PadhaikoAll subjects

Fundamentals of Internet and Networking Protocols

Next Generation Networking with IPv6 (ENCT 387)

Open in the Padhaiko app

1.1 Evolution of the Internet and networking paradigms

Modern computer networking evolved from isolated, proprietary mainframe terminals into a global, packet-switched Internet governed by open standards. Understanding how IPv6 came to be requires tracing how networking architectures and communication paradigms evolved over the last five decades.

Definition

Key Milestones in Internet Evolution

Shifting Networking Paradigms

As network scale and application demands changed, communication paradigms evolved across four dimensions:

Evolution of Networking Paradigms
Paradigm DimensionTraditional Era (1970s–1990s)Modern Next-Generation Era
Switching ModeCircuit-switched voice; early shared-bus LANsAll-IP packet switching (VoIP, video, data converged)
Application ArchitectureHost-to-host and centralized client-serverDistributed cloud microservices, CDN edge, peer-to-peer
Endpoint MobilityFixed wired workstations with static IP addressesRoaming smartphones, vehicles, and wireless IoT sensors
Control & ManagementManual box-by-box CLI configurationSoftware-Defined Networking (SDN), automation, zero-trust
Addressing AssumptionsScarce public IPv4 addresses hidden behind NATAbundant globally unique IPv6 addresses, end-to-end reachability

1.2 OSI versus TCP/IP model

Network protocols are organized into layered architectures so that hardware changes, routing logic, and application software can evolve independently. Two foundational models govern networking education and practice: the ISO OSI Reference Model (7 layers) and the TCP/IP Protocol Suite (4 or 5 layers).

Definition

Encapsulation and Protocol Data Units (PDUs)

As user data travels down the sender's protocol stack, each layer prepends its own control header (and sometimes a trailer at Layer 2)—a process called encapsulation. At the receiver, each layer strips and processes its corresponding header (decapsulation).

OSI 7-Layer Model Mapped to TCP/IP 4-Layer ArchitectureText
  OSI 7-Layer Model                   TCP/IP 4-Layer Model       Key Protocols / Examples
+-------------------------------+   +-------------------------+
| 7. Application                |   |                         |
| 6. Presentation               |   | 4. Application Layer    |  HTTP, DNS, SMTP, DHCPv6, SSH
| 5. Session                    |   |                         |
+-------------------------------+   +-------------------------+
| 4. Transport                  |   | 3. Transport Layer      |  TCP, UDP, SCTP, QUIC
+-------------------------------+   +-------------------------+
| 3. Network                    |   | 2. Internet Layer       |  IPv4, IPv6, ICMPv6, OSPFv3
+-------------------------------+   +-------------------------+
| 2. Data Link                  |   | 1. Network Access Layer |  Ethernet (802.3), PPP, 802.11
| 1. Physical                   |   |    (Link + Physical)    |  UTP, Optical Fiber, Radio
+-------------------------------+   +-------------------------+
AspectOSI Reference ModelTCP/IP Protocol Suite
Number of Layers7 layers (strict separation of Application, Presentation, Session)4 layers (RFC 1122) or 5 layers (splitting Network Access into Link and Physical)
Origin & ApproachDeveloped by ISO as a theoretical reference model before protocolsDeveloped by DARPA/IETF from working protocols first, then documented as a model
Transport GuaranteesConnection-oriented only at Network (in original spec); both at TransportConnectionless at Internet layer (IP); reliable (TCP) or unreliable (UDP) at Transport
Layer CouplingStrict vertical boundaries; layers do not bypass adjacent layersLooser boundaries; applications often handle session/presentation tasks directly
Industry AdoptionUsed as a conceptual teaching and troubleshooting vocabulary (e.g., "Layer 3 switch")Actual foundation of the global Internet (IPv4, IPv6, TCP, UDP, DNS, HTTP)

1.3 IPv4 addressing overview

Internet Protocol version 4 (IPv4), standardized in RFC 791 (1981), provides logical Layer 3 addressing and best-effort packet delivery across interconnected networks.

Definition

IPv4 Address Structure and Notation

An IPv4 address is a 32-bit binary identifier divided into two logical parts:

For human readability, the 32 bits are split into four 8-bit octets written in dotted-decimal notation (each octet from 0 to 255), such as 192.168.10.25. The total theoretical IPv4 address space is 2^32 = 4,294,967,296 (about 4.29 billion) addresses.

