File System — 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 system 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 Billions files · TB volumes · crash-consistent metadata.
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: every object transition is legal and atomic at one aggregate boundary.

↑ top

2. Capacity assumptions

State assumptions before choosing technology. The supplied workload is Billions files · TB volumes · crash-consistent metadata. Convert it into peak throughput, retained records, bandwidth, and failure budget; then connect each number to a component.

Dimension	Working estimate	Design implication
Traffic	Billions files · TB volumes · crash-consistent metadata	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

open/read/write/fsync/close

mkdir/link/unlink/rename

stat/chmod/chown

snapshot/create/restore

Authoritative records and indexes

inode(inode_no PK, type, mode, owner, size, version)

directory_entry(parent_inode, name, child_inode)

extent(inode, logical_start, physical_start, length)

journal(txn_id, metadata_updates, commit_checksum)

Command and event envelopes

# Command envelope
operation: open/read/write/fsync/close
request_id: req_01J...
idempotency_key: idem_7f2...
aggregate_key: file-system_123
expected_version: 17
payload: {...}

# Durable event
event_id: evt_01J...
aggregate_key: file-system_123
event_type: file_system_committed
schema_version: 3
trace_id: tr_8ac...

The authoritative aggregate and idempotency record share a transactional or conditional-write boundary. Event streams, search indexes, caches, analytics, and dashboards are rebuildable projections.

↑ top

4. Why this architecture and what it costs

The boxes are not a shopping list. Each boundary exists because it owns a different invariant, scaling dimension, data lifecycle, security policy, or failure domain. The table makes those choices explicit and shows when the design should remain simpler.

Component	Owned responsibility	Why separate it	Rejected alternative	Cost and evolution trigger
System Call API	Owns authentication, validation, quotas, deadlines, and routing. In this design it participates in: Resolve mount and path components through dentry/inode cache with permission checks.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents the external caller from also owning the contract and failure behavior of VFS Layer.	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 “Power fails during rename” to the external caller.	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 System Call API boundary before extracting or scaling it.
VFS Layer	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Map file offsets to extents and serve reads/writes through page cache.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents System Call API from also owning the contract and failure behavior of Path Resolver.	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 commits metadata before file data” to System Call 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 VFS Layer boundary before extracting or scaling it.
Path Resolver	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Allocate/free blocks and update inode/directory metadata inside journal transaction.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents VFS Layer from also owning the contract and failure behavior of Mount 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 “Hard link survives unlink race” to VFS Layer.	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 Path Resolver boundary before extracting or scaling it.
Mount Table	Owns authoritative versioned state or immutable evidence. In this design it participates in: fsync orders data and journal commit according to durability contract.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Path Resolver from also owning the contract and failure behavior of Dentry / Inode Cache.	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 “Free-space bitmap diverges” to Path Resolver.	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 Mount Table boundary before extracting or scaling it.
Dentry / Inode Cache	Serves a low-latency, disposable projection with authoritative fallback. In this design it participates in: On crash replay committed journal, discard incomplete intents, then scrub allocation/checksum state.	Removes repeated work from the latency-critical path. It prevents Mount Table from also owning the contract and failure behavior of Permission Checker.	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 “Snapshot COW exhausts capacity” to Mount Table.	Costs invalidation, staleness, stampede protection, and memory. Add or split it only after the workload (Billions files · TB volumes · crash-consistent metadata) proves repeated reads dominate. Measure the latency, saturation, lag, or mismatch at the Dentry / Inode Cache boundary before extracting or scaling it.
Permission Checker	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Resolve mount and path components through dentry/inode cache with permission checks.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Dentry / Inode Cache from also owning the contract and failure behavior of Directory B-Tree.	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 “Power fails during rename” to Dentry / Inode 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 Permission Checker boundary before extracting or scaling it.
Directory B-Tree	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Map file offsets to extents and serve reads/writes through page cache.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Permission Checker from also owning the contract and failure behavior of Inode 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 “Journal commits metadata before file data” to Permission Checker.	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 Directory B-Tree boundary before extracting or scaling it.
Inode Table	Owns authoritative versioned state or immutable evidence. In this design it participates in: Allocate/free blocks and update inode/directory metadata inside journal transaction.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Directory B-Tree from also owning the contract and failure behavior of Extent Mapper.	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 “Hard link survives unlink race” to Directory B-Tree.	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 Inode Table boundary before extracting or scaling it.
Extent Mapper	Owns client protocol, request identity, retry, and resume behavior. In this design it participates in: fsync orders data and journal commit according to durability contract.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Inode Table from also owning the contract and failure behavior of Block Allocator.	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 “Free-space bitmap diverges” to Inode Table.	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 Extent Mapper boundary before extracting or scaling it.
Block Allocator	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: On crash replay committed journal, discard incomplete intents, then scrub allocation/checksum state.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Extent Mapper from also owning the contract and failure behavior of Page Cache.	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 “Snapshot COW exhausts capacity” to Extent Mapper.	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 Block Allocator boundary before extracting or scaling it.
Page Cache	Serves a low-latency, disposable projection with authoritative fallback. In this design it participates in: Resolve mount and path components through dentry/inode cache with permission checks.	Removes repeated work from the latency-critical path. It prevents Block Allocator from also owning the contract and failure behavior of Writeback Worker.	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 “Power fails during rename” to Block Allocator.	Costs invalidation, staleness, stampede protection, and memory. Add or split it only after the workload (Billions files · TB volumes · crash-consistent metadata) proves repeated reads dominate. Measure the latency, saturation, lag, or mismatch at the Page Cache boundary before extracting or scaling it.
Writeback Worker	Executes idempotent asynchronous work under lease and retry policy. In this design it participates in: Map file offsets to extents and serve reads/writes through page cache.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Page Cache from also owning the contract and failure behavior of Metadata Journal.	Perform the work before acknowledging the caller. This gives simpler consistency but unbounded tail latency and fragile retries. Here that simpler choice would couple “Journal commits metadata before file data” to Page Cache.	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 Writeback Worker boundary before extracting or scaling it.
Metadata Journal	Preserves durable ordering, replay, and consumer checkpoints. In this design it participates in: Allocate/free blocks and update inode/directory metadata inside journal transaction.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Writeback Worker from also owning the contract and failure behavior of Snapshot / COW.	Call every downstream consumer synchronously. That couples user latency and availability to analytics, notification, indexing, and audit dependencies. Here that simpler choice would couple “Hard link survives unlink race” to Writeback Worker.	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 Metadata Journal boundary before extracting or scaling it.
Snapshot / COW	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: fsync orders data and journal commit according to durability contract.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Metadata Journal from also owning the contract and failure behavior of Fsck / Scrubber.	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 “Free-space bitmap diverges” to Metadata 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 Snapshot / COW boundary before extracting or scaling it.
Fsck / Scrubber	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: On crash replay committed journal, discard incomplete intents, then scrub allocation/checksum state.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Snapshot / COW 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 “Snapshot COW exhausts capacity” to Snapshot / COW.	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 Fsck / Scrubber boundary before extracting or scaling it.

