Workflow Orchestrator — 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 workflow orchestrator 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 100M active workflows · years-long history · deterministic replay.
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 workflow orchestrator 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 100M active workflows · years-long history · deterministic replay. Convert it into peak throughput, retained records, bandwidth, and failure budget; then connect each number to a component.

Dimension	Working estimate	Design implication
Traffic	100M active workflows · years-long history · deterministic replay	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/workflows/{type}/start

POST /v1/workflows/{id}/signals

POST /v1/workflows/{id}/cancel

GET /v1/workflows/{id}/history

Authoritative records and indexes

workflow(id PK, type, state, current_run, version)

history(workflow_id, event_seq, type, payload)

task(queue, task_id, lease, attempt)

timer(workflow_id, fire_at, event_id, state)

Command and event envelopes

# Command envelope
operation: POST /v1/workflows/{type}/start
request_id: req_01J...
idempotency_key: idem_7f2...
aggregate_key: workflow-orchestrator_123
expected_version: 17
payload: {...}

# Durable event
event_id: evt_01J...
aggregate_key: workflow-orchestrator_123
event_type: workflow_orchestrator_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
Workflow Client	Owns client protocol, request identity, retry, and resume behavior. In this design it participates in: Persist workflow-start event and enqueue one deterministic decision task.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents the external caller from also owning the contract and failure behavior of Workflow 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 “Worker code becomes nondeterministic” 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 Workflow Client boundary before extracting or scaling it.
Workflow API	Owns authentication, validation, quotas, deadlines, and routing. In this design it participates in: Executor replays history, runs pure workflow code, and appends commands as events.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Workflow Client from also owning the contract and failure behavior of Namespace / Quota.	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 “Activity succeeds after timeout” to Workflow Client.	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 Workflow API boundary before extracting or scaling it.
Namespace / Quota	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Dispatch activities with stable IDs, retries, heartbeats, and independent queues.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Workflow API from also owning the contract and failure behavior of History 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 “Timer fires twice” to Workflow 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 Namespace / Quota boundary before extracting or scaling it.
History Store	Owns authoritative versioned state or immutable evidence. In this design it participates in: Timers/signals append durable events and wake a new workflow decision.	Defines the source of truth and commit boundary. It prevents Namespace / Quota from also owning the contract and failure behavior of Workflow Task Queue.	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 “History grows beyond replay budget” to Namespace / Quota.	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 History Store boundary before extracting or scaling it.
Workflow Task Queue	Preserves durable ordering, replay, and consumer checkpoints. In this design it participates in: Compact/archive closed histories and support versioned code routing or operator reset.	Decouples bursts and failure from user acknowledgment. It prevents History Store from also owning the contract and failure behavior of Deterministic Executor.	Call every downstream consumer synchronously. That couples user latency and availability to analytics, notification, indexing, and audit dependencies. Here that simpler choice would couple “Deployment removes old workflow version” to History Store.	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 Workflow Task Queue boundary before extracting or scaling it.
Deterministic Executor	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Persist workflow-start event and enqueue one deterministic decision task.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Workflow Task Queue from also owning the contract and failure behavior of Activity Queues.	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 “Worker code becomes nondeterministic” to Workflow Task Queue.	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 Deterministic Executor boundary before extracting or scaling it.
Activity Queues	Preserves durable ordering, replay, and consumer checkpoints. In this design it participates in: Executor replays history, runs pure workflow code, and appends commands as events.	Decouples bursts and failure from user acknowledgment. It prevents Deterministic Executor from also owning the contract and failure behavior of Activity Workers.	Call every downstream consumer synchronously. That couples user latency and availability to analytics, notification, indexing, and audit dependencies. Here that simpler choice would couple “Activity succeeds after timeout” to Deterministic Executor.	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 Activity Queues boundary before extracting or scaling it.
Activity Workers	Executes idempotent asynchronous work under lease and retry policy. In this design it participates in: Dispatch activities with stable IDs, retries, heartbeats, and independent queues.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Activity Queues from also owning the contract and failure behavior of Timer Wheel.	Perform the work before acknowledging the caller. This gives simpler consistency but unbounded tail latency and fragile retries. Here that simpler choice would couple “Timer fires twice” to Activity Queues.	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 Activity Workers boundary before extracting or scaling it.
Timer Wheel	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Timers/signals append durable events and wake a new workflow decision.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Activity Workers from also owning the contract and failure behavior of Signal 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 “History grows beyond replay budget” to Activity Workers.	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 Timer Wheel boundary before extracting or scaling it.
Signal Ingest	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Compact/archive closed histories and support versioned code routing or operator reset.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Timer Wheel from also owning the contract and failure behavior of Visibility Index.	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 “Deployment removes old workflow version” to Timer Wheel.	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 Signal Ingest boundary before extracting or scaling it.
Visibility Index	Serves a low-latency, disposable projection with authoritative fallback. In this design it participates in: Persist workflow-start event and enqueue one deterministic decision task.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Signal Ingest from also owning the contract and failure behavior of Retry / DLQ.	Read authoritative storage for every request. Simpler initially, but latency and source-of-truth load scale together and a hot key becomes a database incident. Here that simpler choice would couple “Worker code becomes nondeterministic” to Signal Ingest.	Costs invalidation, staleness, stampede protection, and memory. Add or split it only after the workload (100M active workflows · years-long history · deterministic replay) proves repeated reads dominate. Measure the latency, saturation, lag, or mismatch at the Visibility Index boundary before extracting or scaling it.
Retry / DLQ	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Executor replays history, runs pure workflow code, and appends commands as events.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Visibility Index from also owning the contract and failure behavior of Versioning 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 “Activity succeeds after timeout” to Visibility Index.	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 Retry / DLQ boundary before extracting or scaling it.
Versioning Router	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Dispatch activities with stable IDs, retries, heartbeats, and independent queues.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Retry / DLQ from also owning the contract and failure behavior of Archival 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 “Timer fires twice” to Retry / DLQ.	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 Versioning Router boundary before extracting or scaling it.
Archival Store	Owns authoritative versioned state or immutable evidence. In this design it participates in: Timers/signals append durable events and wake a new workflow decision.	Defines the source of truth and commit boundary. It prevents Versioning Router from also owning the contract and failure behavior of Operations / Reset.	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 “History grows beyond replay budget” to Versioning Router.	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 Archival Store boundary before extracting or scaling it.
Operations / Reset	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Compact/archive closed histories and support versioned code routing or operator reset.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Archival Store 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 “Deployment removes old workflow version” to Archival 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 Operations / Reset boundary before extracting or scaling it.

