than computational hardness assumptions.
2. Network Virtualization
Network Virtualization decouples logical network topologies and services from underlying physical hardware:
- Overlay Virtual Networks (VXLAN / GENEVE): Encapsulate Layer 2 Ethernet frames inside Layer 3 UDP/IP packets (
VXLANuses UDP port4789and a VNI supporting up to isolated tenant segments, overcoming the limit of IEEE 802.1Q VLANs). - Network Functions Virtualization (NFV): Replaces dedicated hardware appliances (firewalls, load balancers, NAT gateways, WAN optimizers) with Virtual Network Functions (VNFs) running as software VMs/containers on commodity x86 servers.
- 5G Network Slicing: Partitions one shared physical 5G infrastructure into logically isolated end-to-end slices tailored for distinct SLAs (eMBB high bandwidth, URLLC ultra-low latency, mMTC massive IoT).
3. Recent Trends in Networking
- Wi-Fi 6 / 6E & Wi-Fi 7 (IEEE 802.11ax / 802.11be): Uses OFDMA, MU-MIMO, spectrum, and Multi-Link Operation (MLO) for gigabit wireless density.
- HTTP/3 & QUIC: Moves transport from kernel TCP to encrypted UDP-based QUIC to eliminate head-of-line blocking and speed up connection setup (/).
- Zero Trust Network Access (ZTNA) & SASE: Converges cloud-delivered SD-WAN and security inspection into unified identity-aware access.
- AI-Ops & Programmable Data Planes (P4): Uses machine learning for anomaly prediction and P4-programmable switches to parse custom protocols at terabit line rates.
Worked Out Examples
7.1
An OpenFlow SDN switch handles . For each new flow, the very first packet misses the flow table and triggers a Packet-In message to the SDN controller, which replies with a Flow-Mod rule installation. Suppose each established flow subsequently transfers of each directly through the data plane.
- Calculate the control-plane bandwidth (in ) exchanged between the switch and the controller.
- Calculate the data-plane throughput (in ) forwarded by the switch.
- What percentage of total data packets require controller intervention?
Calculating Control-Plane Bandwidth: Each new flow exchanges over the southbound OpenFlow channel. At :
Calculating Data-Plane Throughput and Intervention Ratio: Each flow sends through the hardware flow table:
Since only packet out of every packets per flow triggers a table miss:
- Control-plane bandwidth: ().
- Data-plane throughput: .
- Only of packets trigger controller intervention; the remaining are forwarded at hardware line rate.
7.2
A streaming provider deploys a CDN edge server near a campus. When a requested video segment is cached at the edge (cache hit), the user access latency is . On a cache miss, the request must fetch the segment from the distant origin server with latency . If the CDN achieves a cache hit ratio of ():
- Calculate the average user access latency .
- Compute the latency speedup factor compared to fetching every segment directly from the origin server.
- By what percentage is origin server request traffic reduced?
Computing Average Access Latency: The expected latency is the weighted average of cache-hit and cache-miss latencies:
Computing Speedup Factor and Origin Offload:
- Latency speedup factor relative to direct origin access ():
- Because of requests are served entirely by the edge cache without contacting the origin, origin request traffic drops by .
- Average access latency: .
- Latency speedup factor: .
- Origin server request traffic reduction: .
7.3
In the BB84 Quantum Key Distribution (QKD) protocol, suppose an eavesdropper (Eve) intercepts every transmitted photon, measures it in a randomly chosen rectilinear () or diagonal () basis, and resends a photon in that state to Bob.
- For a single sifted key bit (where Alice and Bob used the same basis), determine the probability that Eve introduces a bit error.
- If Alice and Bob publicly compare sifted check bits to detect eavesdropping, compute the probability (as a percentage) that Eve goes undetected and the probability that Eve is detected.
Determining Single-Bit Error Probability :
- With probability , Eve guesses the same basis as Alice and Bob, introducing error.
- With probability , Eve chooses the wrong basis, randomizing the photon polarization relative to Bob's basis; Bob subsequently obtains the wrong bit with probability .
