Video Streaming Platform — 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 video streaming platform 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 1M uploads/day · exabytes served · startup < 2 s.
Protect the named business invariant under concurrency and partial failure.
Scale by an explicit aggregate/partition key without one global mutex.
Degrade intentionally when cache, queue, store, or regional ownership is unavailable.

Primary invariant: each accepted video streaming platform operation has one stable result and retries do not repeat its business effect.

↑ top

2. Capacity assumptions

State assumptions before choosing technology. The supplied workload is 1M uploads/day · exabytes served · startup < 2 s. Convert it into peak throughput, retained records, bandwidth, and failure budget; then connect each number to a component.

Dimension	Working estimate	Design implication
Traffic	1M uploads/day · exabytes served · startup < 2 s	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/videos/uploadsresumable source session

POST /v1/videos/{id}/publishatomic manifest publish

GET /v1/playback/{id}authorized manifest

POST /v1/qoebatched playback telemetry

Authoritative records and indexes

video(video_id PK, owner_id, state, source_uri, manifest_version)

rendition(video_id, codec, resolution, bitrate, segment_prefix, state)

job(video_id, profile, attempt, lease_until)

manifest(video_id, version, renditions, published_at)

Command and event envelopes

# Command envelope
operation: POST /v1/videos/uploads
request_id: req_01J...
idempotency_key: idem_7f2...
aggregate_key: video-streaming_123
expected_version: 17
payload: {...}

