Chaosnet (1981)

1. Introduction

Chaosnet is a specialized local area network designed for communication between computers situated within a relatively short range—typically $\le 2\text{ km}$.

The designation "Chaosnet" is derived from the system's fundamental architecture: it lacks any form of centralized control.

Origins and Adoption

Developed in 1975 by the Massachusetts Institute of Technology (MIT) within the Artificial Intelligence Laboratory, it initially served as the primary communication backbone for the Lisp Machine system [ CHINUAL , AIM444 ]. Over time, its utility expanded to connect various hardware types across the MIT campus and eventually spread to other research facilities and universities.

The Lisp Machine Ecosystem

The Lisp Machine architecture utilizes a multi-processor approach. In this setup, every active user is provided with a dedicated workstation comprising:

• A medium-scale processor.
• Sufficient local memory.
• A dedicated swapping disk.

Despite this local autonomy, files are managed via a central file-system accessed through the network. This hybrid approach is summarized below:

Feature	Traditional Time-Sharing	Lisp Machine + Chaosnet
Processing	Shared CPU (Time-sliced)	Dedicated CPU per user
File Access	Centralized	Centralized via Chaosnet
Inter-user Comm.	Supported	Supported
Lisp Performance	Slower for massive programs	High efficiency for multi-million word programs
Response Time	Variable	Rapid interactive response

System Requirements and Objectives

Because Chaosnet effectively replaces the local disk in a standard computer, it must meet rigorous standards:

1. Velocity: High throughput and low latency.
2. Resilience: High reliability (hence the decentralized nature).
3. Capacity: Ability to support several dozen connected machines.

Beyond file storage, Chaosnet facilitates access to shared peripherals such as tape drives, printers, and unique I/O devices or specialized processors.

Design Philosophy

The primary objectives were simplicity and high performance. Performance was not sought through complex algorithms, but rather by utilizing a high-speed transmission medium with minimal overhead.

Why Simplicity?

• Maintenance: Since the network connects diverse hardware, multiple implementations exist; lower complexity reduces the maintenance burden for each.
• Management: Simplifies the non-trivial task of managing a heterogeneous network used by various groups.
• Diagnostics: Allows administrators to quickly pinpoint whether a failure is caused by the physical cable, a specific host's software, or its hardware.

Scope and Constraints

The designers intentionally ignored problems that do not typically affect local networks. Consequently, Chaosnet is not suitable for:

• Long-distance or continental links
• Satellite communications
• High-error/noisy links
• Low-speed connections
• Complex multi-path routing

Furthermore, it omits "luxury" features like multiple service levels or native security (though users may implement end-to-end encryption).

Architectural Influence

Chaosnet drew significant inspiration from the early local networking experiments at Xerox PARC [ ETHERNET ]. It is split into two interdependent layers:

1. Hardware: A carrier-sense multiple-access (CSMA) structure.
2. Software: Higher-level protocols based on packets.

While designed for the Chaosnet cable, the software is flexible enough to operate over different media or across multiple linked cables. Its logic incorporates elements from TCP, Arpanet, and Ethernet.

---

2. Hardware Protocol

2.1 The Ether

The physical medium, known as the ether, consists of a semi-rigid 1/2 inch low-loss coaxial cable.

Technical Specifications:

• Termination: $75\Omega$ at both extremities.
• Topology: Strictly linear.
  • Prohibited: Branches, stubs, or circular loops.

Node Composition

A network node is comprised of the following components:

1. Transceiver: Attached directly to the coaxial cable.
2. Flat Cable: A 10-meter link connecting the transceiver to the interface.
3. Interface: Connected to the computer's I/O bus.
4. NCP: The Network Control Program software that manages the network logic.

Transmission Logic

The network operates on a "one-at-a-time" basis. A single node seizes the ether to broadcast a packet. Every other node on the line receives the signal and determines in hardware whether to process the packet or discard it.

Capacity and Scaling:

• Node Limit: Typically a few dozen nodes. This limit is imposed by the electrical degradation caused by the connectors.
• Extension: While repeaters (bidirectional digital amplifiers) can extend the cable length, Chaosnet typically uses bridges.

Bridges are nodes (often pdp11 computers) equipped with two or more interfaces. They serve to:

• Relay packets between different ethers.
• Connect different types of network media.
• Perform auxiliary tasks, such as managing terminals.

Implementation Checklist

• High-speed transmission medium
• Carrier-sense multiple access
• Decentralized control
• Linear coaxial topology
• Global scalability (Out of scope)

Example Packet Logic (Conceptual)

/* Simplified conceptual logic for packet handling */
if (packet.destination == MY_ADDRESS || packet.destination == BROADCAST) {
    receive_packet(packet);
} else {
    ignore_packet();
}