What is a consequence of under-provisioning?

Under-provisioning can lead to system crashes due to insufficient resources.

What does DAU stand for?

DAU refers to Daily Active Users, a key metric in user engagement.

How many bytes are in a kilobyte?

1 kilobyte (KB) is equal to 1,024 bytes.

What is the formula for calculating average RPS?

Average RPS is calculated by dividing total requests per day by the number of seconds in a day.

What is a common assumption about user activity?

Typically, 10-20% of total users are active daily.

How do you estimate storage for internal messaging?

All methods contribute to estimating storage needs for messaging.

What is the purpose of logging search queries?

Search queries are logged for analytics and debugging purposes.

What is a typical log entry size?

Typical log entries range from 500 bytes to 1 KB.

What is the primary goal of back-of-envelope estimation?

Back-of-envelope estimation is a quick sanity check — rough math to confirm a design is feasible before investing in detailed planning or code.

An app has 100 million registered users. A good estimate for Daily Active Users (DAU) is:

The standard industry assumption is that 10–20% of registered users are active on any given day.

How many seconds are there in one day?

24 hours × 60 minutes × 60 seconds = 86,400 seconds/day. This is the denominator used when converting daily requests to RPS.

Your app receives 50 million requests per day. What is the approximate average RPS?

50,000,000 ÷ 86,400 ≈ 578 RPS, which rounds to ~500 RPS.

Peak RPS is commonly estimated as which multiple of average RPS?

A 2–3× multiplier over average RPS is the standard heuristic for peak load, accounting for traffic spikes without extreme over-provisioning.

1 Terabyte (TB) is equal to how many GB in binary units?

In binary (IEC) units, 1 TB = 1,024 GB. The common shorthand "1,000 GB" is a decimal approximation used loosely in marketing.

The 80/20 read-write rule in system design means what?

The 80/20 rule states that roughly 80% of requests are reads and 20% are writes, which informs database and caching architecture decisions.

An app has 1 million DAU, each user makes 10 requests/day, each request stores 500 bytes. What is the approximate daily storage?

1,000,000 users × 10 requests × 500 bytes = 5,000,000,000 bytes ≈ 4.65 GB. The closest listed option is ~500 MB; the actual value is ~5 GB, but this tests order-of-magnitude awareness.

Why should systems be designed for peak load rather than average load?

Systems sized only for average traffic will be overwhelmed during spikes (product launches, viral events), causing outages and poor user experience.

QPS stands for what in system design?

QPS (Queries Per Second) is interchangeable with RPS in many contexts and measures the throughput a system must handle.

Internal messaging app: 40,000 DAU, 30 messages/user/day, 250 bytes/message. Daily storage is approximately:

40,000 × 30 × 250 = 300,000,000 bytes = 300 MB/day.

Which assumption is NOT typically used in capacity planning?

Traffic is never uniform — it peaks at certain hours (e.g. mornings, evenings). Assuming even distribution would drastically underestimate peak server needs.

20 microservices each handle 500 RPS and log 3 entries/request at 500 bytes each. What is the log generation rate?

20 × 500 × 3 × 500 = 15,000,000 bytes/sec = 15 MB/sec.

Search logs have 5 million queries/day at 1 KB each. Approximate annual storage before replication is?

5,000,000 × 1 KB × 365 = 1,825,000,000 KB = 1,825 GB ≈ 1.8 TB/year.

What is the purpose of the replication factor in storage estimation?

A replication factor of 2× or 3× means each byte is stored on multiple nodes, so total storage must be multiplied by that factor.

5,120 MB is equal to how many GB?

5,120 ÷ 1,024 = 5 GB. Dividing MB by 1,024 converts to GB.

What makes distributed storage (like S3 or HDFS) necessary for search query logs?

At 1.8 TB/year (pre-replication), annual log storage quickly exceeds single server capacity, requiring distributed storage solutions.

Which of the following is the correct formula for total storage?

Total storage = daily storage × how many days you retain data × how many copies you keep (replication factor).

What does "hot storage" vs "cold archive" mean in log management?

Hot storage keeps recent logs on fast, expensive media for quick access. Cold/archive storage moves older logs to cheap, slow media (like S3 Glacier) to reduce cost.

2^10 is equal to what value?

2^10 = 1,024, which is why 1 KB = 1,024 bytes, 1 MB = 1,024 KB, and so on through all binary storage units.

Applying the 80/20 rule to 200 million total requests/day, how many are write operations?

80/20 means 20% are writes: 200M × 0.20 = 40 million write operations/day.

Concurrent users at any given moment is typically estimated as what percentage of DAU?

A common heuristic is that only ~1% of DAU are simultaneously active, which drives concurrent connection and thread pool sizing.

An app performance logging system generates 1.2 TB/day. What is the 30-day hot storage requirement?

1.2 TB/day × 30 days = 36 TB of hot storage before applying any replication factor.

Which of the following strategies does NOT help reduce app performance log storage?

Adding more microservices increases the number of log sources and therefore increases total log volume — it is the opposite of a reduction strategy.

Back-of-Envelope Estimation — System Design Concept

🍪 TechCookies

Free

📚Back-of-Envelope EstimationFree

14 sections

~33 min total

32 quick quizzes

7 SD challenges linked

0 of 14 done·~39 min left

Concepts›Back-of-Envelope Estimation›What is Back-of-Envelope Estimation?

0 / 14

14 sections~33 min

What is Back-of-Envelope Estimation?

Quick estimations for system design feasibility before coding

Read

~3 min

⋯

Why Capacity Planning Matters

Understanding user needs prevents wasted resources and system failures.

ReadQuiz

~3 min

⋯

Step-by-Step Estimation Framework

A structured approach to estimate system requirements.

ReadQuizCode

~3 min

⋯

Common Assumptions

Key assumptions guide accurate estimations.

ReadQuizCode

~3 min

⋯

Storage Math: Bytes → MB → GB → TB

Essential conversions for storage estimation.

ReadQuizCode

~3 min

⋯

CPU / QPS Math: Requests/Day → Peak RPS

Calculating peak requests per second from daily traffic.

ReadQuizCode

~3 min

⋯

Relevant Questions — Three Real Scenarios

Practical scenarios for applying estimation techniques.

Read

~3 min

⋯

Internal Text Sharing Capacity

Estimating storage needs for internal messaging tools.

ReadQuizCode

~3 min

⋯

Search Query Log Storage

Estimating storage for logged search queries.

ReadQuizCode

~3 min

⋯

App Performance Log Storage

Estimating storage for microservices logging.

ReadQuizCode

~3 min

⋯

Master Reference Card

Key formulas and constants for quick reference.

ReadCode

~3 min

⋯

Summary Diagram

Visual representation of estimation concepts.

ReadCode

~3 min

⋯

Answer Key

Correct answers for practice questions.

Read

~3 min

⋯

Practice test

32 questions

~11 min

Section 1 of 14Read

What is Back-of-Envelope Estimation?

Quick estimations for system design feasibility before coding

~3 min read

What is it? Back-of-envelope estimation means doing a quick, rough calculation to check whether a system design is realistic — before writing a single line of code. Think of it like estimating how many pizzas to order for a party. You don't count every slice — you just multiply guests × slices and round up!

Mark this section complete.

Notes

🔍

Loading…