System 12.23 — Notification Platform
In one line
The notification system (12.3) elevated to a PLATFORM team product: tenants, templates, delivery SLAs, self-service.
Think of it like this
The same notification idea from earlier, sold as a whole team's shared product used by every other team in the company, instead of built separately by each team.
Key ideas
- 01
Tenants: hundreds of internal teams subscribe to a shared delivery platform — tenant isolation + quotas + self-service console.
- 02
Multi-channel with provider failover: channel → provider pool → circuit breaker per provider → automatic failover on outage.
- 03
Template + content versioning with audit; dynamic rendering; per-tenant approval workflow.
- 04
Delivery guarantees: per-tenant QoS bands (gold = min 99.9% deliver within 5min, silver = best effort).
- 05
Events & analytics: every send emits telemetry (attempts, latency, provider, result) → ops dashboards + tenant billing.
- 06
Batching, bloom/dedupe, quiet-hours sending, per-user channel preference — all platform features, not per-app.
- 07
Interview framing: it's 12.3 + multitenancy + provider circuit breaking + self-service — a platform story.
Code & diagrams
Same pipeline as System 12.3, now exposed as a self-serve product with its own onboarding and dashboard.
Explain without notes
A primary provider fails for 2 hours. Walk the failover, the throttle guard, and what tenants are told.
Practice
Design the tenant model + quota isolation (keys, limits) + the self-service console surface.
Trade-offs
- ↔
One platform serves everyone (efficiency) vs per-team silos (isolation) — the platform must be boring and safe.
Completion checklist
I can present notification-as-a-platform with tenancy, failover, and SLAs.