Content delivery with CDN in HLD - System Design Exercise

Design: Content Delivery Network (CDN) System

Design covers CDN architecture for static content delivery including caching, cache invalidation, and global distribution. Does not cover dynamic content generation or user authentication.

Functional Requirements

FR1: Deliver static content (images, videos, scripts) to users globally with low latency

FR2: Support at least 1 million concurrent users

FR3: Ensure content is cached close to users to reduce origin server load

FR4: Provide cache invalidation to update content when needed

FR5: Handle sudden traffic spikes without downtime

FR6: Support HTTPS for secure content delivery

Non-Functional Requirements

NFR1: 99.9% availability (less than 8.77 hours downtime per year)

NFR2: API response latency p99 under 100ms for cached content

NFR3: Origin server should not be overwhelmed by requests

NFR4: Content freshness must be maintained with cache TTL and invalidation

Think Before You Design

Questions to Ask

❓ Question 1

❓ Question 2

❓ Question 3

❓ Question 4

❓ Question 5

Key Components

Edge servers located globally for caching content

Origin servers hosting the original content

Load balancers to distribute user requests

Cache invalidation mechanism

DNS routing to direct users to nearest edge server

HTTPS termination at edge

Design Patterns

Cache-aside pattern for cache population

Geo-DNS or Anycast for routing users to nearest edge

TTL-based cache expiration

Push vs pull CDN content update strategies

Reference Architecture

DNS Resolver

directs to nearest→Edge CDN Server Cluster

↓

Cache Storage

↓

Origin Server

Components

DNS Resolver

Geo-DNS or Anycast DNS

Direct users to the nearest edge server based on location

Edge CDN Server Cluster

Distributed edge servers with caching software (e.g., Nginx, Varnish)

Cache and serve content close to users to reduce latency

Cache Storage

In-memory or disk cache on edge servers

Store cached content for fast retrieval

Origin Server

Web servers hosting original static content

Serve content when not available in cache

Cache Invalidation Service

API or messaging system

Update or remove cached content when origin content changes

Load Balancer

Global and local load balancers

Distribute incoming requests evenly across edge servers

Request Flow

1. User requests content via browser.

2. DNS resolver directs user to nearest edge CDN server.

3. Edge server checks cache storage for requested content.

4. If content is cached and fresh, edge server returns content immediately.

5. If content is missing or stale, edge server requests content from origin server.

6. Origin server responds with content.

7. Edge server caches the content and returns it to user.

8. Cache invalidation service updates or removes cached content when origin content changes.

9. Subsequent user requests get fresh content from edge cache.

Database Schema

Not applicable as CDN primarily uses cache storage and origin servers; metadata about cached content includes content key, TTL, last updated timestamp, and invalidation status.

Scaling Discussion

Bottlenecks

Origin server overload during cache misses or traffic spikes

Edge server cache storage limits causing frequent cache evictions

DNS resolver latency or failure affecting user routing

Cache invalidation delays causing stale content delivery

Solutions

Use origin server autoscaling and rate limiting to handle load

Increase edge cache capacity and use efficient eviction policies (LRU)

Deploy multiple DNS resolvers with failover and low latency

Implement near real-time cache invalidation with push updates or short TTLs

Interview Tips

Time: Spend 10 minutes clarifying requirements and constraints, 20 minutes designing architecture and data flow, 10 minutes discussing scaling and trade-offs, 5 minutes summarizing.

Explain how CDN reduces latency by caching content close to users

Discuss cache miss handling and origin server protection

Describe cache invalidation strategies to maintain content freshness

Highlight global distribution and DNS routing for scalability

Mention security considerations like HTTPS termination at edge

Practice

1. What is the primary purpose of a Content Delivery Network (CDN)?

easy

A. To store user data permanently

B. To cache content closer to users for faster delivery

C. To replace the origin server completely

D. To encrypt all website content

Solution

1. Step 1: Understand CDN function

   A CDN stores copies of content in multiple locations worldwide to reduce latency.

2. Step 2: Identify main benefit

   This caching near users speeds up content delivery and reduces load on the origin server.