Decision 01

Chosen boundary vs simpler monolith

Persist workflow history, not in-memory stack state; deterministic replay reconstructs decisions after any worker failure.

A single deployable with one transactional store is the right starting point for a small workflow orchestrator workload. The proposed boundary becomes justified when persist workflow history, not in-memory stack state; deterministic replay reconstructs decisions after any worker failure. Separating it contains the named failure “Worker code becomes nondeterministic” without placing unrelated work on the critical path.

Decision 02

Correctness vs availability

Activities perform external side effects and must be idempotent; workflow code only schedules and records decisions.

The design coordinates only where each accepted workflow orchestrator 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

Timers and signals are durable events, which costs history growth but avoids keeping processes or database locks alive for days.

This choice adds dashboards, ownership, schema compatibility, capacity planning, and recovery runbooks. Start with the minimum topology that supports “Dispatch activities with stable IDs, retries, heartbeats, and independent queues.” 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 workflow orchestrator 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

Persist workflow history, not in-memory stack state; deterministic replay reconstructs decisions after any worker failure.

Scale boundary

Activities perform external side effects and must be idempotent; workflow code only schedules and records decisions.

Operational boundary

Timers and signals are durable events, which costs history growth but avoids keeping processes or database locks alive for days.

↑ top

5. Critical flow and algorithm

Persist workflow-start event and enqueue one deterministic decision task.

Executor replays history, runs pure workflow code, and appends commands as events.

Dispatch activities with stable IDs, retries, heartbeats, and independent queues.

Timers/signals append durable events and wake a new workflow decision.

Compact/archive closed histories and support versioned code routing or operator reset.

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

Cache immutable or slowly changing reads with versioned keys, TTL jitter, short negative entries, and request coalescing.

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 workflow orchestrator 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
Worker code becomes nondeterministic	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.
Activity succeeds after 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.
Timer fires twice	Resolve at source of truth with uniqueness, expected version, lease generation, or monotonic cursor.	Reconciliation returns zero invariant gaps; lag/replica health and user-visible SLO return to baseline.
History grows beyond replay budget	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.
Deployment removes old workflow version	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: Worker code becomes nondeterministic
operation_id: op_01J...
aggregate_key: workflow-orchestrator_123
last_durable_version: 18
projection_checkpoint: partition-07@offset-482901
retry_policy: same_identity / bounded_backoff
reconciliation_status: pending
owner: Operations / Reset

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: Worker code becomes nondeterministic — 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 makes workflow code deterministic?

Persist workflow history, not in-memory stack state; deterministic replay reconstructs decisions after any worker failure. Walk the concrete lifecycle step “Persist workflow-start event and enqueue one deterministic decision task.” Then test it against “Worker code becomes nondeterministic”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How are activities retried?

Activities perform external side effects and must be idempotent; workflow code only schedules and records decisions. Walk the concrete lifecycle step “Executor replays history, runs pure workflow code, and appends commands as events.” Then test it against “Activity succeeds after timeout”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How do you version workflow definitions?

Timers and signals are durable events, which costs history growth but avoids keeping processes or database locks alive for days. Walk the concrete lifecycle step “Dispatch activities with stable IDs, retries, heartbeats, and independent queues.” Then test it against “Timer fires twice”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How do timers scale?

Persist workflow history, not in-memory stack state; deterministic replay reconstructs decisions after any worker failure. Walk the concrete lifecycle step “Timers/signals append durable events and wake a new workflow decision.” Then test it against “History grows beyond replay budget”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How do child workflows work?

Activities perform external side effects and must be idempotent; workflow code only schedules and records decisions. Walk the concrete lifecycle step “Compact/archive closed histories and support versioned code routing or operator reset.” Then test it against “Deployment removes old workflow version”. 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 workflow orchestrator 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 workflow orchestrator 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

Workflow Client

Workflow API

Namespace / Quota

History Store

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