Payment Processing — 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 payment processing 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 50K TPS · zero money loss · multi-day reconciliation.
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: no scarce resource or monetary effect is created twice, lost after acknowledgment, or committed without an audit trail.

↑ top

2. Capacity assumptions

State assumptions before choosing technology. The supplied workload is 50K TPS · zero money loss · multi-day reconciliation. Convert it into peak throughput, retained records, bandwidth, and failure budget; then connect each number to a component.

Dimension	Working estimate	Design implication
Traffic	50K TPS · zero money loss · multi-day reconciliation	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/payment_intentscreate idempotent intent

POST /v1/payment_intents/{id}/confirmauthorize/capture

POST /v1/refundsledger-backed refund

POST /v1/webhooks/{processor}signed callback

Authoritative records and indexes

payment_intent(id PK, merchant_id, amount, currency, state, version)

ledger_entry(txn_id, account_id, debit, credit, currency, effective_at)

processor_attempt(intent_id, attempt, provider_ref, status)

idempotency(merchant_id, key UNIQUE, request_hash, response)

Command and event envelopes

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

# Durable event
event_id: evt_01J...
aggregate_key: payment-system_123
event_type: payment_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
Checkout / Merchant	Owns client protocol, request identity, retry, and resume behavior. In this design it participates in: Authenticate merchant, validate amount/currency, and atomically create intent/idempotency result.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents the external caller from also owning the contract and failure behavior of Payment 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 “Client times out after successful capture” 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 Checkout / Merchant boundary before extracting or scaling it.
Payment API	Owns authentication, validation, quotas, deadlines, and routing. In this design it participates in: Score risk and resolve a tokenized funding instrument without exposing raw card data.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Checkout / Merchant from also owning the contract and failure behavior of Idempotency Store.	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 “Processor returns unknown outcome” to Checkout / Merchant.	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 Payment API boundary before extracting or scaling it.
Idempotency Store	Owns authoritative versioned state or immutable evidence. In this design it participates in: Route to a processor and record each uncertain attempt before external I/O.	Defines the source of truth and commit boundary. It prevents Payment API from also owning the contract and failure behavior of Payment Orchestrator.	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 “Webhook arrives before synchronous response” to Payment API.	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 Idempotency Store boundary before extracting or scaling it.
Payment Orchestrator	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Conditionally transition intent and append balanced ledger postings in one boundary.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Idempotency Store from also owning the contract and failure behavior of Risk Engine.	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 “Ledger posting is unbalanced” to Idempotency 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 Payment Orchestrator boundary before extracting or scaling it.
Risk Engine	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Verify/dedupe webhooks and reconcile provider settlements against internal attempts/ledger.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Payment Orchestrator from also owning the contract and failure behavior of Token Vault.	Embed policy in every caller. Rules then drift by deployment version and emergency changes require fleet-wide redeployment. Here that simpler choice would couple “Settlement file disagrees with captured amount” to Payment Orchestrator.	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 Risk Engine boundary before extracting or scaling it.
Token Vault	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Authenticate merchant, validate amount/currency, and atomically create intent/idempotency result.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Risk Engine from also owning the contract and failure behavior of Processor Router.	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 “Client times out after successful capture” to Risk Engine.	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 Token Vault boundary before extracting or scaling it.
Processor Router	Executes idempotent asynchronous work under lease and retry policy. In this design it participates in: Score risk and resolve a tokenized funding instrument without exposing raw card data.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Token Vault from also owning the contract and failure behavior of Processor Adapter.	Perform the work before acknowledging the caller. This gives simpler consistency but unbounded tail latency and fragile retries. Here that simpler choice would couple “Processor returns unknown outcome” to Token Vault.	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 Processor Router boundary before extracting or scaling it.
Processor Adapter	Executes idempotent asynchronous work under lease and retry policy. In this design it participates in: Route to a processor and record each uncertain attempt before external I/O.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Processor Router from also owning the contract and failure behavior of Webhook Gateway.	Perform the work before acknowledging the caller. This gives simpler consistency but unbounded tail latency and fragile retries. Here that simpler choice would couple “Webhook arrives before synchronous response” to Processor Router.	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 Processor Adapter boundary before extracting or scaling it.
Webhook Gateway	Owns authentication, validation, quotas, deadlines, and routing. In this design it participates in: Conditionally transition intent and append balanced ledger postings in one boundary.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Processor Adapter from also owning the contract and failure behavior of Payment State DB.	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 “Ledger posting is unbalanced” to Processor Adapter.	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 Webhook Gateway boundary before extracting or scaling it.
Payment State DB	Owns authoritative versioned state or immutable evidence. In this design it participates in: Verify/dedupe webhooks and reconcile provider settlements against internal attempts/ledger.	Defines the source of truth and commit boundary. It prevents Webhook Gateway from also owning the contract and failure behavior of Double-Entry Ledger.	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 “Settlement file disagrees with captured amount” to Webhook Gateway.	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 Payment State DB boundary before extracting or scaling it.
Double-Entry Ledger	Owns authentication, validation, quotas, deadlines, and routing. In this design it participates in: Authenticate merchant, validate amount/currency, and atomically create intent/idempotency result.	Defines the source of truth and commit boundary. It prevents Payment State DB from also owning the contract and failure behavior of Outbox / Kafka.	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 “Client times out after successful capture” to Payment State DB.	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 Double-Entry Ledger boundary before extracting or scaling it.
Outbox / Kafka	Preserves durable ordering, replay, and consumer checkpoints. In this design it participates in: Score risk and resolve a tokenized funding instrument without exposing raw card data.	Decouples bursts and failure from user acknowledgment. It prevents Double-Entry Ledger from also owning the contract and failure behavior of Reconciliation.	Call every downstream consumer synchronously. That couples user latency and availability to analytics, notification, indexing, and audit dependencies. Here that simpler choice would couple “Processor returns unknown outcome” to Double-Entry Ledger.	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 Outbox / Kafka boundary before extracting or scaling it.
Reconciliation	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Route to a processor and record each uncertain attempt before external I/O.	Proves correctness and supports controlled recovery. It prevents Outbox / Kafka from also owning the contract and failure behavior of Refund / Dispute.	Infer correctness from request logs. Logs do not prove durable business state, detect silent drift, or safely repair ambiguous outcomes. Here that simpler choice would couple “Webhook arrives before synchronous response” to Outbox / Kafka.	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 Reconciliation boundary before extracting or scaling it.
Refund / Dispute	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Conditionally transition intent and append balanced ledger postings in one boundary.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Reconciliation from also owning the contract and failure behavior of Audit / Compliance.	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 “Ledger posting is unbalanced” to Reconciliation.	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 Refund / Dispute boundary before extracting or scaling it.
Audit / Compliance	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Verify/dedupe webhooks and reconcile provider settlements against internal attempts/ledger.	Proves correctness and supports controlled recovery. It prevents Refund / Dispute from also owning the contract and failure behavior of operators and repair tooling.	Infer correctness from request logs. Logs do not prove durable business state, detect silent drift, or safely repair ambiguous outcomes. Here that simpler choice would couple “Settlement file disagrees with captured amount” to Refund / Dispute.	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 Audit / Compliance boundary before extracting or scaling it.