Classful Addressing vs. CIDR

Originally, IPv4 divided addresses into five fixed classes based on the leading bits of the first octet. Because rigid class boundaries wasted huge blocks of addresses, the Internet adopted Classless Inter-Domain Routing (CIDR) (RFC 4632) and Variable Length Subnet Masking (VLSM) in 1993.

IPv4 Classful Address Classes (Historical Reference)
ClassLeading BitsFirst Octet RangeDefault Mask / PrefixIntended Purpose
A01 – 126 (0, 127 reserved)255.0.0.0 (/8)Very large networks (16,777,214 hosts each)
B10128 – 191255.255.0.0 (/16)Medium-to-large networks (65,534 hosts each)
C110192 – 223255.255.255.0 (/24)Small networks (254 hosts each)
D1110224 – 239N/AMulticast group addressing
E1111240 – 255N/AExperimental / reserved for future use

Under CIDR, a network is written as address/prefix-length (for example, 172.16.4.0/22), where /n specifies how many leading bits belong to the network prefix and (32 - n) bits remain for hosts.

Special-Purpose and Private IPv4 Ranges

Key Reserved and Private IPv4 Ranges
Address RangeCIDR PrefixPurpose / RFC
10.0.0.0 – 10.255.255.25510.0.0.0/8Private internal networks (RFC 1918)
172.16.0.0 – 172.31.255.255172.16.0.0/12Private internal networks (RFC 1918)
192.168.0.0 – 192.168.255.255192.168.0.0/16Private internal networks (RFC 1918)
127.0.0.0 – 127.255.255.255127.0.0.0/8Host loopback (127.0.0.1 tests local TCP/IP stack)
169.254.0.0 – 169.254.255.255169.254.0.0/16Link-local / APIPA when DHCP fails (RFC 3927)
100.64.0.0 – 100.127.255.255100.64.0.0/10Shared address space for Carrier-Grade NAT (RFC 6598)

1.4 IPv4 addressing review and limitations

Although CIDR, VLSM, and Network Address Translation (NAT) extended IPv4's lifespan by decades, the protocol exhibits fundamental architectural and operational limitations:

  1. Address Space Exhaustion: With only 32 bits (4.29 billion raw addresses, and even fewer usable after subtracting Class D/E, loopback, private, and subnet overhead), IANA depleted its central pool of free IPv4 /8 blocks in February 2011, and Regional Internet Registries (RIRs like APNIC, ARIN, RIPE) exhausted general pools shortly after.
  2. Broken End-to-End Transparency Due to NAT / CGNAT: Because private IPv4 addresses (10.0.0.0/8, 192.168.0.0/16) are not globally routable, edge routers rewrite source/destination IP addresses and TCP/UDP port numbers using NAT / NAPT. While NAT hides internal devices, it creates major problems:
    • Breaks direct peer-to-peer applications (VoIP, multiplayer gaming, P2P file sharing) requiring complex traversal hacks (STUN, TURN, ICE).
    • Breaks protocols that embed IP addresses inside payloads or authenticate headers (like IPsec AH), forcing NAT-Traversal (NAT-T) encapsulation.
    • Complicates server hosting, logging/forensics (many subscribers share one public IP under Carrier-Grade NAT), and state failover (routers must maintain huge NAT translation tables).
  3. Inefficient Variable-Length Header Processing:
    • The IPv4 header varies from 20 to 60 bytes due to optional fields, forcing routers to inspect header length (IHL) dynamically.
    • Every router along the path decrements the TTL and must recompute the IPv4 Header Checksum at every hop, wasting CPU cycles (even though Layer 2 Ethernet CRC and Layer 4 TCP/UDP checksums already protect data integrity).
  4. In-Network Router Fragmentation: IPv4 allows intermediate routers to fragment oversized packets into smaller chunks, increasing router memory/CPU overhead and causing severe packet loss amplification (losing one fragment discards the whole datagram).
  5. Lack of Native Autoconfiguration, Security, and Flow Support: IPv4 relies on broadcast-heavy ARP (Address Resolution Protocol) and external DHCP servers for configuration. IPsec was added later as an optional add-on, and the 8-bit ToS/DSCP field offers no dedicated hardware-friendly Flow Label for real-time streams.
  6. Routing Table Bloat: Legacy non-aggregated allocations and multihoming force Default-Free Zone (DFZ) BGP routers to maintain hundreds of thousands of IPv4 prefixes.

