Distributed SQL Database — 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 sql database 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 PB-scale · millions TPS · serializable option.
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 PB-scale · millions TPS · serializable option. Convert it into peak throughput, retained records, bandwidth, and failure budget; then connect each number to a component.

Dimension	Working estimate	Design implication
Traffic	PB-scale · millions TPS · serializable option	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

SQL BEGIN/COMMIT transaction

SELECT ... AS OF SYSTEM TIME

ALTER TABLE ... online

WATCH table/range changefeed

Authoritative records and indexes

range(range_id, start_key, end_key, replicas, epoch)

mvcc_cell(key, timestamp, value, tombstone)

transaction(id, read_ts, write_ts, state, participants)

schema(version, descriptor, effective_ts)

Command and event envelopes

# Command envelope
operation: SQL BEGIN/COMMIT transaction
request_id: req_01J...
idempotency_key: idem_7f2...
aggregate_key: distributed-sql_123
expected_version: 17
payload: {...}

# Durable event
event_id: evt_01J...
aggregate_key: distributed-sql_123
event_type: distributed_sql_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
SQL Client	Owns client protocol, request identity, retry, and resume behavior. In this design it participates in: Parse/optimize SQL and route key spans using versioned range descriptors.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents the external caller from also owning the contract and failure behavior of SQL Gateway.	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 “Coordinator dies during two-phase commit” 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 SQL Client boundary before extracting or scaling it.
SQL Gateway	Owns authentication, validation, quotas, deadlines, and routing. In this design it participates in: Choose read timestamp and execute local or distributed reads from MVCC snapshots.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents SQL Client from also owning the contract and failure behavior of Parser / Optimizer.	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 “Hot sequential key overloads one range” to SQL 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 SQL Gateway boundary before extracting or scaling it.
Parser / Optimizer	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: For writes acquire intents/locks, replicate each range command by consensus, and coordinate commit.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents SQL Gateway from also owning the contract and failure behavior of Schema Catalog.	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 “Long read blocks MVCC garbage collection” to SQL Gateway.	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 Parser / Optimizer boundary before extracting or scaling it.
Schema Catalog	Preserves durable ordering, replay, and consumer checkpoints. In this design it participates in: Resolve abandoned intents and publish changefeed after commit timestamp is durable.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Parser / Optimizer from also owning the contract and failure behavior of Range Router.	Call every downstream consumer synchronously. That couples user latency and availability to analytics, notification, indexing, and audit dependencies. Here that simpler choice would couple “Leaseholder moves during transaction” to Parser / Optimizer.	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 Schema Catalog boundary before extracting or scaling it.
Range Router	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Split/merge hot ranges, rebalance replicas, compact MVCC, and run online schema migrations.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Schema Catalog from also owning the contract and failure behavior of Transaction 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 “Schema version changes mid-query” to Schema Catalog.	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 Range Router boundary before extracting or scaling it.
Transaction Coordinator	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Parse/optimize SQL and route key spans using versioned range descriptors.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Range Router from also owning the contract and failure behavior of Timestamp Oracle.	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 “Coordinator dies during two-phase commit” to Range Router.	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 Transaction Coordinator boundary before extracting or scaling it.
Timestamp Oracle	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Choose read timestamp and execute local or distributed reads from MVCC snapshots.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Transaction Coordinator from also owning the contract and failure behavior of Range Leaseholder.	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 “Hot sequential key overloads one range” to Transaction 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 Timestamp Oracle boundary before extracting or scaling it.
Range Leaseholder	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: For writes acquire intents/locks, replicate each range command by consensus, and coordinate commit.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Timestamp Oracle from also owning the contract and failure behavior of Consensus 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 “Long read blocks MVCC garbage collection” to Timestamp Oracle.	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 Range Leaseholder boundary before extracting or scaling it.
Consensus Replicas	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Resolve abandoned intents and publish changefeed after commit timestamp is durable.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Range Leaseholder from also owning the contract and failure behavior of MVCC 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 “Leaseholder moves during transaction” to Range Leaseholder.	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 Consensus Replicas boundary before extracting or scaling it.
MVCC Engine	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Split/merge hot ranges, rebalance replicas, compact MVCC, and run online schema migrations.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Consensus Replicas from also owning the contract and failure behavior of Distributed Lock Table.	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 “Schema version changes mid-query” to Consensus 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 MVCC Engine boundary before extracting or scaling it.
Distributed Lock Table	Owns authoritative versioned state or immutable evidence. In this design it participates in: Parse/optimize SQL and route key spans using versioned range descriptors.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents MVCC Engine from also owning the contract and failure behavior of Changefeed 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 “Coordinator dies during two-phase commit” to MVCC 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 Distributed Lock Table boundary before extracting or scaling it.
Changefeed Log	Preserves durable ordering, replay, and consumer checkpoints. In this design it participates in: Choose read timestamp and execute local or distributed reads from MVCC snapshots.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Distributed Lock Table from also owning the contract and failure behavior of Range Split / Merge.	Call every downstream consumer synchronously. That couples user latency and availability to analytics, notification, indexing, and audit dependencies. Here that simpler choice would couple “Hot sequential key overloads one range” to Distributed Lock Table.	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 Changefeed Log boundary before extracting or scaling it.
Range Split / Merge	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: For writes acquire intents/locks, replicate each range command by consensus, and coordinate commit.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Changefeed Log from also owning the contract and failure behavior of Backup / Restore.	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 “Long read blocks MVCC garbage collection” to Changefeed Log.	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 Range Split / Merge boundary before extracting or scaling it.
Backup / Restore	Owns authoritative versioned state or immutable evidence. In this design it participates in: Resolve abandoned intents and publish changefeed after commit timestamp is durable.	Defines the source of truth and commit boundary. It prevents Range Split / Merge from also owning the contract and failure behavior of Observability.	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 “Leaseholder moves during transaction” to Range Split / Merge.	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 Backup / Restore boundary before extracting or scaling it.
Observability	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Split/merge hot ranges, rebalance replicas, compact MVCC, and run online schema migrations.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Backup / Restore 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 “Schema version changes mid-query” to Backup / Restore.	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 Observability boundary before extracting or scaling it.