# Durable event
event_id: evt_01J...
aggregate_key: video-streaming_123
event_type: video_streaming_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
Creator Client	Owns client protocol, request identity, retry, and resume behavior. In this design it participates in: Verify and probe the committed source before creating rendition jobs.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents the external caller from also owning the contract and failure behavior of Resumable Upload 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 “One rendition repeatedly fails” 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 Creator Client boundary before extracting or scaling it.
Resumable Upload API	Owns authentication, validation, quotas, deadlines, and routing. In this design it participates in: Fan out codec/resolution tasks with deterministic output keys and retry-safe jobs.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Creator Client from also owning the contract and failure behavior of Source Object Store.	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 “Manifest references a missing segment” to Creator 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 Resumable Upload API boundary before extracting or scaling it.
Source Object Store	Owns authoritative versioned state or immutable evidence. In this design it participates in: Validate duration, keyframes, audio sync, and segment checksums for every rendition.	Defines the source of truth and commit boundary. It prevents Resumable Upload API from also owning the contract and failure behavior of Media Probe.	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 “CDN region serves stale authorization” to Resumable Upload API.	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 Source Object Store boundary before extracting or scaling it.
Media Probe	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Atomically publish one manifest version only after the minimum ladder is complete.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Source Object Store from also owning the contract and failure behavior of Workflow Orchestrator.	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 “Upload resumes with conflicting part” to Source 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 Media Probe boundary before extracting or scaling it.
Workflow Orchestrator	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Authorize playback, issue short-lived tokens, and let the player adapt from throughput/buffer.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Media Probe from also owning the contract and failure behavior of Transcode Fleet.	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 “Live encoder falls behind wall clock” to Media Probe.	Costs an interface, deployment, telemetry, ownership, and failure mode. Keep it in-process first; extract when independent scale or reliability pays for the boundary. Measure the latency, saturation, lag, or mismatch at the Workflow Orchestrator boundary before extracting or scaling it.
Transcode Fleet	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Verify and probe the committed source before creating rendition jobs.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Workflow Orchestrator from also owning the contract and failure behavior of Quality Validator.	Perform the work before acknowledging the caller. This gives simpler consistency but unbounded tail latency and fragile retries. Here that simpler choice would couple “One rendition repeatedly fails” to Workflow Orchestrator.	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 Transcode Fleet boundary before extracting or scaling it.
Quality Validator	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Fan out codec/resolution tasks with deterministic output keys and retry-safe jobs.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Transcode Fleet from also owning the contract and failure behavior of Rendition 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 “Manifest references a missing segment” to Transcode Fleet.	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 Quality Validator boundary before extracting or scaling it.
Rendition Store	Owns authoritative versioned state or immutable evidence. In this design it participates in: Validate duration, keyframes, audio sync, and segment checksums for every rendition.	Defines the source of truth and commit boundary. It prevents Quality Validator from also owning the contract and failure behavior of Manifest Publisher.	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 “CDN region serves stale authorization” to Quality Validator.	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 Rendition Store boundary before extracting or scaling it.
Manifest Publisher	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Atomically publish one manifest version only after the minimum ladder is complete.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Rendition Store from also owning the contract and failure behavior of Catalog 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 “Upload resumes with conflicting part” to Rendition 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 Manifest Publisher boundary before extracting or scaling it.
Catalog DB	Preserves durable ordering, replay, and consumer checkpoints. In this design it participates in: Authorize playback, issue short-lived tokens, and let the player adapt from throughput/buffer.	Defines the source of truth and commit boundary. It prevents Manifest Publisher from also owning the contract and failure behavior of Playback API.	Call every downstream consumer synchronously. That couples user latency and availability to analytics, notification, indexing, and audit dependencies. Here that simpler choice would couple “Live encoder falls behind wall clock” to Manifest Publisher.	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 Catalog DB boundary before extracting or scaling it.
Playback API	Owns authentication, validation, quotas, deadlines, and routing. In this design it participates in: Verify and probe the committed source before creating rendition jobs.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Catalog DB from also owning the contract and failure behavior of Token / DRM Service.	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 “One rendition repeatedly fails” to Catalog DB.	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 Playback API boundary before extracting or scaling it.
Token / DRM Service	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Fan out codec/resolution tasks with deterministic output keys and retry-safe jobs.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Playback API from also owning the contract and failure behavior of Multi-CDN Router.	Keep this behavior inside an adjacent service. Prefer that simpler option until independent scale, ownership, policy, or failure isolation is measurable. Here that simpler choice would couple “Manifest references a missing segment” to Playback 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 Token / DRM Service boundary before extracting or scaling it.
Multi-CDN Router	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Validate duration, keyframes, audio sync, and segment checksums for every rendition.	Removes repeated work from the latency-critical path. It prevents Token / DRM Service from also owning the contract and failure behavior of Viewer Player.	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 “CDN region serves stale authorization” to Token / DRM Service.	Costs invalidation, staleness, stampede protection, and memory. Add or split it only after the workload (1M uploads/day · exabytes served · startup < 2 s) proves repeated reads dominate. Measure the latency, saturation, lag, or mismatch at the Multi-CDN Router boundary before extracting or scaling it.
Viewer Player	Owns one bounded domain decision, transition, or operational capability. In this design it participates in: Atomically publish one manifest version only after the minimum ladder is complete.	Separates a distinct scale, ownership, policy, or failure boundary. It prevents Multi-CDN Router from also owning the contract and failure behavior of QoE Event Stream.	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 “Upload resumes with conflicting part” to Multi-CDN 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 Viewer Player boundary before extracting or scaling it.
QoE Event Stream	Preserves durable ordering, replay, and consumer checkpoints. In this design it participates in: Authorize playback, issue short-lived tokens, and let the player adapt from throughput/buffer.	Decouples bursts and failure from user acknowledgment. It prevents Viewer Player from also owning the contract and failure behavior of operators and repair tooling.	Call every downstream consumer synchronously. That couples user latency and availability to analytics, notification, indexing, and audit dependencies. Here that simpler choice would couple “Live encoder falls behind wall clock” to Viewer Player.	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 QoE Event Stream boundary before extracting or scaling it.

Decision 01

Chosen boundary vs simpler monolith

Store the source once, generate segmented bitrate ladders asynchronously, and publish only complete manifests.