1.5 Need and importance of next generating IP addressing

To overcome IPv4's structural bottlenecks rather than layering more workarounds on top of NAT, the IETF standardized IPv6 (initially specified in RFC 1883 / RFC 2460, now RFC 8200). Next-generation IP addressing is essential for modern networks for six core reasons:

Definition

Core Drivers for Next-Generation IP Addressing

  1. Virtually Inexhaustible 128-Bit Address Space: IPv6 increases address length from 32 bits to 128 bits, yielding 2^128 (approximately 3.4 x 10^38) unique addresses—enough to assign globally unique addresses to every smartphone, sensor, vehicle, and appliance without NAT.
  2. Restoration of End-to-End Connectivity: Every device can hold a globally reachable IPv6 address, restoring the original Internet end-to-end model and simplifying VoIP, peer-to-peer services, remote telemetry, and end-to-end IPsec encryption.
  3. Hierarchical Aggregation and Routing Scalability: Large global prefixes assigned cleanly by IANA and RIRs to ISPs allow aggressive route aggregation in the Internet core, keeping backbone routing tables compact.
  4. Simplified, Fixed-Length Packet Header: IPv6 uses a fixed 40-byte basic header with only 8 fields (down from 12+ in IPv4), no header checksum (eliminating per-hop recomputation), and no router fragmentation (sources perform Path MTU Discovery), enabling faster hardware forwarding.
  5. Plug-and-Play Autoconfiguration (SLAAC) & Neighbor Discovery: Hosts can automatically generate their own IPv6 addresses using Stateless Address Autoconfiguration (SLAAC) via ICMPv6 Router Advertisements, replacing broadcast ARP with targeted multicast Neighbor Discovery.
  6. Built-in Support for Security, QoS, and Next-Gen Infrastructures: IPv6 integrates a 20-bit Flow Label for real-time traffic engineering, clean IPsec compatibility without NAT interference, and scalable addressing required for 5G mobile networks, massive IoT deployments, Software-Defined Networking (SDN), and Cloud-Native architectures.

Worked Examples

Example 1

1

A host is configured with the IPv4 CIDR address 192.168.45.130/26. Determine:

  1. The subnet mask in dotted-decimal notation.
  2. The network address (Network ID) and broadcast address of this subnet.
  3. The range of usable host IP addresses and the total number of usable hosts.

Deriving the Subnet Mask from /26: A /26 prefix reserves 26 bits for the network and 32 - 26 = 6 bits for hosts. In binary, 26 ones followed by 6 zeros split into octets is: 11111111.11111111.11111111.11000000 Since the 4th octet has 2 network bits (128 + 64 = 192), or equivalently 256 - 2^6 = 256 - 64 = 192, the subnet mask is 255.255.255.192.

Finding the Block Size, Network ID, and Broadcast Address:

  • Block size (increment) in the 4th octet: 2^6 = 64.
  • The /26 subnets inside 192.168.45.0/24 start at multiples of 64: .0, .64, .128, .192.
  • Since 130 lies between 128 and 191 (130 // 64 = 2, and 2 * 64 = 128), the Network ID is 192.168.45.128.
  • The Broadcast address is the last address before the next subnet (.192), which is 128 + 64 - 1 = 191, giving 192.168.45.191.

Calculating Usable Host Range and Host Count:

  • First usable host: 192.168.45.128 + 1 = 192.168.45.129
  • Last usable host: 192.168.45.191 - 1 = 192.168.45.190
  • Total usable hosts: 2^6 - 2 = 64 - 2 = 62 hosts.
Answer:
  1. Subnet Mask: 255.255.255.192
  2. Network ID: 192.168.45.128; Broadcast Address: 192.168.45.191
  3. Usable Host Range: 192.168.45.129 to 192.168.45.190 (62 usable hosts)
Example 2

2

An organization is allocated the IPv4 CIDR block 172.16.0.0/22 and must assign subnets using VLSM for four requirements:

Design a contiguous VLSM allocation from 172.16.0.0/22 (largest requirement first), state the prefix and usable host range for each subnet, and compute how many addresses remain unallocated.

