Scalability Interview Questions - Hard

calendarDec 13, 2025 · 2 min read · scalability interview hard distributed-systems performance  ·
Share on: linkedincopy

Hard-level scalability interview questions covering extreme scale, global distribution, and advanced optimization.

Q1: Design a globally distributed database with strong consistency.

Answer:

Challenge: CAP theorem - can't have all three (Consistency, Availability, Partition tolerance).

Two-Phase Commit (2PC):

Spanner-like Architecture (Google Spanner):

Q2: How do you handle 1 million concurrent WebSocket connections?

Answer:

Connection Distribution:

Message Broadcasting:

Optimization Techniques:

Q3: Design a system to handle 1 billion daily active users.

Answer:

Scale Requirements:

  • 1B DAU
  • Peak: 100K requests/sec
  • 99.99% uptime
  • <100ms latency globally

Regional Architecture:

Data Sharding Strategy:

Q4: Implement distributed rate limiting across data centers.

Answer:

Challenge: Maintain accurate counts across regions with minimal latency.

Quota Allocation Strategy:

Token Bucket with Gossip Protocol:

Q5: Design auto-scaling for unpredictable traffic spikes.

Answer:

Predictive Scaling:

Multi-Tier Scaling:

Scaling Policies:

Q6: How do you handle database migrations at scale with zero downtime?

Answer:

Expand-Contract Pattern:

Online Schema Change Tools:

Q7: Design a system to deduplicate 1PB of data.

Answer:

Chunking Strategy:

Distributed Deduplication:

Bloom Filter for Scale:

Q8: Implement distributed tracing for microservices.

Answer:

Trace Context Propagation:

Trace Visualization:

Q9: Design a system for real-time analytics on streaming data.

Answer:

Lambda Architecture:

Windowing Strategies:

Q10: How do you handle cascading failures in microservices?

Answer:

Prevention Strategies:

Bulkhead Pattern:

Summary

Hard scalability topics:

  • Global Consistency: Spanner, 2PC, consensus
  • Million Connections: WebSocket scaling, event-driven I/O
  • Billion Users: Regional architecture, sharding
  • Distributed Rate Limiting: Quota allocation, gossip
  • Auto-Scaling: Predictive, multi-tier
  • Zero-Downtime Migrations: Expand-contract pattern
  • Deduplication at Scale: Content-addressable storage, Bloom filters
  • Distributed Tracing: Span propagation, visualization
  • Real-Time Analytics: Stream processing, windowing
  • Cascading Failures: Circuit breakers, bulkheads

These techniques enable building systems at extreme scale with high reliability.

Related Snippets