PadhaikoAll subjects

Advanced Networking Applications

Next Generation Networking with IPv6 (ENCT 387)

Open in the Padhaiko app

7.1 Fundamentals of SDN

Traditional IP networks bundle the control plane (routing protocols deciding paths) and the data plane (forwarding tables moving packets) inside every individual router or switch. Configuring policies network-wide requires touching dozens or thousands of devices manually via CLI. Software-Defined Networking (SDN) decouples these planes to make networks programmable as a single logical fabric.

Definition

The Three Planes of SDN Architecture

  1. Data (Infrastructure) Plane: Physical and virtual switches/routers that perform high-speed Match-Action packet forwarding using hardware flow tables (no local routing intelligence required).
  2. Control Plane (SDN Controller): A logically centralized software controller (e.g., ONOS, OpenDaylight, Ryu) maintaining a global topology view and programming forwarding rules into data-plane devices via Southbound APIs (OpenFlow, NETCONF/YANG, P4Runtime, gNMI).
  3. Application / Management Plane: Network applications (traffic engineering, load balancing, firewalling, zero-trust automation) interacting with the controller via Northbound APIs (REST, gRPC, Python/Java SDKs).
Software-Defined Networking (SDN) Three-Layer ArchitectureText
+-------------------------------------------------------------------+
|  Application Plane: Traffic Engineering | Firewall | Load Balancer|
+-------------------------------------------------------------------+
                         ^   Northbound API (REST / gRPC)
                         v
+-------------------------------------------------------------------+
|  Control Plane:     Logically Centralized SDN Controller          |
|                     (Global Topology, Path Computation, Policy)   |
+-------------------------------------------------------------------+
                         ^   Southbound API (OpenFlow / NETCONF / P4)
                         v
+-------------------------------------------------------------------+
|  Data Plane:        Programmable Switches & Routers               |
|                     (Match-Action Flow Tables -> Wire-Speed Fwd)  |
+-------------------------------------------------------------------+
AspectTraditional Distributed NetworkingSoftware-Defined Networking (SDN)
Control & Data PlanesTightly coupled inside every router/switchPhysically decoupled; centralized logical controller
Configuration ModelBox-by-box manual CLI or vendor-specific scriptsCentralized programmatic APIs & automation
Topology VisibilityLocal view (each router builds its own view via IGPs)Global network-wide graph visible to controller apps
Traffic EngineeringDestination-based shortest path (cost/hops)Multi-field Match-Action rules (Src/Dst, Flow Label, QoS)
Hardware FlexibilityFixed vendor ASIC protocol logicProgrammable pipelines (e.g., OpenFlow, P4)

7.2 Introduction to IBN and NDN

Beyond programmable forwarding in SDN, next-generation architectures explore declarative automation (IBN) and content-centric routing (NDN).

Intent-Based Networking (IBN)

Intent-Based Networking (IBN) elevates network management from imperative commands ("configure VLAN 20, ACL 105, and OSPF cost 10 on switch X") to high-level declarative business intent ("ensure telemedicine video traffic between Clinic A and Hospital B stays encrypted with latency under 20 ms").

Definition

The Four-Stage Closed-Loop IBN Lifecycle

  1. Translation & Validation: Parses human/API business intent, verifies feasibility against resources, and compiles it into concrete network policies (routing, QoS, IPsec, ACLs).
  2. Activation (Automated Deployment): Pushes validated configurations across switches, routers, and SDN controllers via NETCONF/YANG, gNMI, or OpenFlow without manual CLI edits.
  3. Assurance (Continuous Telemetry): Continuously monitors streaming telemetry, latency, packet loss, and flow compliance to verify that the live network satisfies the declared intent.
  4. Remediation & Optimization: When link failures or congestion threaten the intent, the system automatically reroutes flows or adjusts queues in a closed feedback loop.

Named Data Networking (NDN)

Named Data Networking (NDN) is a clean-slate Information-Centric Networking (ICN) architecture that replaces host-centric IP addresses (where data lives) with hierarchical content names (what data is requested, e.g., /edu/ioe/enct387/notes/ch7/v1).

AspectIPv6 Host-Centric NetworkingNamed Data Networking (NDN)
Core Identifier128-bit interface/host address (2001:db8::1)Variable-length hierarchical content name (/org/app/file)
Forwarding StateStateless FIB on destination IPv6 prefixStateful PIT + FIB + Content Store cache
In-Network CachingHandled at app layer (CDNs, HTTP proxies)Native at network layer in every router's Content Store
Security ModelChannel-based (TLS / IPsec tunnel between hosts)Object-based (every Data packet is signed by producer)

7.3 IPv6 as the foundations of 5G network

5G mobile networks serve three diverse service classes—eMBB (high-throughput broadband), URLLC (mission-critical low-latency control), and mMTC (massive IoT)—requiring flexible, cloud-native packet cores where IPv6 is foundational.