Total Available Addresses in 172.16.0.0/22: With prefix /22, there are 32 - 22 = 10 host bits, giving 2^10 = 1024 total addresses spanning 172.16.0.0 through 172.16.3.255.

Determining Required Host Bits (2^h - 2 >= required) and Prefix Length (32 - h) for Each Subnet:

  • Department A (400 hosts): 2^9 - 2 = 510 >= 400 -> needs h = 9 bits -> /23 (block size = 512 addresses = two /24 blocks).
  • Department B (200 hosts): 2^8 - 2 = 254 >= 200 -> needs h = 8 bits -> /24 (block size = 256 addresses).
  • Department C (100 hosts): 2^7 - 2 = 126 >= 100 -> needs h = 7 bits -> /25 (block size = 128 addresses).
  • WAN Link (2 hosts): 2^2 - 2 = 2 >= 2 -> needs h = 2 bits -> /30 (block size = 4 addresses).

Assigning Contiguous Blocks Without Overlap:

  1. Department A (/23): 172.16.0.0/23 (covers 172.16.0.0 to 172.16.1.255; usable 172.16.0.1 – 172.16.1.254).
  2. Department B (/24): Next free address is 172.16.2.0 -> 172.16.2.0/24 (covers 172.16.2.0 to 172.16.2.255; usable 172.16.2.1 – 172.16.2.254).
  3. Department C (/25): Next free address is 172.16.3.0 -> 172.16.3.0/25 (covers 172.16.3.0 to 172.16.3.127; usable 172.16.3.1 – 172.16.3.126).
  4. WAN Link (/30): Next free address is 172.16.3.128 -> 172.16.3.128/30 (covers 172.16.3.128 to 172.16.3.131; usable 172.16.3.129 – 172.16.3.130).

Computing Remaining Unallocated Addresses: Total allocated addresses = 512 + 256 + 128 + 4 = 900 addresses. Remaining addresses (172.16.3.132 – 172.16.3.255) = 1024 - 900 = 124 addresses.

Answer:
  • Department A: 172.16.0.0/23 (usable 172.16.0.1 – 172.16.1.254)
  • Department B: 172.16.2.0/24 (usable 172.16.2.1 – 172.16.2.254)
  • Department C: 172.16.3.0/25 (usable 172.16.3.1 – 172.16.3.126)
  • WAN Link: 172.16.3.128/30 (usable 172.16.3.129 – 172.16.3.130)
  • Remaining unallocated addresses: 124 addresses (172.16.3.132 – 172.16.3.255)
Example 3

3

Compare IPv4 (20-byte base header) and IPv6 (40-byte base header) in terms of:

  1. Exact total address space size in IPv4 vs. IPv6 and the expansion ratio.
  2. IP-layer header overhead percentage for a small VoIP packet carrying a 60-byte voice payload over an 8-byte UDP header.
  3. IP-layer header overhead percentage for a bulk data packet carrying a 1400-byte payload over a 20-byte TCP header.

Comparing Address Spaces:

  • IPv4 (32-bit): 2^32 = 4,294,967,296 addresses (~4.29 billion).
  • IPv6 (128-bit): 2^128 ≈ 3.4028 x 10^38 addresses.
  • Expansion Ratio: 2^128 / 2^32 = 2^96 ≈ 7.92 x 10^28 times larger than IPv4.

IP Header Overhead for Small VoIP Packet (8-byte UDP + 60-byte payload):

  • Over IPv4: Total packet size = 20 + 8 + 60 = 88 bytes. IPv4 header overhead = (20 / 88) x 100% = 22.73%.
  • Over IPv6: Total packet size = 40 + 8 + 60 = 108 bytes. IPv6 header overhead = (40 / 108) x 100% = 37.04%.

IP Header Overhead for Bulk TCP Packet (20-byte TCP + 1400-byte payload):

  • Over IPv4: Total packet size = 20 + 20 + 1400 = 1440 bytes. IPv4 header overhead = (20 / 1440) x 100% = 1.39%.
  • Over IPv6: Total packet size = 40 + 20 + 1400 = 1460 bytes. IPv6 header overhead = (40 / 1460) x 100% = 2.74%.
Answer:
  1. IPv4 has 4,294,967,296 addresses; IPv6 has 2^128 (~3.4 x 10^38) addresses (2^96 times larger).
  2. For the 60-byte VoIP/UDP packet, IP header overhead is 22.73% (IPv4) vs. 37.04% (IPv6).
  3. For the 1400-byte bulk TCP packet, IP header overhead drops to 1.39% (IPv4) vs. 2.74% (IPv6), showing minimal overhead penalty for normal data traffic despite 4x longer addresses.

