Skip to content

Link-Activation Program

Activation process — instantiates Spanning Connectivity Formation

Crosses the connectivity threshold by enabling, repairing, authorizing, or synchronizing a deliberately chosen set of links until isolated components merge into one spanning network.

Link-Activation Program is the execution mechanism that actually turns the links on. It takes a chosen set of candidate links and a budget, and drives the network across an explicit spanning criterion by working through the set — enabling dormant links, repairing broken ones, authorizing forbidden ones, and synchronizing out-of-step ones. Unlike a mesh flood, it is targeted: a small, deliberately selected set of high-leverage links activated to merge large components. And unlike an adapter rollout, the links here already carry meaning once live — the barrier is that they are off, not that the two sides cannot understand each other. Its defining discipline is that it stops when the spanning criterion is met, not when the link list or the budget runs out.

Example

A regional internet provider's traffic to a neighbouring network currently hairpins hundreds of miles through a distant transit provider: the two networks are reachable in principle but not directly connected, so paths are slow and every byte is billed twice. The program joins an Internet Exchange Point and activates a chosen set of BGP peering sessions with the networks carrying the most local traffic. Each activation is really an authorize-and-synchronize: a peering agreement plus a matched routing configuration on both sides. The budget covers exchange-port fees and engineering time, spent first on the peers that pull the most traffic off transit. The program stops when the spanning criterion is met — every major local network directly reachable in one hop — after which local traffic stays local and two transit-dependent islands are one directly-connected region.

How it works

  • Take the prioritized link set. The program consumes a ranked candidate set (from Bridge-Edge Prioritization); it does not compute the leverage itself.
  • Resolve what "activation" means per link. Enable, repair, authorize, or synchronize — an authorization-limited link needs negotiation, not engineering, and vice versa.
  • Spend the budget largest-merge-first. Front-load the links that join the biggest still-separate components, so reachability climbs fastest per unit spent.
  • Check the criterion and stop. Halt at connectivity achieved, not at links exhausted — the criterion, not the backlog, is the terminating condition.

Tuning parameters

  • Activation-mode mix — how many links are off for engineering reasons (enable/repair) versus permission reasons (authorize/synchronize); the two need entirely different work.
  • Sequencing — biggest-merge-first (fastest reachability) versus cheapest-first (early wins, momentum, but slower merging).
  • Budget reserve — how much to hold back for links that prove harder to activate than scoped.
  • Stop-criterion strictness — halt at a single spanning path, or keep going for a redundant one (which hands off to Redundant Network Link).
  • Reversibility — whether a newly authorized link is provisional and can be pulled if it misbehaves, or permanent once lit.

When it helps, and when it misleads

Its strength is concentrating scarce activation effort on the few links that genuinely merge components, turning "we added a lot of links" into "we became reachable." Its failure mode is activating links inside an already-connected component: much new capacity, zero new reachability, because the merging leverage was miscomputed or ignored. The value lives in the bridge links that span the gaps between components, not the redundant ones added where a path already exists.[1] The classic misuse is running the program against a link-count target rather than the spanning criterion, so it reports success while components stay split. The discipline is to score success by components merged and let the spanning criterion, not the budget, decide when to stop.

How it implements the components

Link-Activation Program fills the execution side of the archetype's machinery — turning a chosen set of links into a crossing:

  • candidate_activation_set — it operates the deliberately chosen set of links to be enabled, repaired, authorized, or synchronized.
  • activation_or_repair_budget — it allocates the effort and money across that set, front-loading the highest-merging links.
  • spanning_connectivity_criterion — the criterion is its stop rule; the program drives the network until the criterion is satisfied.

It does not compute which links carry the most merging leverage (component_merging_leverage_score — Bridge-Edge Prioritization and Component-Merge Simulation), fix incompatibility rather than absence (Interoperability Adapter Rollout), or keep the links alive afterward (Link-Maintenance Sweep).

References

[1] The connectivity value of a link depends on whether it bridges otherwise-separate components rather than adding another path within one — the intuition behind Granovetter's observation that weak, bridging ties carry information between clusters that dense in-group ties cannot. A link-activation program earns its keep on bridge links, not redundant ones. ↩