System Design Interview Flow

System design interviews are a critical part of technical hiring for senior engineering roles. In this post, we’ll explore the key principles, structure, and topics you need to master to excel in a 1-hour system design interview. The discussion includes essential steps like clarifying requirements, API design, data modeling, failure design, and post-interview reflection.

---

Key Steps in a System Design Interview

1. Clarify Requirements and Discuss Tradeoffs

Start by clarifying functional and non-functional requirements, such as:

• Functional: User roles, API endpoints, and system behavior.

• Non-functional: Queries Per Second (QPS), P99 latency, scalability, and fault-tolerance.

Approach:

• Spend ≤10 minutes to brainstorm and discuss requirements.

• Estimate numbers like daily active users and request rates.

• Ask about scalability: Should we start simple and scale up or design for scale immediately?

• Explore edge cases and ensure nothing is missed.

"Everything is a tradeoff. Improving one aspect (like latency) may increase cost or complexity."

---

2. Draft the API Specification

• Quickly sketch out GET, POST, PUT, and DELETE endpoints.

• Define the request/response structure.

• Mention endpoints like /health (for monitoring), /signup, and /login but keep it concise.

Tip: Propose an API draft within 5 minutes to leave room for deeper discussions.

---

3. Design the Data Model

• Decide whether to design from scratch or use existing databases.

• Avoid shared databases across services to prevent schema conflicts and resource contention.

• Key considerations:

  • Table schema.

  • Data access patterns (CRUD operations).

  • Techniques for preventing concurrent update conflicts.

---

4. Discuss Failures, Logging, and Monitoring

Failure Design

Explore graceful degradation and fault-tolerance:

• Silent failures, retry strategies, and monitoring for failures.

Observability

Logging, monitoring, and alerting ensure system health:

• Golden Signals:

  1. Latency

  2. Traffic, measured in HTTP requests per second

  3. Errors

  4. Saturation, depending on whether our system’s constraint is CPU, memory, or I/O, we can set utilization targets that should not be exceeded

• Use structured logs and standardized time formats. Avoid logging sensitive information (PII). Some useful things which can be tracked:

  • Contain a unique identifier to trace requests across services and share between users and developers.

  • Timestamps should use the same time zone and time format.

  • Should contain a unique identifier to trace requests across services .

  • Categorize log entries. Start with debug, info, and error.

Tools: Prometheus, Grafana, ELK (Elasticsearch, Logstash, Kibana).

---

5. Explore Search Implementation

Search is a common requirement for most systems. Three techniques include:

1. SQL + LIKE queries (not scalable).

2. Client-side libraries (for small datasets).

3. Search Engines like Elasticsearch (recommended for PB-scale text data).

Steps to Add Elasticsearch

1. Create indices for the documents.

2. Use the Bulk API to keep indices updated.

3. Handle fuzzy searches using Elasticsearch's query DSL.

---

6. Time Management and Driving the Discussion

• Focus on what matters most: requirements, API design, data model, and system components.

• Use diagrams (high-level and low-level) as visual aids.

• Keep the interviewer engaged by suggesting directions to explore further.

Remember: A 50-minute interview cannot cover everything. Communicate breadth and depth effectively.

---

Post-Interview Reflection and Self-Assessment

Write a Reflection

Immediately after the interview:

1. Document everything:

   • What was asked.

   • What you said.

   • Diagrams drawn.

2. Assess what went well and what didn’t.

Analyze Your Performance

• Did you miss any details?

• Were there any gaps in communication?

• Were you able to manage time effectively?

Research and Improve

• Compare your design to existing resources:

  • High Scalability Blog

  • YouTube discussions

  • System design books

• Identify knowledge gaps and refine your approach.

---

Interviewing the Company

System design interviews are two-way. You are also evaluating the company:

• What’s the technology stack?

• How competent is the engineering leadership?

• Are their development and CI/CD practices solid?

• What opportunities exist for learning and growth?

---

Summary: Principles to Follow

1. Clarify requirements and tradeoffs early.

2. Time-box discussions: API specification, data model, failures, and observability.

3. Use diagrams to illustrate designs effectively.

4. Cover logging, monitoring, alerting, and observability.

5. Reflect and self-assess after interviews to improve.

6. Interview the company to ensure it aligns with your goals.

---

Additional Resources:

• High Scalability Blog

• Google SRE Book: Four Golden Signals

• Elasticsearch Guide