Notification System — System Design Academy

Complete System Design Tutorial

The design separates the latency-sensitive online path, the correctness-critical durable transition, and asynchronous projection or repair work. Each chapter below explains one decision shown in the diagrams.

1. Requirements and scope

Functional requirements

Execute the primary notification system workflow and return an observable result.
Expose the concrete operations shown below with stable retry identities.
Allow operators to inspect, disable, repair, or replay ambiguous state.
Keep derived analytics and notifications off the correctness-critical path.

Non-functional requirements

Design for 500M notifications/day · burst 250K/s · minutes of retention.
Protect the named business invariant under concurrency and partial failure.
Scale by an explicit aggregate/partition key without one global mutex.
Degrade intentionally when cache, queue, store, or regional ownership is unavailable.

Primary invariant: each accepted notification system operation has one stable result and retries do not repeat its business effect.

↑ top

2. Capacity assumptions

State assumptions before choosing technology. The supplied workload is 500M notifications/day · burst 250K/s · minutes of retention. Convert it into peak throughput, retained records, bandwidth, and failure budget; then connect each number to a component.

Dimension	Working estimate	Design implication
Traffic	500M notifications/day · burst 250K/s · minutes of retention	Separate online compute, durable state, and asynchronous consumers.
Peak factor	5–10× average	Admission control, autoscaling headroom, and hot-key protection.
Stored state	write rate × retention × record bytes	Partition count, indexes, compaction, backup, and archival tier.
Network	peak operations × payload size	Batching, compression, edge placement, and egress budget.
Availability	99.9–99.99%	Multi-zone replicas, tested failover, and dependency error budgets.
Recovery	RPO/RTO by invariant	Ack boundary, replay retention, reconciliation, and operator runbooks.

Do not cargo-cult the numbers: show the arithmetic, name the uncertain variable, and say which production metric will replace the interview estimate.

↑ top

3. API and data model

POST /v1/notificationsaccept one idempotent send

POST /v1/campaignsenqueue fan-out campaign

DELETE /v1/notifications/{id}cancel if not claimed

GET /v1/notifications/{id}delivery status

Authoritative records and indexes

notification(id PK, user_id, template_id, priority, schedule_at, state)

delivery(notification_id, channel, attempt, provider_id, status)

preference(user_id, channel, topic, enabled, quiet_hours)

UNIQUE producer_id,idempotency_key

Command and event envelopes

# Command envelope
operation: POST /v1/notifications
request_id: req_01J...
idempotency_key: idem_7f2...
aggregate_key: notification-system_123
expected_version: 17
payload: {...}

# Durable event
event_id: evt_01J...
aggregate_key: notification-system_123
event_type: notification_system_committed
schema_version: 3
trace_id: tr_8ac...

The authoritative aggregate and idempotency record share a transactional or conditional-write boundary. Event streams, search indexes, caches, analytics, and dashboards are rebuildable projections.

↑ top

4. Why this architecture and what it costs

The boxes are not a shopping list. Each boundary exists because it owns a different invariant, scaling dimension, data lifecycle, security policy, or failure domain. The table makes those choices explicit and shows when the design should remain simpler.

