Media storage and CDN in HLD - Scalability & System Analysis

Scalability Analysis - Media storage and CDN

Growth Table: Media Storage and CDN Scaling

Users	Media Storage	CDN Usage	Network Traffic	Latency
100 users	Single storage server, local disk	Minimal or no CDN	Low, direct fetch from storage	Low latency, direct access
10,000 users	Distributed storage cluster, object storage	Basic CDN with few edge nodes	Moderate, some caching	Improved latency via CDN edges
1,000,000 users	Highly scalable object storage (S3-like), multi-region	Global CDN with many edge locations	High, CDN offloads origin	Low latency globally
100,000,000 users	Multi-cloud, geo-redundant storage, sharded data	Advanced CDN with dynamic content optimization	Very high, optimized delivery	Consistent low latency worldwide

First Bottleneck

At small scale, the media storage server disk I/O and network bandwidth limit throughput.

At medium scale (~10K users), the origin storage bandwidth and read latency become bottlenecks.

At large scale (1M+ users), the CDN edge cache capacity and cache miss rate impact performance.

Without CDN, origin servers get overwhelmed by traffic spikes.

Scaling Solutions

• Horizontal scaling: Add more storage nodes and CDN edge servers to distribute load.
• Caching: Use CDN to cache media close to users, reducing origin load.
• Sharding: Partition media storage by region or content type to improve access speed.
• Multi-region replication: Store copies of media in multiple geographic locations.
• Compression and optimization: Reduce media size for faster delivery.
• Load balancing: Distribute requests evenly across storage and CDN nodes.

Back-of-Envelope Cost Analysis

Assuming 1M users, each streaming 1 video per day of 5 MB:

• Requests per second (QPS): ~12 (1M users * 1 request / 86400 seconds)
• Daily data transfer: 5 TB (1M * 5 MB)
• Bandwidth needed at origin: Reduced by CDN cache hit ratio (e.g., 90% cache hit reduces origin bandwidth to 0.5 TB)
• Storage needed: Depends on retention, e.g., 30 days = 150 TB
• Network bandwidth: CDN edges handle most traffic, origin bandwidth is bottleneck without caching

Interview Tip

Start by defining user scale and media size.

Identify origin storage limits and CDN role early.

Discuss caching strategies and geographic distribution.

Explain how to handle cache misses and data consistency.

Always mention cost and latency trade-offs.

Self Check Question

Your media storage origin handles 1000 QPS. Traffic grows 10x. What do you do first?

Answer: Deploy or increase CDN edge caching to offload origin servers and reduce direct requests, preventing origin overload.

Key Result

Media storage scales by adding distributed storage and using CDN caching to offload origin servers; the first bottleneck is origin bandwidth and disk I/O, solved by CDN and horizontal scaling.