Exercises

Practice Exercises — Fundamentals of Internet and Networking Protocols

Question 1

Compare the OSI 7-layer Reference Model and the TCP/IP 4-layer Protocol Suite. List the Protocol Data Unit (PDU) name at each layer of the TCP/IP stack and explain how encapsulation works as data moves from sender to receiver.

Architectural Differences Between OSI and TCP/IP:

  • Layers: OSI defines 7 layers (splitting Application, Presentation, and Session into separate layers, and Physical and Data Link separately), whereas TCP/IP merges Application/Presentation/Session into one Application layer and Physical/Data Link into the Network Access layer (4 layers total, or 5 if Link and Physical are shown separately).
  • Design Order: OSI was specified as a conceptual standard before widespread implementations existed; TCP/IP grew out of working DARPA protocols (TCP, IP, ARP, DNS) and was codified afterward.
  • Layer Boundaries: OSI enforces strict layer isolation, whereas TCP/IP allows practical cross-layer optimization.

PDUs and Encapsulation: At the sender, Application data (Message) passes down to Transport, which adds a TCP/UDP header (Segment / Datagram); the Internet layer adds an IP header (Packet); the Network Access layer adds a MAC header and FCS trailer (Frame) and transmits Bits on the wire. The receiver reverses the process (decapsulation) layer by layer.

Answer:

OSI has 7 theoretical layers designed before implementation, whereas TCP/IP has 4 practical layers built around working Internet protocols. TCP/IP PDUs from top to bottom are Message (Application), Segment/Datagram (Transport), Packet (Internet), and Frame/Bits (Network Access), formed via encapsulation at the sender and stripped via decapsulation at the receiver.

Question 2

Given the IPv4 address 10.20.30.105/27:

  1. Find the dotted-decimal subnet mask.
  2. Determine the Network ID, Broadcast address, and valid host address range.
  3. Calculate the total number of /27 subnets that can be carved out of a /24 block and the usable hosts per subnet.

Subnet Mask for /27: There are 32 - 27 = 5 host bits. Block size in the 4th octet = 2^5 = 32. The 4th octet of the mask is 256 - 32 = 224, giving 255.255.255.224.

Network ID, Broadcast Address, and Host Range:

  • Dividing 105 by block size 32 gives 105 // 32 = 3, so the subnet starts at 3 * 32 = 96.
  • Network ID: 10.20.30.96
  • Broadcast Address: 96 + 32 - 1 = 127 -> 10.20.30.127
  • Usable Host Range: 10.20.30.97 to 10.20.30.126

Subnets per /24 and Usable Hosts:

  • Borrowing 27 - 24 = 3 bits from a /24 yields 2^3 = 8 subnets.
  • Each /27 subnet provides 2^5 - 2 = 32 - 2 = 30 usable hosts.
Answer:
  1. Subnet mask: 255.255.255.224
  2. Network ID: 10.20.30.96, Broadcast: 10.20.30.127, Usable range: 10.20.30.97 – 10.20.30.126
  3. 8 subnets per /24, with 30 usable hosts per subnet.
Question 3

Explain five major technical limitations of IPv4 and discuss why Network Address Translation (NAT) cannot serve as a permanent solution to IPv4 exhaustion.

Five Major Limitations of IPv4:

  1. 32-bit Address Exhaustion: Only ~4.29 billion addresses exist, already depleted at IANA and RIRs.
  2. Variable Header & Per-Hop Checksum: 20–60 byte header with options and a header checksum that must be recalculated at every router hop as TTL decrements.
  3. Router-Based Fragmentation: Intermediate routers fragment oversized packets, degrading forwarding speed and amplifying packet loss.
  4. Broadcast-Dependent Configuration: Relies on broadcast ARP and external DHCP rather than built-in stateless autoconfiguration.
  5. Routing Table Bloat & Add-on Security: Legacy allocations bloat BGP tables; IPsec was retrofitted rather than integrated with end-to-end reachability.