- Therefore, the error probability per sifted check bit is:
Calculating Detection Probability over Check Bits:
- Single-bit error probability introduced by Eve: .
- Over check bits, probability of going undetected is , and probability of detection is .
Practice Exercises
Practice Exercises — SDN, Network Security, NDN/IBN & Quantum Networks
Explain the architecture of Software-Defined Networking (SDN) by describing its three functional planes and the roles of Northbound and Southbound interfaces. State two key advantages of SDN over traditional networking.
SDN Planes and Interfaces:
- Data Plane: Hardware/virtual switches executing fast packet forwarding using match-action flow tables.
- Control Plane: Logically centralized SDN controller maintaining global topology and programming switch flow tables via Southbound APIs (e.g., OpenFlow).
- Application Plane: Traffic engineering, security, and load-balancing apps communicating with the controller via Northbound APIs (e.g., REST).
Key Advantages:
- Centralized Global Visibility and Programmability: Simplifies traffic engineering and dynamic rerouting without per-box CLI scripting.
- Vendor Neutrality and Rapid Deployment: New network policies roll out as software upgrades rather than waiting for proprietary switch firmware updates.
SDN separates forwarding (Data Plane), centralized intelligence (Control Plane), and policy apps (Application Plane), linked via Southbound (OpenFlow) and Northbound (REST) APIs. Its main advantages are centralized global programmability and agile, vendor-neutral policy automation.
Distinguish between passive and active network security attacks, give two examples of each, and explain how the CIA Triad guides defensive countermeasures.
Passive vs. Active Attacks:
- Passive Attacks: Observe or monitor network traffic without modifying data or disrupting service (e.g., eavesdropping/packet sniffing, traffic analysis). Defended primarily by Confidentiality mechanisms (TLS, IPsec, VPN encryption).
- Active Attacks: Alter data, inject bogus packets, or disrupt availability (e.g., Man-in-the-Middle, DDoS, replay, ARP spoofing). Defended by Integrity (HMAC/AEAD, digital signatures), Authentication (certificates), and Availability controls (firewalls, rate limiting, redundancy).
Passive attacks (sniffing, traffic analysis) spy on traffic without altering it and are countered by encryption (Confidentiality). Active attacks (MitM, DDoS, replay) modify or disrupt traffic and require cryptographic Integrity, Authentication, firewalls, and DDoS mitigation (Availability).
Compare Content Delivery Networks (CDNs), Named Data Networking (NDN), and Intent-Based Networking (IBN) in terms of their primary goal and key operating mechanism.
Comparing the Three Paradigms:
- CDN: Reduces latency and origin load by caching popular web/video content on geographically distributed edge servers using DNS/Anycast redirection.
- NDN: Replaces host-IP addressing with hierarchical content names; consumers send
Interestpackets and routers useCS(cache),PIT(pending requests), andFIB(name routing) to return signedDatapackets from any cache or producer. - IBN: Elevates network operations to declarative business intent, automatically translating goals into SDN configurations and continuously verifying compliance via closed-loop telemetry.
CDNs cache content at ISP-edge servers to lower latency; NDN routes packets by content name using router CS/PIT/FIB tables instead of IP addresses; IBN translates high-level business intents into automated configurations with continuous closed-loop verification.
A CDN edge cache serves hits in , while cache misses take from the origin server. If the cache hit ratio is (), calculate the average access latency and the percentage reduction in latency compared to having no CDN.
Computing Average Latency and Reduction:
Without a CDN, every request takes , so the percentage reduction is:
Average access latency is , representing a reduction compared to direct origin access.
An SDN switch processes . Each new flow sends a Packet-In to the controller and receives a Flow-Mod reply. Thereafter, each flow forwards data packets of each in the data plane. Calculate the control-plane bitrate and data-plane bitrate (both in ).
Computing Control-Plane and Data-Plane Bitrates:
- Control-plane traffic per flow :
- Data-plane traffic per flow :
Control-plane bitrate is and data-plane bitrate is .