Distributed Message Queue — 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 distributed message queue 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 10M messages/s · petabytes retained · p99 < 20 ms.
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: acknowledged writes survive the configured failure model and replicas converge.

↑ top

2. Capacity assumptions

State assumptions before choosing technology. The supplied workload is 10M messages/s · petabytes retained · p99 < 20 ms. Convert it into peak throughput, retained records, bandwidth, and failure budget; then connect each number to a component.

Dimension	Working estimate	Design implication
Traffic	10M messages/s · petabytes retained · p99 < 20 ms	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

PRODUCE(topic, key, batch, producer_epoch, sequence)

FETCH(topic, partition, offset, max_bytes)

COMMIT(group, partition, offset)

ADMIN create/expand topicversioned metadata

Authoritative records and indexes

record(topic, partition, offset, timestamp, key, value, producer_seq)

partition(topic, id, leader, replicas, epoch, high_watermark)

group_offset(group, topic, partition, offset, generation)

segment(base_offset, max_offset, checksum, tier_uri)

Command and event envelopes

# Command envelope
operation: PRODUCE(topic, key, batch, producer_epoch, sequence)
request_id: req_01J...
idempotency_key: idem_7f2...
aggregate_key: message-queue_123
expected_version: 17
payload: {...}

# Durable event
event_id: evt_01J...
aggregate_key: message-queue_123
event_type: message_queue_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
Producer	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Validate schema and choose partition by stable key or explicit partition.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents the external caller from also owning the contract and failure behavior of Schema Registry.	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 “Leader fails after local append before quorum” 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 Producer boundary before extracting or scaling it.
Schema Registry	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Leader dedupes producer epoch/sequence, appends batch, and replicates to followers.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Producer from also owning the contract and failure behavior of Partitioner.	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 “Consumer rebalance overlaps processing” to Producer.	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 Schema Registry boundary before extracting or scaling it.
Partitioner	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Advance high watermark after in-sync quorum and acknowledge according to durability policy.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Schema Registry from also owning the contract and failure behavior of Broker Front Door.	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 “Slow replica falls out of ISR” to Schema Registry.	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 Partitioner boundary before extracting or scaling it.
Broker Front Door	Owns authentication, validation, quotas, deadlines, and routing. In this design it participates in: Consumers fetch ordered batches and commit offsets under a group generation.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Partitioner from also owning the contract and failure behavior of Partition Leader.	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 “Producer epoch is reused after retry” to Partitioner.	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 Broker Front Door boundary before extracting or scaling it.
Partition Leader	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Retention/compaction closes segments and optionally moves immutable data to object storage.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Broker Front Door from also owning the contract and failure behavior of Follower Replicas.	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 “Compaction removes required tombstone” to Broker Front Door.	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 Partition Leader boundary before extracting or scaling it.
Follower Replicas	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Validate schema and choose partition by stable key or explicit partition.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Partition Leader from also owning the contract and failure behavior of Commit Quorum.	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 “Leader fails after local append before quorum” to Partition Leader.	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 Follower Replicas boundary before extracting or scaling it.
Commit Quorum	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Leader dedupes producer epoch/sequence, appends batch, and replicates to followers.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Follower Replicas from also owning the contract and failure behavior of Segment Log.	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 “Consumer rebalance overlaps processing” to Follower Replicas.	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 Commit Quorum boundary before extracting or scaling it.
Segment Log	Preserves durable ordering, replay, and consumer checkpoints. In this design it participates in: Advance high watermark after in-sync quorum and acknowledge according to durability policy.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Commit Quorum from also owning the contract and failure behavior of Index / Page Cache.	Call every downstream consumer synchronously. That couples user latency and availability to analytics, notification, indexing, and audit dependencies. Here that simpler choice would couple “Slow replica falls out of ISR” to Commit Quorum.	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 Segment Log boundary before extracting or scaling it.
Index / Page Cache	Serves a low-latency, disposable projection with authoritative fallback. In this design it participates in: Consumers fetch ordered batches and commit offsets under a group generation.	Removes repeated work from the latency-critical path. It prevents Segment Log from also owning the contract and failure behavior of Controller Quorum.	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 “Producer epoch is reused after retry” to Segment Log.	Costs invalidation, staleness, stampede protection, and memory. Add or split it only after the workload (10M messages/s · petabytes retained · p99 < 20 ms) proves repeated reads dominate. Measure the latency, saturation, lag, or mismatch at the Index / Page Cache boundary before extracting or scaling it.
Controller Quorum	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Retention/compaction closes segments and optionally moves immutable data to object storage.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Index / Page Cache from also owning the contract and failure behavior of Consumer Group Coordinator.	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 “Compaction removes required tombstone” to Index / Page Cache.	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 Controller Quorum boundary before extracting or scaling it.
Consumer Group Coordinator	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Validate schema and choose partition by stable key or explicit partition.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Controller Quorum from also owning the contract and failure behavior of Consumer.	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 “Leader fails after local append before quorum” to Controller Quorum.	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 Consumer Group Coordinator boundary before extracting or scaling it.
Consumer	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Leader dedupes producer epoch/sequence, appends batch, and replicates to followers.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Consumer Group Coordinator from also owning the contract and failure behavior of Offset 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 “Consumer rebalance overlaps processing” to Consumer Group Coordinator.	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 Consumer boundary before extracting or scaling it.
Offset Store	Owns authoritative versioned state or immutable evidence. In this design it participates in: Advance high watermark after in-sync quorum and acknowledge according to durability policy.	Defines the source of truth and commit boundary. It prevents Consumer from also owning the contract and failure behavior of Tiered Object Store.	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 “Slow replica falls out of ISR” to Consumer.	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 Offset Store boundary before extracting or scaling it.
Tiered Object Store	Owns authoritative versioned state or immutable evidence. In this design it participates in: Consumers fetch ordered batches and commit offsets under a group generation.	Defines the source of truth and commit boundary. It prevents Offset Store from also owning the contract and failure behavior of Compaction / Retention.	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 “Producer epoch is reused after retry” to Offset Store.	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 Tiered Object Store boundary before extracting or scaling it.
Compaction / Retention	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Retention/compaction closes segments and optionally moves immutable data to object storage.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Tiered Object 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 “Compaction removes required tombstone” to Tiered Object 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 Compaction / Retention boundary before extracting or scaling it.