Component	Owned responsibility	Why separate it	Rejected alternative	Cost and evolution trigger
Product Service	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Resolve recipient identity, consent, quiet hours, and required channels.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents the external caller from also owning the contract and failure behavior of Notification API.	Keep this behavior inside an adjacent service. Prefer that simpler option until independent scale, ownership, policy, or failure isolation is measurable. Here that simpler choice would couple “Provider returns success after worker timeout” to the external caller.	Costs an interface, deployment, telemetry, ownership, and failure mode. Keep it in-process first; extract when independent scale or reliability pays for the boundary. Measure the latency, saturation, lag, or mismatch at the Product Service boundary before extracting or scaling it.
Notification API	Owns authentication, validation, quotas, deadlines, and routing. In this design it participates in: Render a versioned template and persist the notification before acknowledgment.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Product Service from also owning the contract and failure behavior of Preference Service.	Put protocol, authentication, quotas, and business transitions in one process. Acceptable at low scale, but it couples external traffic risk to domain releases. Here that simpler choice would couple “Bulk campaign starves password resets” to Product Service.	Costs another hop, policy distribution, and a high-blast-radius fleet. Introduce it when many clients or services need one protocol and protection boundary. Measure the latency, saturation, lag, or mismatch at the Notification API boundary before extracting or scaling it.
Preference Service	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Route by priority and channel to replayable partitions.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Notification API from also owning the contract and failure behavior of Template Renderer.	Keep this behavior inside an adjacent service. Prefer that simpler option until independent scale, ownership, policy, or failure isolation is measurable. Here that simpler choice would couple “Preference changes after scheduling” to Notification API.	Costs an interface, deployment, telemetry, ownership, and failure mode. Keep it in-process first; extract when independent scale or reliability pays for the boundary. Measure the latency, saturation, lag, or mismatch at the Preference Service boundary before extracting or scaling it.
Template Renderer	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: A channel worker claims a delivery lease, applies provider rate limits, and sends.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Preference Service from also owning the contract and failure behavior of Schedule Store.	Keep this behavior inside an adjacent service. Prefer that simpler option until independent scale, ownership, policy, or failure isolation is measurable. Here that simpler choice would couple “Poison template repeatedly fails rendering” to Preference Service.	Costs an interface, deployment, telemetry, ownership, and failure mode. Keep it in-process first; extract when independent scale or reliability pays for the boundary. Measure the latency, saturation, lag, or mismatch at the Template Renderer boundary before extracting or scaling it.
Schedule Store	Owns authoritative versioned state or immutable evidence. In this design it participates in: Verify provider callbacks, dedupe by provider event ID, and update the ledger.	Defines the source of truth and commit boundary. It prevents Template Renderer from also owning the contract and failure behavior of Priority Router.	Let each service persist its own interpretation. That removes one clear commit boundary and makes retries, support, and reconciliation ambiguous. Here that simpler choice would couple “Provider outage fills retry queues” to Template Renderer.	Costs schema migration, replication, backup, and shard ownership. Split storage only when access patterns, retention, or compliance boundaries materially differ. Measure the latency, saturation, lag, or mismatch at the Schedule Store boundary before extracting or scaling it.
Priority Router	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Resolve recipient identity, consent, quiet hours, and required channels.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Schedule Store from also owning the contract and failure behavior of Kafka Topics.	Keep this behavior inside an adjacent service. Prefer that simpler option until independent scale, ownership, policy, or failure isolation is measurable. Here that simpler choice would couple “Provider returns success after worker timeout” to Schedule Store.	Costs an interface, deployment, telemetry, ownership, and failure mode. Keep it in-process first; extract when independent scale or reliability pays for the boundary. Measure the latency, saturation, lag, or mismatch at the Priority Router boundary before extracting or scaling it.
Kafka Topics	Preserves durable ordering, replay, and consumer checkpoints. In this design it participates in: Render a versioned template and persist the notification before acknowledgment.	Decouples bursts and failure from user acknowledgment. It prevents Priority Router from also owning the contract and failure behavior of Email Worker.	Call every downstream consumer synchronously. That couples user latency and availability to analytics, notification, indexing, and audit dependencies. Here that simpler choice would couple “Bulk campaign starves password resets” to Priority Router.	Costs eventual consistency, schema evolution, replay tooling, idempotent consumers, and lag operations. Earn it when bursts or downstream failure must not block acknowledgment. Measure the latency, saturation, lag, or mismatch at the Kafka Topics boundary before extracting or scaling it.
Email Worker	Executes idempotent asynchronous work under lease and retry policy. In this design it participates in: Route by priority and channel to replayable partitions.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Kafka Topics from also owning the contract and failure behavior of Push Worker.	Perform the work before acknowledging the caller. This gives simpler consistency but unbounded tail latency and fragile retries. Here that simpler choice would couple “Preference changes after scheduling” to Kafka Topics.	Costs leases, checkpoints, retries, DLQs, and lag operations. Separate once execution time or burstiness no longer fits the request deadline. Measure the latency, saturation, lag, or mismatch at the Email Worker boundary before extracting or scaling it.
Push Worker	Executes idempotent asynchronous work under lease and retry policy. In this design it participates in: A channel worker claims a delivery lease, applies provider rate limits, and sends.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Email Worker from also owning the contract and failure behavior of SMS Worker.	Perform the work before acknowledging the caller. This gives simpler consistency but unbounded tail latency and fragile retries. Here that simpler choice would couple “Poison template repeatedly fails rendering” to Email Worker.	Costs leases, checkpoints, retries, DLQs, and lag operations. Separate once execution time or burstiness no longer fits the request deadline. Measure the latency, saturation, lag, or mismatch at the Push Worker boundary before extracting or scaling it.
SMS Worker	Executes idempotent asynchronous work under lease and retry policy. In this design it participates in: Verify provider callbacks, dedupe by provider event ID, and update the ledger.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Push Worker from also owning the contract and failure behavior of Provider Adapters.	Perform the work before acknowledging the caller. This gives simpler consistency but unbounded tail latency and fragile retries. Here that simpler choice would couple “Provider outage fills retry queues” to Push Worker.	Costs leases, checkpoints, retries, DLQs, and lag operations. Separate once execution time or burstiness no longer fits the request deadline. Measure the latency, saturation, lag, or mismatch at the SMS Worker boundary before extracting or scaling it.
Provider Adapters	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Resolve recipient identity, consent, quiet hours, and required channels.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents SMS Worker from also owning the contract and failure behavior of Delivery Ledger.	Keep this behavior inside an adjacent service. Prefer that simpler option until independent scale, ownership, policy, or failure isolation is measurable. Here that simpler choice would couple “Provider returns success after worker timeout” to SMS Worker.	Costs an interface, deployment, telemetry, ownership, and failure mode. Keep it in-process first; extract when independent scale or reliability pays for the boundary. Measure the latency, saturation, lag, or mismatch at the Provider Adapters boundary before extracting or scaling it.
Delivery Ledger	Owns authentication, validation, quotas, deadlines, and routing. In this design it participates in: Render a versioned template and persist the notification before acknowledgment.	Defines the source of truth and commit boundary. It prevents Provider Adapters from also owning the contract and failure behavior of DLQ / Replay.	Let each service persist its own interpretation. That removes one clear commit boundary and makes retries, support, and reconciliation ambiguous. Here that simpler choice would couple “Bulk campaign starves password resets” to Provider Adapters.	Costs schema migration, replication, backup, and shard ownership. Split storage only when access patterns, retention, or compliance boundaries materially differ. Measure the latency, saturation, lag, or mismatch at the Delivery Ledger boundary before extracting or scaling it.
DLQ / Replay	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Route by priority and channel to replayable partitions.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Delivery Ledger from also owning the contract and failure behavior of Webhook Ingest.	Keep this behavior inside an adjacent service. Prefer that simpler option until independent scale, ownership, policy, or failure isolation is measurable. Here that simpler choice would couple “Preference changes after scheduling” to Delivery Ledger.	Costs an interface, deployment, telemetry, ownership, and failure mode. Keep it in-process first; extract when independent scale or reliability pays for the boundary. Measure the latency, saturation, lag, or mismatch at the DLQ / Replay boundary before extracting or scaling it.
Webhook Ingest	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: A channel worker claims a delivery lease, applies provider rate limits, and sends.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents DLQ / Replay from also owning the contract and failure behavior of operators and repair tooling.	Keep this behavior inside an adjacent service. Prefer that simpler option until independent scale, ownership, policy, or failure isolation is measurable. Here that simpler choice would couple “Poison template repeatedly fails rendering” to DLQ / Replay.	Costs an interface, deployment, telemetry, ownership, and failure mode. Keep it in-process first; extract when independent scale or reliability pays for the boundary. Measure the latency, saturation, lag, or mismatch at the Webhook Ingest boundary before extracting or scaling it.