Why NAT Is Not a Permanent Solution:

  • Breaks End-to-End Principle: Incoming connections to internal servers fail without manual port forwarding, breaking peer-to-peer apps, VoIP, and SIP.
  • Stateful Bottleneck & Port Exhaustion: Carrier-Grade NAT (CGNAT) devices must track millions of flows and exhaust 65,535 TCP/UDP ports per public IP.
  • Interferes with IPsec & Accountability: Modifying IP/port headers breaks IPsec AH integrity checks and obscures user traceability across shared public IPs.
Answer:

IPv4 suffers from 32-bit address exhaustion, per-hop header checksum recomputation, router-side fragmentation, broadcast-heavy configuration, and routing table bloat. NAT only masks address scarcity by breaking end-to-end transparency, introducing stateful bottlenecks and port exhaustion, and interfering with P2P protocols and IPsec.

Question 4

A small campus receives 198.51.100.0/24 and must create four subnets using VLSM:

Assign the CIDR block and usable host range for each subnet in decreasing order of size, and find the number of addresses left over.

Determining Subnet Sizes:

  • Lab 1 (100 hosts): 2^7 - 2 = 126 >= 100 -> /25 (128 addresses).
  • Lab 2 (50 hosts): 2^6 - 2 = 62 >= 50 -> /26 (64 addresses).
  • Admin (25 hosts): 2^5 - 2 = 30 >= 25 -> /27 (32 addresses).
  • Router Link (2 hosts): 2^2 - 2 = 2 >= 2 -> /30 (4 addresses).

Contiguous Assignment from 198.51.100.0/24 (256 total addresses):

  1. Lab 1: 198.51.100.0/25 (usable .1 – .126, broadcast .127)
  2. Lab 2: 198.51.100.128/26 (usable .129 – .190, broadcast .191)
  3. Admin: 198.51.100.192/27 (usable .193 – .222, broadcast .223)
  4. Router Link: 198.51.100.224/30 (usable .225 – .226, broadcast .227)

Leftover Addresses: Total used = 128 + 64 + 32 + 4 = 228 addresses. Remaining = 256 - 228 = 28 addresses (198.51.100.228 – 198.51.100.255).

Answer:
  • Lab 1: 198.51.100.0/25 (usable .1 – .126)
  • Lab 2: 198.51.100.128/26 (usable .129 – .190)
  • Admin: 198.51.100.192/27 (usable .193 – .222)
  • Router Link: 198.51.100.224/30 (usable .225 – .226)
  • Remaining: 28 addresses (.228 – .255)
Question 5

Why is next-generation IP addressing (IPv6) considered vital for IoT, 5G, and cloud-scale networks? Summarize six key architectural advantages of IPv6 over IPv4.

Vital Role in IoT, 5G, and Cloud Networks: 5G and IoT connect tens of billions of sensors, vehicles, and mobile UEs, while cloud platforms spin up millions of ephemeral containers and virtual machines. IPv4 cannot provide globally unique addresses at this scale without multi-layered NAT, which adds latency, management complexity, and state overhead.

Six Key Advantages of IPv6 Over IPv4:

  1. 128-bit Address Space: 2^128 addresses eliminate scarcity and restore end-to-end IP reachability.
  2. Fixed 40-Byte Header: Faster router parsing with no options inside the main header and no per-hop header checksum.
  3. No Router Fragmentation: End hosts use Path MTU Discovery, preventing intermediate router bottlenecks.
  4. Stateless Address Autoconfiguration (SLAAC): Plug-and-play host configuration without requiring broadcast ARP or mandatory DHCP servers.
  5. Hierarchical Route Aggregation: Clean prefix delegation keeps global BGP routing tables smaller.
  6. Built-in Flow Label & IPsec Readiness: Dedicated 20-bit Flow Label supports QoS and real-time classification, while end-to-end addressing enables clean IPsec deployment.
Answer:

IPv6 provides the massive 128-bit address space needed for billions of IoT, 5G, and cloud endpoints without NAT. Its six core advantages are: (1) 128-bit global addressing, (2) fixed 40-byte header with no per-hop checksum, (3) source-only fragmentation via PMTUD, (4) plug-and-play SLAAC, (5) hierarchical route aggregation, and (6) native Flow Label QoS support and clean end-to-end IPsec compatibility.

Next Generation Internet Protocol →