Search and recommendation in HLD - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Architecture

intermediate

Design a scalable search system for millions of products

You need to design a search system that can handle millions of products with fast response times. Which architecture component is most critical to ensure scalability and low latency?

A A distributed index with sharding and replication

B A batch processing system that updates search results once a day

C A single powerful database server with all data loaded in memory

D A client-side search using local browser storage

💡 Hint

Think about how to handle large data and many users at the same time.

Distributed indexing with sharding and replication allows the system to split data across servers and handle many queries in parallel, ensuring scalability and low latency.

❓ scaling

intermediate

Handling high query volume in recommendation system

Your recommendation system receives a sudden spike in user requests. Which approach best helps maintain performance under high load?

A Store recommendations only in a slow disk-based database

B Recompute recommendations from scratch for every request

C Cache popular recommendations in memory close to users

D Send all requests to a single recommendation server

💡 Hint

Think about reducing repeated work and fast access.

Caching popular recommendations reduces computation and database hits, allowing the system to serve many users quickly during spikes.

❓ tradeoff

advanced

Choosing between real-time and batch recommendation updates

Which tradeoff is true when choosing real-time recommendation updates over batch updates?

A Real-time updates provide fresher recommendations but require more computing resources

B Real-time updates reduce resource usage but increase recommendation delay

C Batch updates provide fresher recommendations than real-time updates

D Batch updates require more frequent user interaction to trigger updates

💡 Hint

Consider freshness versus resource cost.

Real-time updates keep recommendations fresh but need continuous computation, increasing resource use compared to batch processing.

🧠 Conceptual

advanced

Understanding cold start problem in recommendation systems

What is the cold start problem in recommendation systems?

A System crash due to too many recommendations

B Difficulty recommending items to new users with no history

C Slow search queries due to large index size

D Users receiving too many irrelevant recommendations

💡 Hint

Think about what happens when the system has no data about a user.

The cold start problem occurs when the system cannot generate personalized recommendations because it lacks user interaction data.

❓ estimation

expert

Estimating storage needs for a search index

You have 100 million documents averaging 1 KB each. The search index typically requires 30% of the original data size. How much storage is needed for the index?

A Approximately 10 GB

B Approximately 300 GB

C Approximately 3 TB

D Approximately 30 GB

💡 Hint

Calculate 30% of total data size.

100 million documents * 1 KB = 100 million KB = 100 GB. 30% of 100 GB is 30 GB.