Decision 01

Chosen boundary vs simpler monolith

Partition by a stable key to preserve local ordering; no system offers useful global ordering at this scale.

A single deployable with one transactional store is the right starting point for a small distributed message queue workload. The proposed boundary becomes justified when partition by a stable key to preserve local ordering; no system offers useful global ordering at this scale. Separating it contains the named failure “Leader fails after local append before quorum” without placing unrelated work on the critical path.

Decision 02

Correctness vs availability

Append sequentially, batch network/disk writes, and replicate before acknowledging according to durability needs.

The design coordinates only where acknowledged writes survive the configured failure model and replicas converge. 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

Track consumer offsets separately from retention so consumers can replay without blocking deletion policy.

This choice adds dashboards, ownership, schema compatibility, capacity planning, and recovery runbooks. Start with the minimum topology that supports “Advance high watermark after in-sync quorum and acknowledge according to durability policy.” 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 acknowledged writes survive the configured failure model and replicas converge. Durable projections, notifications, indexing, analytics, and repair are isolated so their load or failure cannot corrupt the acknowledged operation.

Correctness boundary

Partition by a stable key to preserve local ordering; no system offers useful global ordering at this scale.

Scale boundary

Append sequentially, batch network/disk writes, and replicate before acknowledging according to durability needs.

Operational boundary

Track consumer offsets separately from retention so consumers can replay without blocking deletion policy.

↑ top

5. Critical flow and algorithm

Validate schema and choose partition by stable key or explicit partition.

Leader dedupes producer epoch/sequence, appends batch, and replicates to followers.

Advance high watermark after in-sync quorum and acknowledge according to durability policy.

Consumers fetch ordered batches and commit offsets under a group generation.

Retention/compaction closes segments and optionally moves immutable data to object storage.

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

acknowledged writes survive the configured failure model and replicas converge. 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
Leader fails after local append before quorum	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.
Consumer rebalance overlaps processing	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.
Slow replica falls out of ISR	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.
Producer epoch is reused after retry	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.
Compaction removes required tombstone	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: Leader fails after local append before quorum
operation_id: op_01J...
aggregate_key: message-queue_123
last_durable_version: 18
projection_checkpoint: partition-07@offset-482901
retry_policy: same_identity / bounded_backoff
reconciliation_status: pending
owner: Compaction / Retention

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: Leader fails after local append before quorum — 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 does leader election work?

Partition by a stable key to preserve local ordering; no system offers useful global ordering at this scale. Walk the concrete lifecycle step “Validate schema and choose partition by stable key or explicit partition.” Then test it against “Leader fails after local append before quorum”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How do consumer groups rebalance?

Append sequentially, batch network/disk writes, and replicate before acknowledging according to durability needs. Walk the concrete lifecycle step “Leader dedupes producer epoch/sequence, appends batch, and replicates to followers.” Then test it against “Consumer rebalance overlaps processing”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How do transactions work?

Track consumer offsets separately from retention so consumers can replay without blocking deletion policy. Walk the concrete lifecycle step “Advance high watermark after in-sync quorum and acknowledge according to durability policy.” Then test it against “Slow replica falls out of ISR”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How do you avoid duplicate processing?

Partition by a stable key to preserve local ordering; no system offers useful global ordering at this scale. Walk the concrete lifecycle step “Consumers fetch ordered batches and commit offsets under a group generation.” Then test it against “Producer epoch is reused after retry”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How does tiered storage change brokers?

Append sequentially, batch network/disk writes, and replicate before acknowledging according to durability needs. Walk the concrete lifecycle step “Retention/compaction closes segments and optionally moves immutable data to object storage.” Then test it against “Compaction removes required tombstone”. 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	acknowledged writes survive the configured failure model and replicas converge.
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 acknowledged writes survive the configured failure model and replicas converge. The online path stays bounded, authoritative state is versioned and partitioned, and asynchronous work is replayable and observable.”

↑ top

Producer

Schema Registry

Partitioner

Broker Front Door

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