A single deployable with one transactional store is the right starting point for a small video streaming platform workload. The proposed boundary becomes justified when store the source once, generate segmented bitrate ladders asynchronously, and publish only complete manifests. Separating it contains the named failure “One rendition repeatedly fails” without placing unrelated work on the critical path.

Decision 02

Correctness vs availability

Put immutable segments behind a CDN; keep authorization and playback manifests at the control plane.

The design coordinates only where each accepted video streaming platform operation has one stable result and retries do not repeat its business effect. Everywhere else it accepts bounded staleness or asynchronous propagation. The alternative—strong coordination for every projection—simplifies the mental model but increases tail latency and turns partitions into broad write unavailability.

Decision 03

Operational cost and evolution

Choose bitrate using measured throughput and buffer health; prefetch conservatively on mobile.

This choice adds dashboards, ownership, schema compatibility, capacity planning, and recovery runbooks. Start with the minimum topology that supports “Validate duration, keyframes, audio sync, and segment checksums for every rendition.” and extract another service only when measured throughput, blast radius, data lifecycle, or team ownership changes independently.

Why this design: the synchronous path contains only the decisions required to protect each accepted video streaming platform operation has one stable result and retries do not repeat its business effect. Durable projections, notifications, indexing, analytics, and repair are isolated so their load or failure cannot corrupt the acknowledged operation.

Correctness boundary

Store the source once, generate segmented bitrate ladders asynchronously, and publish only complete manifests.

Scale boundary

Put immutable segments behind a CDN; keep authorization and playback manifests at the control plane.

Operational boundary

Choose bitrate using measured throughput and buffer health; prefetch conservatively on mobile.

↑ top

5. Critical flow and algorithm

Verify and probe the committed source before creating rendition jobs.

Fan out codec/resolution tasks with deterministic output keys and retry-safe jobs.

Validate duration, keyframes, audio sync, and segment checksums for every rendition.

Atomically publish one manifest version only after the minimum ladder is complete.

Authorize playback, issue short-lived tokens, and let the player adapt from throughput/buffer.

Retry-safe mutation skeleton

identity = authenticate(request)
prior = idempotency.get(identity, request.key)
if prior: return prior.result

current = authoritative_store.read(partition_key)
next_state, events = domain.transition(current, request)
authoritative_store.compare_and_set(current.version, next_state)
idempotency.commit(request.key, next_state.result)
outbox.append(events)
return next_state.result

Exactly-once business effect: the network can duplicate requests. Stable identities, conditional state transitions, immutable event IDs, and reconciliation make the user-visible effect happen once.

↑ top

6. Caching, hot keys, and backpressure

Cache policy

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

Hot partition or tenant

Measure per-key and per-tenant skew. Replicate safe reads, isolate the offender, salt only when ordering permits, and enforce budgets.

Backpressure

Bound queues, buffers, connection pools, and concurrency. Shed low-priority work and scale from lag/oldest-item age rather than CPU alone.

Stampede and retry storms

Use single-flight/coalescing, jittered expiry, bounded retries, exponential backoff, and circuit breakers so one dependency failure does not multiply traffic.

Cache outage rule: never allow every miss to fall through simultaneously. Use local protection, request coalescing, circuit breakers, bounded pools, and controlled fallback.

↑ top

7. Partitioning and consistency

partition = stable_hash(primary_or_aggregate_key) → shard owner
write = conditional_update(expected_version)
read_projection = versioned and rebuildable

Strong boundary

each accepted video streaming platform operation has one stable result and retries do not repeat its business effect. Use a uniqueness constraint, serializable/conditional write, expected version, or fencing token at this boundary.

Eventual boundary

Analytics, notifications, search, read models, and operational dashboards may lag when freshness is observable and replay exists.