5G Core Architecture and IPv6 PDU Sessions

Segment Routing over IPv6 (SRv6 — RFC 8754 / RFC 8986)

SRv6 turns the 128-bit IPv6 address into a programmable Segment Identifier (SID) split into Locator : Function : Arguments:

The ingress router inserts a Segment Routing Header (SRH, Routing Header Type 4) listing the ordered SID path; transit routers forward using standard IPv6 Longest Prefix Match without keeping per-slice MPLS/RSVP state.


7.4 IPv6 integration SDN/IBN/NDN

Modern carrier and cloud backbones integrate IPv6 directly with SDN, IBN, and NDN to combine global reachability with programmability:

  1. IPv6 + SDN (Programmable Match-Action & SRv6):
    • OpenFlow 1.3+ and P4 switches natively match 128-bit IPv6 headers, Traffic Class (DSCP/ECN), and the 20-bit Flow Label to steer encrypted flows without inspecting TCP/UDP ports.
    • Pairing an SDN Controller with SRv6 solves SDN's core TCAM scalability bottleneck: instead of pushing millions of per-flow state rules into every core switch, the SDN controller programs an SRv6 SID list only at the network ingress, while the core forwards using compact IPv6 Locator routes.
  2. IPv6 + IBN (Declarative Slicing & Policy Automation):
    • Nibble-aligned IPv6 addressing (/48, /56, /64) and unique non-NATed endpoint identities allow an IBN engine to translate high-level intent (such as "isolate 5G URLLC robotic slice with <10 ms delay") directly into deterministic IPv6 prefix ACLs, DSCP markings, and SRv6 transport paths pushed via NETCONF/YANG, while telemetry on the IPv6 Flow Label verifies SLA assurance continuously.
  3. IPv6 + NDN (Overlay Transport & Coexistence):
    • Because the global Internet runs on IPv6/IPv4 rather than pure Layer 2 NDN, NDN deployments commonly encapsulate Interest and Data packets inside IPv6/UDP overlay tunnels across wide-area backbones.
    • IPv6 multicast (ff02::/16) supports local NDN Interest discovery on Ethernet segments, while SDN/SRv6 underlays provide guaranteed bandwidth slices linking distributed NDN Content Stores.

Worked Examples

Example 1

1

A 5G mobile operator allocates the IPv6 block 2001:db8:5a00::/40 to a regional User Plane Function (UPF) to assign one dedicated /64 prefix per UE PDU session via SLAAC. Furthermore, the operator's 5G transport network uses SRv6 with a 128-bit SID format of 48-bit Locator : 16-bit Function : 64-bit Arguments.

  1. How many simultaneous /64 UE PDU sessions can this /40 pool support?
  2. Verify the bit allocation of the SRv6 SID and calculate the total outer IPv6 + SRH header size (in bytes) when a packet carries a Segment Routing Header (SRH) containing 3 SIDs (assuming an 8-byte fixed SRH base header plus 16 bytes per 128-bit SID).

1. Calculating Simultaneous 5G /64 PDU Sessions: Prefix bits available between /40 and /64 = 64 - 40 = 24 bits. Number of /64 UE prefixes = 2^24 = 16,777,216 simultaneous sessions.

2. SRv6 SID Structure and Header Overhead:

  • Total SID width = 48 (Locator) + 16 (Function) + 64 (Arguments) = 128 bits.
  • An SRH for 3 128-bit (16-byte) SIDs has length: SRH size = 8 (Fixed SRH header) + (3 x 16) = 8 + 48 = 56 bytes.
  • Combined with the 40-byte IPv6 base header: Total outer IPv6 + SRH overhead = 40 + 56 = 96 bytes.
Answer:
  1. The /40 pool supports 16,777,216 simultaneous /64 UE PDU sessions.
  2. Total SID length = 128 bits; total IPv6 + 3-SID SRH overhead = 96 bytes (40 B base + 56 B SRH).
Example 2

2

Consider a carrier core containing 8 transit switches serving 50 edge PoPs and 20,000 simultaneous active customer flows.

  1. In a stateful per-flow OpenFlow SDN design where every transit switch installs an individual rule per active flow, how many total flow entries are consumed across the 8 transit switches?
  2. If the SDN controller switches to SRv6 stateless steering—pushing the SID path list only at the ingress edge so each of the 8 transit switches only needs 1 aggregate /48 Locator route per edge PoP—how many routing entries are needed across the 8 transit switches, and by what factor is core state reduced?