Decision 01

Chosen boundary vs simpler monolith

Use an idempotency key for every intent and model payment as a state machine, not one synchronous call.

A single deployable with one transactional store is the right starting point for a small payment processing workload. The proposed boundary becomes justified when use an idempotency key for every intent and model payment as a state machine, not one synchronous call. Separating it contains the named failure “Client times out after successful capture” without placing unrelated work on the critical path.

Decision 02

Correctness vs availability

Record every balance movement in an immutable double-entry ledger; derived balances can be rebuilt.

The design coordinates only where no scarce resource or monetary effect is created twice, lost after acknowledgment, or committed without an audit trail. 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

Treat provider webhooks as untrusted, duplicated, and out of order; verify, dedupe, and reconcile.

This choice adds dashboards, ownership, schema compatibility, capacity planning, and recovery runbooks. Start with the minimum topology that supports “Route to a processor and record each uncertain attempt before external I/O.” 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 no scarce resource or monetary effect is created twice, lost after acknowledgment, or committed without an audit trail. Durable projections, notifications, indexing, analytics, and repair are isolated so their load or failure cannot corrupt the acknowledged operation.

Correctness boundary

Use an idempotency key for every intent and model payment as a state machine, not one synchronous call.

Scale boundary

Record every balance movement in an immutable double-entry ledger; derived balances can be rebuilt.

Operational boundary

Treat provider webhooks as untrusted, duplicated, and out of order; verify, dedupe, and reconcile.

↑ top

5. Critical flow and algorithm

Authenticate merchant, validate amount/currency, and atomically create intent/idempotency result.

Score risk and resolve a tokenized funding instrument without exposing raw card data.

Route to a processor and record each uncertain attempt before external I/O.

Conditionally transition intent and append balanced ledger postings in one boundary.

Verify/dedupe webhooks and reconcile provider settlements against internal attempts/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

no scarce resource or monetary effect is created twice, lost after acknowledgment, or committed without an audit trail. 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
Client times out after successful capture	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.
Processor returns unknown outcome	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.
Webhook arrives before synchronous response	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.
Ledger posting is unbalanced	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.
Settlement file disagrees with captured amount	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: Client times out after successful capture
operation_id: op_01J...
aggregate_key: payment-system_123
last_durable_version: 18
projection_checkpoint: partition-07@offset-482901
retry_policy: same_identity / bounded_backoff
reconciliation_status: pending
owner: Audit / Compliance

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: Client times out after successful capture — 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

What if the client times out after capture?

Use an idempotency key for every intent and model payment as a state machine, not one synchronous call. Walk the concrete lifecycle step “Authenticate merchant, validate amount/currency, and atomically create intent/idempotency result.” Then test it against “Client times out after successful capture”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How do refunds work?

Record every balance movement in an immutable double-entry ledger; derived balances can be rebuilt. Walk the concrete lifecycle step “Score risk and resolve a tokenized funding instrument without exposing raw card data.” Then test it against “Processor returns unknown outcome”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How do you reconcile with processors?

Treat provider webhooks as untrusted, duplicated, and out of order; verify, dedupe, and reconcile. Walk the concrete lifecycle step “Route to a processor and record each uncertain attempt before external I/O.” Then test it against “Webhook arrives before synchronous response”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How is PCI scope reduced?

Use an idempotency key for every intent and model payment as a state machine, not one synchronous call. Walk the concrete lifecycle step “Conditionally transition intent and append balanced ledger postings in one boundary.” Then test it against “Ledger posting is unbalanced”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

What does exactly-once mean for money?

Record every balance movement in an immutable double-entry ledger; derived balances can be rebuilt. Walk the concrete lifecycle step “Verify/dedupe webhooks and reconcile provider settlements against internal attempts/ledger.” Then test it against “Settlement file disagrees with captured amount”. 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	no scarce resource or monetary effect is created twice, lost after acknowledgment, or committed without an audit trail.
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 no scarce resource or monetary effect is created twice, lost after acknowledgment, or committed without an audit trail. The online path stays bounded, authoritative state is versioned and partitioned, and asynchronous work is replayable and observable.”

↑ top

Checkout / Merchant

Payment API

Idempotency Store

Payment Orchestrator

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