Back-of-the-envelope estimation in HLD - System Design Exercise

Design: Back-of-the-envelope Estimation for System Design

Focus on quick, approximate calculations for system scale and performance. Detailed design and implementation are out of scope.

Functional Requirements

FR1: Estimate system capacity quickly with limited data

FR2: Calculate rough resource needs (servers, storage, bandwidth)

FR3: Provide latency and throughput approximations

FR4: Support decision making in early design phases

Non-Functional Requirements

NFR1: Estimates must be done within minutes

NFR2: Accuracy within an order of magnitude is acceptable

NFR3: Use simple math and assumptions

NFR4: No detailed data or simulations required

Think Before You Design

Questions to Ask

❓ Question 1

❓ Question 2

❓ Question 3

❓ Question 4

❓ Question 5

Key Components

User load estimations

Request size and frequency

Network bandwidth

Storage capacity

Compute resources (CPU, memory)

Design Patterns

Load estimation

Capacity planning

Latency and throughput approximation

Resource scaling heuristics

Reference Architecture

No specific architecture diagram needed for estimation; focus on calculation steps.

Components

User Load Estimator

Manual calculation or spreadsheet

Estimate number of concurrent users and requests per second

Data Size Calculator

Manual calculation or spreadsheet

Estimate average data size per request and total data volume

Resource Estimator

Manual calculation or spreadsheet

Estimate CPU, memory, storage, and bandwidth needs

Request Flow

1. 1. Define expected number of users and requests per second.

2. 2. Estimate average data size per request.

3. 3. Calculate total data throughput (requests per second * data size).

4. 4. Estimate storage needs based on data retention policies.

5. 5. Approximate CPU and memory based on request complexity.

6. 6. Check network bandwidth requirements against available capacity.

7. 7. Adjust estimates for peak load and safety margin.

Database Schema

Not applicable for back-of-the-envelope estimation.

Scaling Discussion

Bottlenecks

Underestimating peak traffic leading to resource exhaustion

Ignoring variability in request size or complexity

Not accounting for network latency or failures

Overlooking data growth over time

Solutions

Include safety margins (e.g., 2x peak load) in estimates

Use conservative assumptions for request size and complexity

Plan for horizontal scaling and load balancing

Regularly update estimates with real usage data

Interview Tips

Time: Spend 5 minutes clarifying assumptions, 15 minutes performing rough calculations, 5 minutes discussing results and trade-offs, 5 minutes answering questions.

Clarify assumptions and constraints before estimating

Explain each step of the calculation clearly

Discuss trade-offs between accuracy and speed

Mention how estimates guide design decisions

Highlight importance of revisiting estimates with real data