1. Per-Flow OpenFlow Core State: Each of the 8 transit switches stores 20,000 flow entries: Total OpenFlow core entries = 8 x 20,000 = 160,000 entries.

2. SRv6 Aggregate Locator Core State and Reduction Factor: Each of the 8 transit switches only stores 1 Locator prefix per edge PoP (50 entries): Total SRv6 core entries = 8 x 50 = 400 entries. State reduction factor = 160,000 / 400 = 400x reduction.

Answer:
  1. Per-flow OpenFlow consumes 160,000 core entries.
  2. SRv6 Locator forwarding requires only 400 core entries—a 400x reduction in transit state.
Example 3

3

(Exam-style Architecture Walkthrough) Two students, Host A and Host B, sit behind NDN router R1, which connects to R2 and then to a university content producer hosting /ioe/enct387/ch7/slides.pdf.

  1. Trace step-by-step how R1 and R2 forward the Interest and Data packets when Host A requests the file first, and then Host B requests the exact same file seconds later.
  2. Explain how an Intent-Based Networking (IBN) system over an IPv6/SDN campus backbone would enforce and assure low-latency delivery for this lecture stream.

1. NDN Interest/Data Forwarding and Caching Walkthrough:

  • Host A's Request: Host A sends an Interest(/ioe/enct387/ch7/slides.pdf) to R1. R1 checks its Content Store (CS) (miss), inserts an entry into its PIT recording Host A's incoming face, and checks its FIB for /ioe/enct387 to forward the Interest to R2. R2 likewise logs R1 in its PIT and forwards to the producer.
  • Data Return: The producer signs and returns the Data packet. R2 matches its PIT, caches a copy in its CS, forwards Data to R1, and removes the PIT entry. R1 caches a copy in its CS and delivers Data to Host A.
  • Host B's Request: Seconds later, Host B sends the same Interest to R1. R1 hits its Content Store (CS) immediately and returns the cached, producer-signed Data packet directly to Host B without forwarding upstream to R2 or the producer.

2. IBN Closed-Loop Enforcement Over IPv6/SDN:

  1. Translation: Administrator inputs intent "Guarantee <= 15 ms latency for /ioe/enct387 streaming to Campus B." IBN maps this to IPv6 prefix policies, DSCP AF41, Flow Label matching, and an SRv6 low-latency SID path.
  2. Activation: Controller programs ingress SDN switches and SRv6 policies automatically via NETCONF/P4.
  3. Assurance & Remediation: Streaming telemetry tracks real-time delay across the IPv6 path; if link congestion raises latency above 15 ms, the IBN controller automatically updates the ingress SRv6 SID list to steer flows onto an uncongested fiber path.
Answer:
  1. Host A's Interest creates PIT breadcrumbs via FIB lookup across R1 -> R2 -> Producer; the returning signed Data follows the PIT back and populates each router's Content Store (CS) so Host B's subsequent Interest is served directly from R1's local CS cache.
  2. IBN translates the SLA intent into IPv6 DSCP/Flow-Label/SRv6 rules, activates them via SDN southbound APIs, and continuously assures and reroutes traffic via closed-loop telemetry.

Exercises

Practice Exercises — Advanced Networking Applications

Question 1

Explain the architecture of Software-Defined Networking (SDN) by distinguishing the Data Plane, Control Plane, and Application Plane, as well as Southbound and Northbound APIs. List three operational advantages of SDN over traditional distributed routing.

SDN Planes and APIs:

  • Data Plane: Programmable hardware/virtual switches executing fast Match-Action forwarding.
  • Control Plane: Logically centralized SDN controller maintaining a global topology graph and installing forwarding state via Southbound APIs (OpenFlow, P4Runtime, NETCONF/YANG).
  • Application Plane: Traffic-engineering, security, and orchestration apps communicating with the controller via Northbound APIs (REST, gRPC).

Three Operational Advantages:

  1. Centralized Programmability: Eliminates error-prone manual CLI edits across individual devices.
  2. Global Traffic Optimization: Computes globally optimal paths instead of purely local shortest-path decisions.
  3. Vendor Abstraction & Rapid Innovation: Software apps can introduce new forwarding/QoS behaviors without replacing switch hardware.
Answer:

SDN separates the Data Plane (switches), Control Plane (centralized controller via Southbound OpenFlow/NETCONF/P4), and Application Plane (via Northbound REST/gRPC). Its key advantages are centralized programmability, global path optimization, and hardware-independent innovation.

Question 2

Define Intent-Based Networking (IBN) and explain each phase of its four-stage closed-loop lifecycle. How does IBN differ from standard imperative SDN scripting?

