File Storage & Sharing — 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 file storage & sharing 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 10 PB stored · 50 GB objects · 100K concurrent uploads.
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: a published version is complete, checksummed, authorized, and never silently corrupted.

↑ top

2. Capacity assumptions

State assumptions before choosing technology. The supplied workload is 10 PB stored · 50 GB objects · 100K concurrent uploads. Convert it into peak throughput, retained records, bandwidth, and failure budget; then connect each number to a component.

Dimension	Working estimate	Design implication
Traffic	10 PB stored · 50 GB objects · 100K concurrent uploads	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/uploadscreate resumable session

PUT /v1/uploads/{id}/parts/{n}checksum part

POST /v1/uploads/{id}/commitpublish manifest

GET /v1/changes?cursor=incremental sync

Authoritative records and indexes

file(file_id PK, parent_id, owner_id, current_version, etag)

version(file_id, version, manifest_uri, size, content_hash)

upload(upload_id, expected_parts, received_bitmap, expires_at)

journal(owner_id, sequence, change_type, file_id)

Command and event envelopes

# Command envelope
operation: POST /v1/uploads
request_id: req_01J...
idempotency_key: idem_7f2...
aggregate_key: file-storage_123
expected_version: 17
payload: {...}

# Durable event
event_id: evt_01J...
aggregate_key: file-storage_123
event_type: file_storage_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
Desktop / Web Client	Owns client protocol, request identity, retry, and resume behavior. In this design it participates in: Create an upload session and return signed part URLs.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents the external caller from also owning the contract and failure behavior of Global 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 “Commit arrives before final part replication” 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 Desktop / Web Client boundary before extracting or scaling it.
Global API	Owns authentication, validation, quotas, deadlines, and routing. In this design it participates in: Upload chunks independently; verify length and checksum at the storage edge.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Desktop / Web Client from also owning the contract and failure behavior of Auth / Sharing.	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 “Two devices update the same base version” to Desktop / Web 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 Global API boundary before extracting or scaling it.
Auth / Sharing	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Build an immutable ordered manifest only when all required parts exist.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Global API from also owning the contract and failure behavior of Metadata Service.	Embed policy in every caller. Rules then drift by deployment version and emergency changes require fleet-wide redeployment. Here that simpler choice would couple “Shared-link permission is revoked while cached” to Global 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 Auth / Sharing boundary before extracting or scaling it.
Metadata Service	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Atomically advance current_version and append one change-journal entry.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Auth / Sharing from also owning the contract and failure behavior of Upload 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 “Chunk dedupe leaks cross-tenant existence” to Auth / Sharing.	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 Metadata Service boundary before extracting or scaling it.
Upload Coordinator	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Sync clients resume from a monotonic cursor and resolve divergent edits as versions.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Metadata Service from also owning the contract and failure behavior of Chunk Hash Service.	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 “Journal cursor points to compacted history” to Metadata Service.	Costs an interface, deployment, telemetry, ownership, and failure mode. Keep it in-process first; extract when independent scale or reliability pays for the boundary. Measure the latency, saturation, lag, or mismatch at the Upload Coordinator boundary before extracting or scaling it.
Chunk Hash Service	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Create an upload session and return signed part URLs.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Upload Coordinator from also owning the contract and failure behavior of Multipart Session 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 “Commit arrives before final part replication” to Upload 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 Chunk Hash Service boundary before extracting or scaling it.
Multipart Session Store	Owns authoritative versioned state or immutable evidence. In this design it participates in: Upload chunks independently; verify length and checksum at the storage edge.	Defines the source of truth and commit boundary. It prevents Chunk Hash Service from also owning the contract and failure behavior of Object Blob 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 “Two devices update the same base version” to Chunk Hash Service.	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 Multipart Session Store boundary before extracting or scaling it.
Object Blob Store	Owns authoritative versioned state or immutable evidence. In this design it participates in: Build an immutable ordered manifest only when all required parts exist.	Defines the source of truth and commit boundary. It prevents Multipart Session Store from also owning the contract and failure behavior of Version Manifest DB.	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 “Shared-link permission is revoked while cached” to Multipart Session 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 Object Blob Store boundary before extracting or scaling it.
Version Manifest DB	Owns authoritative versioned state or immutable evidence. In this design it participates in: Atomically advance current_version and append one change-journal entry.	Defines the source of truth and commit boundary. It prevents Object Blob Store from also owning the contract and failure behavior of Change Journal.	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 “Chunk dedupe leaks cross-tenant existence” to Object Blob 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 Version Manifest DB boundary before extracting or scaling it.
Change Journal	Preserves durable ordering, replay, and consumer checkpoints. In this design it participates in: Sync clients resume from a monotonic cursor and resolve divergent edits as versions.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Version Manifest DB from also owning the contract and failure behavior of Sync Cursor Service.	Call every downstream consumer synchronously. That couples user latency and availability to analytics, notification, indexing, and audit dependencies. Here that simpler choice would couple “Journal cursor points to compacted history” to Version Manifest DB.	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 Change Journal boundary before extracting or scaling it.
Sync Cursor Service	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Create an upload session and return signed part URLs.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Change Journal from also owning the contract and failure behavior of Virus Scanner.	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 “Commit arrives before final part replication” to Change Journal.	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 Sync Cursor Service boundary before extracting or scaling it.
Virus Scanner	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Upload chunks independently; verify length and checksum at the storage edge.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Sync Cursor Service from also owning the contract and failure behavior of Thumbnail Workers.	Embed policy in every caller. Rules then drift by deployment version and emergency changes require fleet-wide redeployment. Here that simpler choice would couple “Two devices update the same base version” to Sync Cursor Service.	Costs an interface, deployment, telemetry, ownership, and failure mode. Keep it in-process first; extract when independent scale or reliability pays for the boundary. Measure the latency, saturation, lag, or mismatch at the Virus Scanner boundary before extracting or scaling it.
Thumbnail Workers	Executes idempotent asynchronous work under lease and retry policy. In this design it participates in: Build an immutable ordered manifest only when all required parts exist.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Virus Scanner from also owning the contract and failure behavior of CDN / Download.	Perform the work before acknowledging the caller. This gives simpler consistency but unbounded tail latency and fragile retries. Here that simpler choice would couple “Shared-link permission is revoked while cached” to Virus Scanner.	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 Thumbnail Workers boundary before extracting or scaling it.
CDN / Download	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Atomically advance current_version and append one change-journal entry.	Removes repeated work from the latency-critical path. It prevents Thumbnail Workers from also owning the contract and failure behavior of operators and repair tooling.	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 “Chunk dedupe leaks cross-tenant existence” to Thumbnail Workers.	Costs invalidation, staleness, stampede protection, and memory. Add or split it only after the workload (10 PB stored · 50 GB objects · 100K concurrent uploads) proves repeated reads dominate. Measure the latency, saturation, lag, or mismatch at the CDN / Download boundary before extracting or scaling it.