Decision 01

Chosen boundary vs simpler monolith

Range partitions preserve ordered scans while splitting hot ranges; hash sharding is simpler for point workloads but destroys locality.

A single deployable with one transactional store is the right starting point for a small distributed sql database workload. The proposed boundary becomes justified when range partitions preserve ordered scans while splitting hot ranges; hash sharding is simpler for point workloads but destroys locality. Separating it contains the named failure “Coordinator dies during two-phase commit” without placing unrelated work on the critical path.

Decision 02

Correctness vs availability

Consensus replicates each range; distributed transactions coordinate only the ranges touched and pay latency proportional to geography.

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

MVCC enables consistent snapshots and online reads, but garbage collection must honor active transactions, backups, and change feeds.

This choice adds dashboards, ownership, schema compatibility, capacity planning, and recovery runbooks. Start with the minimum topology that supports “For writes acquire intents/locks, replicate each range command by consensus, and coordinate commit.” 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

Range partitions preserve ordered scans while splitting hot ranges; hash sharding is simpler for point workloads but destroys locality.

Scale boundary

Consensus replicates each range; distributed transactions coordinate only the ranges touched and pay latency proportional to geography.

Operational boundary

MVCC enables consistent snapshots and online reads, but garbage collection must honor active transactions, backups, and change feeds.

↑ top

5. Critical flow and algorithm

Parse/optimize SQL and route key spans using versioned range descriptors.

Choose read timestamp and execute local or distributed reads from MVCC snapshots.

For writes acquire intents/locks, replicate each range command by consensus, and coordinate commit.

Resolve abandoned intents and publish changefeed after commit timestamp is durable.

Split/merge hot ranges, rebalance replicas, compact MVCC, and run online schema migrations.

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

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
Coordinator dies during two-phase commit	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.
Hot sequential key overloads one range	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.
Long read blocks MVCC garbage collection	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.
Leaseholder moves during transaction	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.
Schema version changes mid-query	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: Coordinator dies during two-phase commit
operation_id: op_01J...
aggregate_key: distributed-sql_123
last_durable_version: 18
projection_checkpoint: partition-07@offset-482901
retry_policy: same_identity / bounded_backoff
reconciliation_status: pending
owner: Observability

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: Coordinator dies during two-phase commit — 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 two-phase commit recover?

Range partitions preserve ordered scans while splitting hot ranges; hash sharding is simpler for point workloads but destroys locality. Walk the concrete lifecycle step “Parse/optimize SQL and route key spans using versioned range descriptors.” Then test it against “Coordinator dies during two-phase commit”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How are ranges split?

Consensus replicates each range; distributed transactions coordinate only the ranges touched and pay latency proportional to geography. Walk the concrete lifecycle step “Choose read timestamp and execute local or distributed reads from MVCC snapshots.” Then test it against “Hot sequential key overloads one range”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How is serializable isolation implemented?

MVCC enables consistent snapshots and online reads, but garbage collection must honor active transactions, backups, and change feeds. Walk the concrete lifecycle step “For writes acquire intents/locks, replicate each range command by consensus, and coordinate commit.” Then test it against “Long read blocks MVCC garbage collection”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

What happens to long transactions?

Range partitions preserve ordered scans while splitting hot ranges; hash sharding is simpler for point workloads but destroys locality. Walk the concrete lifecycle step “Resolve abandoned intents and publish changefeed after commit timestamp is durable.” Then test it against “Leaseholder moves during transaction”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How do schema changes run online?

Consensus replicates each range; distributed transactions coordinate only the ranges touched and pay latency proportional to geography. Walk the concrete lifecycle step “Split/merge hot ranges, rebalance replicas, compact MVCC, and run online schema migrations.” Then test it against “Schema version changes mid-query”. 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

SQL Client

SQL Gateway

Parser / Optimizer

Schema Catalog

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