Operation	Consistency	Reason
Authoritative mutation	Conditional / serializable	Protect uniqueness, ownership, money, inventory, order, or legal state transition.
Read after own write	Primary, session token, or cache warm	The caller must observe the result it just committed.
Ordinary read projection	Bounded stale where safe	Replicas and caches improve latency when version/freshness is visible.
Analytics and operations	Eventual + replay	Freshness is less important than isolating the online path.

Route every mutation for one aggregate to one owner/partition epoch.
Use cursors rather than offsets for changing ordered collections.
Retain tombstones until replication and repair windows close.
Move shard ownership through versioned maps and dual-read verification.

↑ top

8. Failure drills and recovery

Failure	Runtime behavior	Recovery proof
One rendition repeatedly fails	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.
Manifest references a missing segment	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.
CDN region serves stale authorization	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.
Upload resumes with conflicting part	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.
Live encoder falls behind wall clock	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: One rendition repeatedly fails
operation_id: op_01J...
aggregate_key: video-streaming_123
last_durable_version: 18
projection_checkpoint: partition-07@offset-482901
retry_policy: same_identity / bounded_backoff
reconciliation_status: pending
owner: QoE Event Stream

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: One rendition repeatedly fails — 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 resumable uploads work?

Store the source once, generate segmented bitrate ladders asynchronously, and publish only complete manifests. Walk the concrete lifecycle step “Verify and probe the committed source before creating rendition jobs.” Then test it against “One rendition repeatedly fails”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How do you pick a bitrate ladder?

Put immutable segments behind a CDN; keep authorization and playback manifests at the control plane. Walk the concrete lifecycle step “Fan out codec/resolution tasks with deterministic output keys and retry-safe jobs.” Then test it against “Manifest references a missing segment”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How do live streams differ?

Choose bitrate using measured throughput and buffer health; prefetch conservatively on mobile. Walk the concrete lifecycle step “Validate duration, keyframes, audio sync, and segment checksums for every rendition.” Then test it against “CDN region serves stale authorization”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

How do you protect paid content?

Store the source once, generate segmented bitrate ladders asynchronously, and publish only complete manifests. Walk the concrete lifecycle step “Atomically publish one manifest version only after the minimum ladder is complete.” Then test it against “Upload resumes with conflicting part”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

What happens when transcoding partially fails?

Put immutable segments behind a CDN; keep authorization and playback manifests at the control plane. Walk the concrete lifecycle step “Authorize playback, issue short-lived tokens, and let the player adapt from throughput/buffer.” Then test it against “Live encoder falls behind wall clock”. State the retry identity, durable checkpoint, recovery signal, and cost of the alternative.

Answer pattern: state the invariant, choose a mechanism, walk one failure, name the recovery signal, and explain the cost of the alternative.

↑ top

11. How to present this in 45 minutes

Time	What to cover
0–5 min	Clarify scope, workload, SLOs, and the correctness invariant.
5–10 min	Estimate throughput/storage and define concrete APIs and records.
10–20 min	Draw online, durable-write, and asynchronous/repair paths.
20–30 min	Explain the exact lifecycle, partition key, and concurrency control.
30–38 min	Discuss caching, hot keys, replication, and backpressure.
38–43 min	Run failure drills, security, observability, and regional evolution.
43–45 min	Summarize the five most important choices and trade-offs.

Final decision summary

Invariant	each accepted video streaming platform operation has one stable result and retries do not repeat its business effect.
Partitioning	Stable aggregate/key hash with versioned ownership and measured hot-key isolation.
Commit	Expected version or uniqueness at the authoritative state boundary.
Async work	Durable outbox/log, idempotent consumers, checkpoints, and replay.
Recovery	Same operation identity, durable truth lookup, and business reconciliation.

Closing summary

CLOSING “The design protects each accepted video streaming platform operation has one stable result and retries do not repeat its business effect. The online path stays bounded, authoritative state is versioned and partitioned, and asynchronous work is replayable and observable.”

↑ top

Creator Client

Resumable Upload API

Source Object Store

Media Probe

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