3. Final Answer:

   To cache content closer to users for faster delivery -> Option B

4. Quick Check:

   CDN = caching near users [OK]

Hint: CDN = cache content near users for speed [OK]

Common Mistakes:

• Thinking CDN replaces origin server
• Confusing CDN with data storage
• Assuming CDN encrypts all content

2. Which of the following is the correct sequence when a user requests content served via a CDN?

easy

A. User -> Origin Server -> CDN -> User

B. Origin Server -> User -> CDN

C. User -> CDN -> Origin Server (if needed) -> CDN -> User

D. User -> CDN -> User -> Origin Server

Solution

1. Step 1: Trace request flow in CDN

   User first contacts the CDN edge server; if content is cached, it responds immediately.

2. Step 2: Understand cache miss handling

   If content is not cached, CDN fetches it from the origin server, caches it, then serves user.

3. Final Answer:

   User -> CDN -> Origin Server (if needed) -> CDN -> User -> Option C

4. Quick Check:

   Request flow = CDN first, origin fallback [OK]

Hint: User hits CDN first, origin only if cache miss [OK]

Common Mistakes:

• Assuming user contacts origin server first
• Ignoring CDN cache miss flow
• Confusing response direction

3. Consider this simplified code snippet representing a CDN cache check:

cache = {"index.html": "content"}
request = "style.css"

if request in cache:
    response = cache[request]
else:
    response = "fetch from origin"
print(response)

What will be printed?

medium

A. fetch from origin

B. KeyError

C. content

D. None

Solution

1. Step 1: Check if request key exists in cache

   The request is "style.css" which is not in cache keys (only "index.html" exists).

2. Step 2: Determine else branch execution

   Since request not found, else branch runs, setting response to "fetch from origin".

3. Final Answer:

   fetch from origin -> Option A

4. Quick Check:

   Cache miss prints fetch from origin [OK]

Hint: If key missing in cache, else branch runs [OK]

Common Mistakes:

• Assuming cache returns default value automatically
• Expecting KeyError without else
• Confusing cache keys with values

4. A CDN is configured but users report slow content loading. Which of these is the most likely cause?

medium

A. The CDN is caching all content correctly

B. Users have slow internet unrelated to CDN

C. The origin server is down but CDN cache is fresh

D. The CDN edge servers are geographically far from users

Solution

1. Step 1: Analyze CDN slow delivery causes

   If CDN edge servers are far from users, latency increases, causing slow loading.

2. Step 2: Evaluate other options

   The CDN is caching all content correctly means caching works, so unlikely cause. The origin server is down but CDN cache is fresh means cache serves content fast. Users have slow internet unrelated to CDN is user-side, not CDN issue.

3. Final Answer:

   The CDN edge servers are geographically far from users -> Option D

4. Quick Check:

   Distance to CDN edge affects speed [OK]

Hint: Far CDN edges cause slow delivery [OK]

Common Mistakes:

• Blaming origin server when cache is fresh
• Ignoring CDN server location
• Assuming caching always fixes speed

5. You want to design a CDN system that handles sudden traffic spikes without overloading the origin server. Which combination of strategies is best?

hard

A. Use aggressive caching with long TTL and multiple CDN edge locations

B. Disable caching and let all requests go to origin server

C. Use a single CDN edge server with short TTL caching

D. Serve content only from origin server with load balancer

Solution

1. Step 1: Understand traffic spike handling

   Long TTL caching reduces origin hits; multiple edges distribute load geographically.

2. Step 2: Evaluate options for scalability

   Disable caching and let all requests go to origin server and D overload origin server. Use a single CDN edge server with short TTL caching limits caching benefits and single edge risks bottleneck.

3. Final Answer:

   Use aggressive caching with long TTL and multiple CDN edge locations -> Option A

4. Quick Check:

   Long TTL + many edges = scalable CDN [OK]

Hint: Long cache + many edges handle spikes best [OK]

Common Mistakes:

• Disabling caching under traffic spikes
• Relying on single edge server
• Ignoring TTL impact on origin load