Distributed caching (Redis, Memcached) in HLD - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Architecture

intermediate

Designing a distributed cache with Redis

You need to design a distributed caching layer using Redis for a web application that handles millions of users. Which architecture component is essential to ensure data consistency and high availability?

A Deploy Redis as a single instance on a powerful server

B Use Redis Sentinel for automatic failover and monitoring

C Use Memcached with no replication for simplicity

D Store cache data only on the application server memory

💡 Hint

Think about how Redis handles failover and keeps the cache available during node failures.

Redis Sentinel provides monitoring, notification, and automatic failover to keep the cache highly available and consistent. Single instances or local memory do not provide failover, and Memcached lacks built-in replication.

❓ scaling

intermediate

Scaling Memcached for large datasets

Your application uses Memcached to cache user session data. The dataset is growing beyond the capacity of a single Memcached server. What is the best approach to scale Memcached horizontally?

A Use Redis instead because Memcached cannot scale horizontally

B Increase the RAM of the single Memcached server to hold all data

C Use consistent hashing to distribute keys across multiple Memcached nodes

D Store all data in a single Memcached node and replicate it to others

💡 Hint

Consider how Memcached clients decide which server to use for a given key.

Consistent hashing allows clients to distribute keys evenly across multiple Memcached servers, enabling horizontal scaling. Increasing RAM on one server has limits, and Memcached supports horizontal scaling via client-side hashing.

❓ tradeoff

advanced

Choosing between Redis and Memcached for session caching

Your team must choose between Redis and Memcached for caching user sessions. Which tradeoff is most important when deciding?

A Redis supports data persistence and complex data types, Memcached is simpler and faster for pure key-value caching

B Memcached supports transactions, Redis does not

C Redis cannot scale horizontally, Memcached can

D Memcached supports data persistence, Redis does not

💡 Hint

Think about features and use cases of Redis and Memcached.

Redis supports persistence and complex data types like lists and sets, making it versatile. Memcached is simpler and optimized for fast key-value caching but lacks persistence and complex data structures.

🧠 Conceptual

advanced

Cache invalidation strategies in distributed caching

Which cache invalidation strategy is best suited for a distributed cache to maintain data freshness without overwhelming the backend database?

A Write all updates only to the cache and never to the database

B Always delete cache entries immediately after every database update

C Never expire cache entries and rely on manual cache clearing

D Use time-to-live (TTL) expiration combined with lazy loading on cache miss

💡 Hint

Consider balancing freshness and load on the database.

TTL expiration ensures cache entries expire after a set time, and lazy loading reloads data on cache miss. Immediate deletion can cause high load, and never expiring risks stale data. Writing only to cache breaks data consistency.

❓ estimation

expert

Estimating cache capacity for a high-traffic system

Your system expects 10 million active users, each with an average session size of 2 KB. You want to cache 30% of active sessions simultaneously in Redis. Estimate the minimum memory required for the Redis cache, considering 20% overhead for metadata and replication.

A Approximately 7.2 GB

B Approximately 4.8 GB

C Approximately 12 GB

D Approximately 3.6 GB

💡 Hint

Calculate total data size, then add overhead.

30% of 10 million users = 3 million sessions. Each session 2 KB → 6 GB raw data. Adding 20% overhead → 6 GB * 1.2 = 7.2 GB minimum memory.