Decision 01

Chosen boundary vs simpler monolith

Use multipart uploads with checksums, resumable sessions, and content-addressed chunks.

A single deployable with one transactional store is the right starting point for a small file storage & sharing workload. The proposed boundary becomes justified when use multipart uploads with checksums, resumable sessions, and content-addressed chunks. Separating it contains the named failure “Commit arrives before final part replication” without placing unrelated work on the critical path.

Decision 02

Correctness vs availability

Keep metadata strongly consistent while blobs replicate asynchronously across zones.

The design coordinates only where a published version is complete, checksummed, authorized, and never silently corrupted. 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

Drive sync from a monotonic change journal; model versions and permissions independently.

This choice adds dashboards, ownership, schema compatibility, capacity planning, and recovery runbooks. Start with the minimum topology that supports “Build an immutable ordered manifest only when all required parts exist.” 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 a published version is complete, checksummed, authorized, and never silently corrupted. Durable projections, notifications, indexing, analytics, and repair are isolated so their load or failure cannot corrupt the acknowledged operation.

Correctness boundary

Use multipart uploads with checksums, resumable sessions, and content-addressed chunks.

Scale boundary

Keep metadata strongly consistent while blobs replicate asynchronously across zones.

Operational boundary

Drive sync from a monotonic change journal; model versions and permissions independently.

↑ top

5. Critical flow and algorithm

Create an upload session and return signed part URLs.

Upload chunks independently; verify length and checksum at the storage edge.

Build an immutable ordered manifest only when all required parts exist.

Atomically advance current_version and append one change-journal entry.

Sync clients resume from a monotonic cursor and resolve divergent edits as versions.

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

a published version is complete, checksummed, authorized, and never silently corrupted. 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
Commit arrives before final part replication	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.
Two devices update the same base version	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.
Shared-link permission is revoked while cached	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.
Chunk dedupe leaks cross-tenant existence	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.
Journal cursor points to compacted history	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: Commit arrives before final part replication
operation_id: op_01J...
aggregate_key: file-storage_123
last_durable_version: 18
projection_checkpoint: partition-07@offset-482901
retry_policy: same_identity / bounded_backoff
reconciliation_status: pending
owner: CDN / Download

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: Commit arrives before final part replication — alert before the user-visible invariant is violated.

Regional evolution

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

↑ top

10. Interviewer follow-ups

How do you resume an interrupted upload?

Use multipart uploads with checksums, resumable sessions, and content-addressed chunks. Walk the concrete lifecycle step “Create an upload session and return signed part URLs.” Then test it against “Commit arrives before final part replication”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How do clients detect conflicts?

Keep metadata strongly consistent while blobs replicate asynchronously across zones. Walk the concrete lifecycle step “Upload chunks independently; verify length and checksum at the storage edge.” Then test it against “Two devices update the same base version”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How do shared links expire?

Drive sync from a monotonic change journal; model versions and permissions independently. Walk the concrete lifecycle step “Build an immutable ordered manifest only when all required parts exist.” Then test it against “Shared-link permission is revoked while cached”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How do you deduplicate content safely?

Use multipart uploads with checksums, resumable sessions, and content-addressed chunks. Walk the concrete lifecycle step “Atomically advance current_version and append one change-journal entry.” Then test it against “Chunk dedupe leaks cross-tenant existence”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How do you migrate storage tiers?

Keep metadata strongly consistent while blobs replicate asynchronously across zones. Walk the concrete lifecycle step “Sync clients resume from a monotonic cursor and resolve divergent edits as versions.” Then test it against “Journal cursor points to compacted history”. 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	a published version is complete, checksummed, authorized, and never silently corrupted.
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 a published version is complete, checksummed, authorized, and never silently corrupted. The online path stays bounded, authoritative state is versioned and partitioned, and asynchronous work is replayable and observable.”

↑ top

Desktop / Web Client

Global API

Auth / Sharing

Metadata Service

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