Programmable Messaging API  ·  View 19 of 22  ·  Operations

Observability

Signal by stage, with one row for the failures that produce no error at all.

Editable source SVG draw.io All views
Submit Admission Ledger Scheduling Attempt and session Ingest and state Rate and saturation Accepted per second Rejections by reason Write latency p99 Queue age by class Lease waits per route Receipts per second Correctness Dedup hit rate Consent refusals Append failures TTL expiries Ambiguous submit rate Uncorrelatable receipts The ones nobody else watches Accepted but never attempted Rule set version in use Projection lag Bulk starving interactive Delivery rate collapse Unknown-state share SLO and alert Submit availability 99.99 Consent eval 99.99 RPO 0 verified Interactive age p99 3 s Route breaker trips Receipt visible p95 2 s Evidence and cost Per-tenant submit trail Rule version per refusal Ledger write cost share Fairness share per route Cost per segment by route Margin and quality together Observability — Signal By Stage, And The Ones Nobody Else Watches Row three is the point of the view: silent collapse and unknown share are the failures with no error. v 1.0 · owner Integration Platform Architecture

Decisions

  • Row three is the reason the view exists: delivery-rate collapse, unknown share, projection lag and bulk starving interactive are all invisible to conventional monitoring.
  • Unknown share is tracked per route and treated as a route-quality signal, so a carrier that never reports loses traffic rather than hiding.
  • Cost per segment and observed quality are reported together, because least-cost routing that halves delivery rate is not a saving.

Assumptions

  • Interactive queue age p99 ≤ 3 s regardless of bulk volume; receipt visible p95 ≤ 2 s.
  • Platform compute ≤ 8% of carrier segment cost, of which the ledger and receipt write path is ≤ 35%.

Risks

  • "Accepted but never attempted" should always be zero; an alert that never fires is the one most likely to be deleted.
◀  Delivery Pipeline All views Route Lifecycle  ▶
Solution Architecture v1.0 · Google Cloud · Integration Platform Architecture · 2026-10 · 22 views · 16 architecture decision records19-observability