The Complex Legacy of Networking and the Missed Opportunities of IPv6

I attended an IETF meeting to see the presentation of TCP BBR, which was met with cautious optimism. The event was filled with discussions about IPv6, the successor to IPv4, which many believe will eventually replace it. However, IPv6 is often criticized for its complexity compared to IPv4, which raises the question of why it wasn't designed as a simple extension of IPv4 with more address bits.

In the early days of networking, systems like the telephone network used physical circuit switching, which didn't require addresses because each connection was a direct line. As technology evolved, time-division multiplexing allowed for more efficient use of wires, leading to the development of IP addresses and routing. Local area networks (LANs) introduced bus networks, where multiple devices could connect to a single wire, leading to the creation of MAC addresses to manage device identification.

Ethernet emerged as a dominant LAN technology, using 48-bit MAC addresses to uniquely identify devices. However, as networks grew, the limitations of bus networks became apparent, leading to the development of bridging and switching to interconnect multiple networks. This system, while fast, was complex and difficult to debug.

The Internet's evolution involved connecting entire LANs over long distances, which required new protocols to manage congestion and routing. The introduction of ARP and DHCP added layers of complexity, as these protocols were necessary to map IP addresses to MAC addresses and automate IP address configuration.

Despite the progress, the reliance on bridging and the intertwining of Ethernet and IP protocols created a tangled web of dependencies. The IETF's vision for IPv6 was to simplify this by eliminating bus networks, broadcasts, and MAC addresses, allowing for more efficient routing and address allocation. However, the transition to IPv6 has been slow, and the legacy of IPv4 and Ethernet remains entrenched.

The concept of mobile IP, which allows devices to maintain connections while moving across networks, was not addressed in the initial design of IPv6. This oversight has led to reliance on layer 2 bridging to maintain connectivity, especially in mobile and Wi-Fi networks. The potential solution involves using protocols like QUIC, which can handle mobility by using unique session identifiers, allowing for seamless roaming without relying on traditional IP and MAC address configurations.

Ultimately, the dream of a simplified, elegant Internet remains unfulfilled, as legacy systems and protocols continue to dominate. The transition to a more efficient network architecture requires overcoming significant technical and infrastructural challenges.