1. Don’t panic, but carry a towel
Douglas Adams’ The Hitchhiker’s Guide to the Galaxy advises travelers to always know where their towel is: a simple, reliable piece of infrastructure that solves many problems. A self-reconfiguring network needs its own towel — a few simple, reliable primitives that the cells can use to find each other, authenticate, and route around trouble.
This entry lists what the desktop would have to provide for self-reconfiguring networks to work.
2. Physical provisions
The network is only as flexible as the physical layer allows:
- Redundant paths: each cell must have at least two independent ways to reach the rest of the network.
- Common interfaces: every cell uses the same electrical and protocol standards, even if the internal implementation differs.
- Hot-swap support: links must tolerate appearance and disappearance of cells without requiring a global reset.
- Deterministic latency bounds: real-time traffic such as control loops or synchronization must not be disrupted by reconfiguration.
3. Discovery and identity
The cells must know who they are and who their neighbors are:
- Persistent identity: each cell has a unique, cryptographically backed identity.
- Neighbor discovery: cells announce themselves and listen for announcements on each physical link.
- Capability advertisement: a cell reports what services it provides, such as compute, storage, power, or routing.
- Topology map: the federation maintains a distributed, eventually consistent map of who is present and how they are connected.
4. Policy and security
A network that reconfigures itself must still obey rules:
- Authentication: a new cell must prove it belongs before it is trusted.
- Authorization: not every cell gets access to every service.
- Policy constraints: ground can define global rules, such as “the power bus controller must always have two independent paths to ground.”
- Quarantine: a misbehaving cell can be isolated while remaining physically connected.
5. Decision and verification
The cells need a way to decide and to prove their decisions were correct:
- Health metrics: each link and cell reports latency, error rate, packet loss, and power status.
- Routing algorithm: a distributed algorithm chooses routes based on metrics and policy.
- Audit log: every reconfiguration is logged with cause and effect so ground can replay it.
- Rollback: if a new topology causes worse performance, the network can revert to a known-good state.
6. What this changes
- Self-reconfiguration requires investment in identity, discovery, policy, and audit infrastructure.
- Many of these provisions are worth making even for a ground-commanded network.
- The next entry will decide whether the desktop should design for this capability.