Decision 01

Chosen boundary vs simpler monolith

Separate acceptance from delivery; return after a durable enqueue and process channels asynchronously.

A single deployable with one transactional store is the right starting point for a small notification system workload. The proposed boundary becomes justified when separate acceptance from delivery; return after a durable enqueue and process channels asynchronously. Separating it contains the named failure “Provider returns success after worker timeout” without placing unrelated work on the critical path.

Decision 02

Correctness vs availability

Use idempotency keys, per-channel retry policies, dead-letter queues, and provider circuit breakers.

The design coordinates only where each accepted notification system operation has one stable result and retries do not repeat its business effect. Everywhere else it accepts bounded staleness or asynchronous propagation. The alternative—strong coordination for every projection—simplifies the mental model but increases tail latency and turns partitions into broad write unavailability.

Decision 03

Operational cost and evolution

Resolve templates and preferences before fan-out; preserve a ledger for audit and user-visible status.

This choice adds dashboards, ownership, schema compatibility, capacity planning, and recovery runbooks. Start with the minimum topology that supports “Route by priority and channel to replayable partitions.” and extract another service only when measured throughput, blast radius, data lifecycle, or team ownership changes independently.

Why this design: the synchronous path contains only the decisions required to protect each accepted notification system operation has one stable result and retries do not repeat its business effect. Durable projections, notifications, indexing, analytics, and repair are isolated so their load or failure cannot corrupt the acknowledged operation.

Correctness boundary

Separate acceptance from delivery; return after a durable enqueue and process channels asynchronously.

Scale boundary

Use idempotency keys, per-channel retry policies, dead-letter queues, and provider circuit breakers.