IBN vs. Imperative Scripting: Imperative scripts specify how to configure each device step-by-step, whereas IBN accepts declarative high-level business intent (what outcome is desired, such as SLA latency or isolation) and continuously verifies that the network keeps honoring it.

Four-Stage Closed-Loop Lifecycle:

  1. Translation: Converts high-level intent into validated network policies.
  2. Activation: Automatically pushes configurations across devices via SDN/NETCONF APIs.
  3. Assurance: Collects real-time telemetry to verify intent compliance continuously.
  4. Remediation: Automatically adjusts paths or policies whenever telemetry detects drift or SLA violation.
Answer:

IBN manages networks by desired business outcomes rather than low-level device commands, operating through a continuous closed loop of Translation -> Activation -> Assurance -> Remediation.

Question 3

Describe how Named Data Networking (NDN) replaces host-based addressing with content-based networking. Explain the exact roles of Interest packets, Data packets, the Content Store (CS), the Pending Interest Table (PIT), and the Forwarding Information Base (FIB).

NDN Packet Types and Router Tables:

  • Interest Packet: Sent by a consumer specifying the hierarchical name of desired content.
  • Data Packet: Returned by a producer or cache containing the Name, Content, and a cryptographic producer signature.
  • Content Store (CS): In-network router cache serving repeat requests locally.
  • Pending Interest Table (PIT): Records incoming faces for unanswered Interests, aggregates duplicate Interests, and guides returning Data back along the reverse path.
  • Forwarding Information Base (FIB): Routes Interests upstream based on longest prefix match on content names.
Answer:

NDN routes by hierarchical content names using Interest and signed Data packets: routers check the CS cache first, aggregate/track pending requests in the PIT, and forward cache-misses upstream via the FIB, returning Data along PIT entries while caching in CS.

Question 4

Why is IPv6 essential to 5G networks (eMBB, URLLC, mMTC, and SBA)?

  1. If a 5G UPF holds a /44 IPv6 pool to assign /64 prefixes via SLAAC to mobile devices, how many simultaneous UE PDU sessions can it support?
  2. Compute the combined IPv6 base header + SRv6 Segment Routing Header (SRH) size in bytes for a 5G transport packet carrying 4 SIDs (8-byte fixed SRH base + 16 bytes per SID).

Why IPv6 Underpins 5G: 5G requires connecting billions of mMTC IoT endpoints without NAT bottlenecks, assigning dedicated /64 prefixes per UE PDU session via SLAAC, running HTTP/2 Service-Based Architecture (SBA) control planes over IPv6, and steering network slices deterministically via SRv6.

Numerical Calculations:

  1. UE PDU Sessions: 64 - 44 = 20 bits -> 2^20 = 1,048,576 simultaneous /64 sessions.
  2. IPv6 + 4-SID SRH Overhead: 40 (base IPv6) + 8 (SRH fixed) + 4 x 16 (4 SIDs) = 40 + 8 + 64 = 112 bytes.
Answer:

IPv6 enables massive mMTC addressing, native /64 per-UE SLAAC PDU sessions, SBA transport, and SRv6 slicing. A /44 pool supports 1,048,576 /64 UE sessions, and a 4-SID SRv6 header adds up to 112 bytes total (40 B base + 72 B SRH).

Question 5

Explain how IPv6 integrates with SDN, IBN, and NDN in modern network architectures—highlighting the specific roles of SRv6, the 20-bit IPv6 Flow Label, nibble-aligned prefixing, and IPv6 overlay tunneling.

Integration Across SDN, IBN, and NDN:

  • IPv6 + SDN: OpenFlow/P4 switches match 128-bit IPv6 prefixes and the 20-bit Flow Label (preserving flow visibility even under encryption), while SRv6 lets an SDN controller encode forwarding paths into edge SRHs so core routers stay stateless.
  • IPv6 + IBN: Clean nibble-aligned IPv6 hierarchies and non-NATed endpoint addresses let an IBN controller map high-level SLA/security intents directly into NETCONF/YANG ACLs, DSCP marks, and SRv6 slices with continuous telemetry verification.
  • IPv6 + NDN: NDN Interest/Data streams traverse wide-area networks inside IPv6/UDP overlay tunnels (with IPv6 multicast for local discovery and SRv6 slices interconnecting distributed Content Stores).
Answer:

IPv6 integrates with SDN via Flow-Label matching and stateless SRv6 steering, empowers IBN through nibble-aligned deterministic policy translation and non-NATed assurance telemetry, and transports NDN across WANs via IPv6/UDP overlays and SRv6 transport slices.

← IPv6 Deployment and Network Design