Attestation store layout and read cache¶
- Kind: decision
- Date: 2026-10-27
- Source: moved from
how-it-works/ci/attestation-store.md; #11972, #11973, #11979, #15554 - Shaped: Attestation store: layout and read cache
Note
This is a recommendation. #11973 is labeled needs-decision and stays open until the owner accepts or
changes it.
Context¶
The decision service (#11979) needs a schema for attestations in Neon. Already decided: Neon is the only authoritative store, rows are append-only, expiry is applied at read time, and no GitHub-side cache may be the sole basis for a skip.
Options¶
- One
attestationstable for every tier, or one table per tier. - A read cache in v1, or none.
Decision¶
One table. Every read is “tier T, input hash H”. Per-tier tables would give three identical read paths and three migration tracks for no query that needs the difference. The per-tier payload schemas already differ in the envelope bytes, and the type tag routes validation; columns are not where that variety belongs.
No cache in v1. A PR makes a handful of lookups (one per gated job, at job start). Neon in the same region as the service answers a primary-key index lookup in single-digit milliseconds. The cost that can matter is a scale-to-zero cold start, which a cache does not fix (the first request after idle still pays it). The answer to that is keeping the compute warm or a pooled connection. If a cache is needed later it stays in-process in the Go service: anything outside the process boundary is the same trust class as the GitHub-side caches already refused.
Consequences¶
The page records the layout, the revisit triggers (p95 above 250 ms warm, or Neon connection limits) and the rules any later cache must follow.