Operational boundary

Resolve templates and preferences before fan-out; preserve a ledger for audit and user-visible status.

↑ top

5. Critical flow and algorithm

Resolve recipient identity, consent, quiet hours, and required channels.

Render a versioned template and persist the notification before acknowledgment.

Route by priority and channel to replayable partitions.

A channel worker claims a delivery lease, applies provider rate limits, and sends.

Verify provider callbacks, dedupe by provider event ID, and update the ledger.

Retry-safe mutation skeleton

identity = authenticate(request)
prior = idempotency.get(identity, request.key)
if prior: return prior.result

current = authoritative_store.read(partition_key)
next_state, events = domain.transition(current, request)
authoritative_store.compare_and_set(current.version, next_state)
idempotency.commit(request.key, next_state.result)
outbox.append(events)
return next_state.result

Exactly-once business effect: the network can duplicate requests. Stable identities, conditional state transitions, immutable event IDs, and reconciliation make the user-visible effect happen once.

↑ top

6. Caching, hot keys, and backpressure

Cache policy

Add a cache only for a measured repeated read; do not place mutable correctness state in a disposable layer.

Hot partition or tenant

Measure per-key and per-tenant skew. Replicate safe reads, isolate the offender, salt only when ordering permits, and enforce budgets.

Backpressure

Bound queues, buffers, connection pools, and concurrency. Shed low-priority work and scale from lag/oldest-item age rather than CPU alone.

Stampede and retry storms

Use single-flight/coalescing, jittered expiry, bounded retries, exponential backoff, and circuit breakers so one dependency failure does not multiply traffic.

Cache outage rule: never allow every miss to fall through simultaneously. Use local protection, request coalescing, circuit breakers, bounded pools, and controlled fallback.

↑ top

7. Partitioning and consistency

partition = stable_hash(primary_or_aggregate_key) → shard owner
write = conditional_update(expected_version)
read_projection = versioned and rebuildable

Strong boundary

each accepted notification system operation has one stable result and retries do not repeat its business effect. Use a uniqueness constraint, serializable/conditional write, expected version, or fencing token at this boundary.

Eventual boundary

Analytics, notifications, search, read models, and operational dashboards may lag when freshness is observable and replay exists.

Operation	Consistency	Reason
Authoritative mutation	Conditional / serializable	Protect uniqueness, ownership, money, inventory, order, or legal state transition.
Read after own write	Primary, session token, or cache warm	The caller must observe the result it just committed.
Ordinary read projection	Bounded stale where safe	Replicas and caches improve latency when version/freshness is visible.
Analytics and operations	Eventual + replay	Freshness is less important than isolating the online path.

Route every mutation for one aggregate to one owner/partition epoch.
Use cursors rather than offsets for changing ordered collections.
Retain tombstones until replication and repair windows close.
Move shard ownership through versioned maps and dual-read verification.

↑ top

8. Failure drills and recovery

Failure	Runtime behavior	Recovery proof
Provider returns success after worker timeout	Query/retry with the same operation identity; do not issue a second business effect.	Reconciliation returns zero invariant gaps; lag/replica health and user-visible SLO return to baseline.
Bulk campaign starves password resets	Isolate the key/tenant/priority, apply admission control, and preserve durable work.	Reconciliation returns zero invariant gaps; lag/replica health and user-visible SLO return to baseline.
Preference changes after scheduling	Stop unsafe progress, serve only explicitly safe degraded behavior, and recover from durable state.	Reconciliation returns zero invariant gaps; lag/replica health and user-visible SLO return to baseline.
Poison template repeatedly fails rendering	Stop unsafe progress, serve only explicitly safe degraded behavior, and recover from durable state.	Reconciliation returns zero invariant gaps; lag/replica health and user-visible SLO return to baseline.
Provider outage fills retry queues	Stop unsafe progress, serve only explicitly safe degraded behavior, and recover from durable state.	Reconciliation returns zero invariant gaps; lag/replica health and user-visible SLO return to baseline.

Recovery checkpoint

# Recovery checkpoint
scenario: Provider returns success after worker timeout
operation_id: op_01J...
aggregate_key: notification-system_123
last_durable_version: 18
projection_checkpoint: partition-07@offset-482901
retry_policy: same_identity / bounded_backoff
reconciliation_status: pending
owner: Webhook Ingest

Every dependency needs a timeout, retry budget, circuit policy, and explicit fail-open, fail-closed, stale, or read-only mode. “Highly available” is incomplete until degraded behavior is named.

