G04-TS: No
Rationale
For a single tailnet, the documented managed architecture places its complete control plane on one shard, served by one Go process with an SQLite database. More shards distribute different tailnets; migration moves a tailnet between shards rather than partitioning its coordination work. Thus the selected single-tailnet network retains an obligatory coordinator capacity boundary and does not meet G04.
Assessment method
Primary deployed-architecture description; scoped bottleneck inference, no load test
Source context
Tailscale database architecture, Alex Chan, 2026-08-12
Outstanding qualification
Scope is one tailnet under the architecture described by Tailscale on 2026-08-12. The overall service does scale horizontally across tailnets. Migration, recovery and larger hosts remain possible; this is not a measured node ceiling, a claim that the whole service has one coordinator, or a G05 global-plausibility verdict. Existing peer traffic can continue during control-plane outages, while new discovery and updates depend on the shard. A redesigned intra-tailnet partitioning mechanism or a separately specified multi-tailnet composition would require reassessment.
Public sources
Editorial evidence references
The following identifiers refer to research records not included in this public package; they are not public links.
- round595.html
- round595-sources.json