Decision 01

Chosen boundary vs simpler monolith

Separate directory names from inode identity so links and renames do not copy file metadata or data blocks.

A single deployable with one transactional store is the right starting point for a small file system workload. The proposed boundary becomes justified when separate directory names from inode identity so links and renames do not copy file metadata or data blocks. Separating it contains the named failure “Power fails during rename” without placing unrelated work on the critical path.

Decision 02

Correctness vs availability

Journal metadata intent before applying scattered updates; copy-on-write is an alternative with different fragmentation and snapshot costs.

The design coordinates only where every object transition is legal and atomic at one aggregate boundary. 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

Page/block caching accelerates I/O but writeback ordering must preserve fsync and crash-consistency contracts.

This choice adds dashboards, ownership, schema compatibility, capacity planning, and recovery runbooks. Start with the minimum topology that supports “Allocate/free blocks and update inode/directory metadata inside journal transaction.” 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 every object transition is legal and atomic at one aggregate boundary. Durable projections, notifications, indexing, analytics, and repair are isolated so their load or failure cannot corrupt the acknowledged operation.

Correctness boundary

Separate directory names from inode identity so links and renames do not copy file metadata or data blocks.

Scale boundary

Journal metadata intent before applying scattered updates; copy-on-write is an alternative with different fragmentation and snapshot costs.

Operational boundary

Page/block caching accelerates I/O but writeback ordering must preserve fsync and crash-consistency contracts.

↑ top

5. Critical flow and algorithm

Resolve mount and path components through dentry/inode cache with permission checks.

Map file offsets to extents and serve reads/writes through page cache.

Allocate/free blocks and update inode/directory metadata inside journal transaction.

fsync orders data and journal commit according to durability contract.

On crash replay committed journal, discard incomplete intents, then scrub allocation/checksum state.

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

every object transition is legal and atomic at one aggregate boundary. 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
Power fails during rename	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 commits metadata before file data	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.
Hard link survives unlink race	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.
Free-space bitmap diverges	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.
Snapshot COW exhausts capacity	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.

Recovery checkpoint

# Recovery checkpoint
scenario: Power fails during rename
operation_id: op_01J...
aggregate_key: file-system_123
last_durable_version: 18
projection_checkpoint: partition-07@offset-482901
retry_policy: same_identity / bounded_backoff
reconciliation_status: pending
owner: Fsck / Scrubber

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: Power fails during rename — 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 rename stay atomic?

Separate directory names from inode identity so links and renames do not copy file metadata or data blocks. Walk the concrete lifecycle step “Resolve mount and path components through dentry/inode cache with permission checks.” Then test it against “Power fails during rename”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How are hard links modeled?

Journal metadata intent before applying scattered updates; copy-on-write is an alternative with different fragmentation and snapshot costs. Walk the concrete lifecycle step “Map file offsets to extents and serve reads/writes through page cache.” Then test it against “Journal commits metadata before file data”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How does journaling recover?

Page/block caching accelerates I/O but writeback ordering must preserve fsync and crash-consistency contracts. Walk the concrete lifecycle step “Allocate/free blocks and update inode/directory metadata inside journal transaction.” Then test it against “Hard link survives unlink race”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How are free blocks allocated?

Separate directory names from inode identity so links and renames do not copy file metadata or data blocks. Walk the concrete lifecycle step “fsync orders data and journal commit according to durability contract.” Then test it against “Free-space bitmap diverges”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How do snapshots change the design?

Journal metadata intent before applying scattered updates; copy-on-write is an alternative with different fragmentation and snapshot costs. Walk the concrete lifecycle step “On crash replay committed journal, discard incomplete intents, then scrub allocation/checksum state.” Then test it against “Snapshot COW exhausts capacity”. 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	every object transition is legal and atomic at one aggregate boundary.
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 every object transition is legal and atomic at one aggregate boundary. The online path stays bounded, authoritative state is versioned and partitioned, and asynchronous work is replayable and observable.”

↑ top

System Call API

VFS Layer

Path Resolver

Mount Table

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