↑ top

9. Security, regions, and observability

Security and abuse

Authenticate at ingress and authorize the exact resource/tenant on every operation.
Encrypt transport and durable data; tokenize or isolate sensitive material.
Rate-limit by tenant, actor, resource, and network; detect enumeration and anomalous fan-out.
Redact secrets and personal data before logs or analytical storage.
Audit privileged mutations, support access, policy changes, and forced recovery.

Signals that prove correctness

User-visible success rate and p50/p95/p99 latency.
Timeouts, circuit state, saturation, and replica health by dependency.
Queue lag, oldest-item age, retries, and dead-letter growth.
Conflict, duplicate-suppression, and reconciliation mismatch counts.
Business drill: Provider returns success after worker timeout — alert before the user-visible invariant is violated.

Regional evolution

Begin multi-zone with tested failover, backups, and reconciliation.
Add global stateless ingress and regional read projections.
Keep one write owner per aggregate or partition where coordination is expensive.
Move ownership with epochs/fencing; avoid silent simultaneous writers.
Use expand/contract schemas, canaries, shadow reads, and reversible cutovers.

↑ top

10. Interviewer follow-ups

How do scheduled notifications work?

Separate acceptance from delivery; return after a durable enqueue and process channels asynchronously. Walk the concrete lifecycle step “Resolve recipient identity, consent, quiet hours, and required channels.” Then test it against “Provider returns success after worker timeout”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How do you enforce user preferences?

Use idempotency keys, per-channel retry policies, dead-letter queues, and provider circuit breakers. Walk the concrete lifecycle step “Render a versioned template and persist the notification before acknowledgment.” Then test it against “Bulk campaign starves password resets”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How do you avoid duplicate sends?

Resolve templates and preferences before fan-out; preserve a ledger for audit and user-visible status. Walk the concrete lifecycle step “Route by priority and channel to replayable partitions.” Then test it against “Preference changes after scheduling”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How do you handle provider failover?

Separate acceptance from delivery; return after a durable enqueue and process channels asynchronously. Walk the concrete lifecycle step “A channel worker claims a delivery lease, applies provider rate limits, and sends.” Then test it against “Poison template repeatedly fails rendering”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How would priority traffic preempt bulk campaigns?

Use idempotency keys, per-channel retry policies, dead-letter queues, and provider circuit breakers. Walk the concrete lifecycle step “Verify provider callbacks, dedupe by provider event ID, and update the ledger.” Then test it against “Provider outage fills retry queues”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

Answer pattern: state the invariant, choose a mechanism, walk one failure, name the recovery signal, and explain the cost of the alternative.

↑ top

11. How to present this in 45 minutes

Time	What to cover
0–5 min	Clarify scope, workload, SLOs, and the correctness invariant.
5–10 min	Estimate throughput/storage and define concrete APIs and records.
10–20 min	Draw online, durable-write, and asynchronous/repair paths.
20–30 min	Explain the exact lifecycle, partition key, and concurrency control.
30–38 min	Discuss caching, hot keys, replication, and backpressure.
38–43 min	Run failure drills, security, observability, and regional evolution.
43–45 min	Summarize the five most important choices and trade-offs.

Final decision summary

Invariant	each accepted notification system operation has one stable result and retries do not repeat its business effect.
Partitioning	Stable aggregate/key hash with versioned ownership and measured hot-key isolation.
Commit	Expected version or uniqueness at the authoritative state boundary.
Async work	Durable outbox/log, idempotent consumers, checkpoints, and replay.
Recovery	Same operation identity, durable truth lookup, and business reconciliation.

Closing summary

CLOSING “The design protects each accepted notification system operation has one stable result and retries do not repeat its business effect. The online path stays bounded, authoritative state is versioned and partitioned, and asynchronous work is replayable and observable.”

↑ top

Product Service

Notification API

Preference Service

Template Renderer

Complete System Design Tutorial

1. Requirements and scope

Functional requirements

Non-functional requirements

2. Capacity assumptions

3. API and data model

Authoritative records and indexes

Command and event envelopes

4. Why this architecture and what it costs

Chosen boundary vs simpler monolith

Correctness vs availability

Operational cost and evolution

5. Critical flow and algorithm

Retry-safe mutation skeleton

6. Caching, hot keys, and backpressure

7. Partitioning and consistency

Strong boundary

Eventual boundary

8. Failure drills and recovery

Recovery checkpoint

9. Security, regions, and observability

Security and abuse

Signals that prove correctness

Regional evolution

10. Interviewer follow-ups

11. How to present this in 45 minutes

Final decision summary

Closing summary