Junior Case Studies Interview Questions (50 MCQs)

Question 1

In a system design interview, what is usually the right first step?

Accepted Answer

Clarifying requirements and scope by asking questions

Answer

Immediately drawing a full diagram with every service

Answer

Picking a database and writing the table schema

Answer

Starting to write code for the main endpoints

Question 2

Which item is a functional requirement for a URL shortener?

Accepted Answer

A user can turn a long URL into a short link

Answer

The system must be available 99.99% of the time

Answer

Redirect latency must stay under 50 milliseconds

Answer

The system must handle 100 million links per month

Question 3

Which item is an example of a non-functional requirement?

Accepted Answer

Feed load p99 latency must be under 200ms

Answer

A user can publish a new post

Answer

A user can follow other users

Answer

A user can like posts

Question 4

With 8.6 million requests per day, what is the approximate average QPS?

Accepted Answer

About 100 requests per second

Answer

About 10 requests per second

Answer

About 1,000 requests per second

Answer

About 10,000 requests per second

Question 5

What is the typical traffic pattern of a URL shortener?

Accepted Answer

Read-heavy; redirects far outnumber link creations

Answer

Write-heavy; link creation exceeds redirects

Answer

Reads and writes are almost always equal

Answer

Most traffic comes from deleting links

Question 6

Why is base62 used for the short code in a URL shortener?

Accepted Answer

It only uses `[0-9a-zA-Z]`, which is URL-safe

Answer

Its output is encrypted and impossible to guess

Answer

It always makes the output exactly 6 characters

Answer

It fully prevents hash collisions

Question 7

Roughly how many combinations does a 7-character base62 code have?

Accepted Answer

About 3.5 trillion

Answer

About 3.5 million

Answer

About 3.5 billion

Answer

About 80 million

Question 8

What is the main difference between `301` and `302` redirects in a URL shortener?

Accepted Answer

`301` is permanent and browser-cached; `302` is temporary

Answer

`302` is permanent and browser-cached; `301` is temporary

Answer

`301` is only for HTTPS and `302` only for HTTP

Answer

Both are identical except for the status number

Question 9

If we create the short code by taking the first 7 chars of an MD5 of the URL, what is the main issue?

Accepted Answer

Collisions are possible and must be detected and resolved

Answer

MD5 gives different output for the same input

Answer

MD5 output contains only digits

Answer

MD5 does not work for long strings

Question 10

What does a Key Generation Service (KGS) do in a URL shortener?

Accepted Answer

Pre-generates and stores unique keys to hand out on demand

Answer

Encrypts the long URL so it cannot be read

Answer

Counts the number of clicks for each link

Answer

Deletes expired links from the database

Question 11

What parts does a Snowflake ID typically consist of?

Accepted Answer

Timestamp, machine/worker ID, and a sequence number

Answer

URL hash, username, and a random number

Answer

Server IP address, port, and process number

Answer

A version 4 UUID plus a checksum

Question 12

What is one downside of UUID v4 as a database primary key?

Accepted Answer

It is random and 128-bit, which hurts ordered index locality

Answer

Generating it requires a central server

Answer

Its collision probability is very high

Answer

It can only be generated on a single machine

Question 13

What is the problem with using a single database's auto-increment for IDs in a distributed system?

Accepted Answer

It becomes a single point of failure and write bottleneck

Answer

The generated IDs will be duplicated

Answer

The IDs have no time ordering

Answer

The IDs grow longer than 128 bits

Question 14

In a pastebin design, where is paste content usually stored?

Accepted Answer

In object storage like S3, with metadata in a database

Answer

Only in the application server's RAM

Answer

Inside the user's browser cookie

Answer

In the web server's log file

Question 15

Which approach is common and simple for removing expired pastes?

Accepted Answer

A periodic cleanup job plus an expiry check on read

Answer

Manual deletion by an admin every few months

Answer

Restarting the database every night

Answer

Closing connections of users with old pastes

Question 16

What is the main purpose of a rate limiter in an API?

Accepted Answer

Limiting how many requests each client makes per time window

Answer

Compressing responses to reduce bandwidth

Answer

Encrypting traffic between client and server

Answer

Distributing requests across different servers

Question 17

When a rate limiter rejects a request, which HTTP status is appropriate?

Accepted Answer

`429 Too Many Requests`

Answer

`403 Forbidden`

Answer

`503 Service Unavailable`

Answer

`400 Bad Request`

Question 18

How does the token bucket algorithm work?

Accepted Answer

Tokens are added at a fixed rate and each request consumes one

Answer

Requests are queued and drained at a fixed rate

Answer

Only requests per calendar minute are counted

Answer

Exact timestamps of all requests are stored in a log

Question 19

What is the well-known problem with the fixed window counter algorithm?

Accepted Answer

A burst at a window boundary can let through up to twice the limit

Answer

It needs memory proportional to the number of requests

Answer

It never allows any burst

Answer

It can only be implemented on one server

Question 20

What is commonly used to store rate limiter counters across multiple servers?

Accepted Answer

A shared in-memory store like Redis

Answer

A local variable in each server's memory

Answer

A table in an analytics data warehouse

Answer

A text file on each server's disk

Question 21

In a news feed, what does 'fan-out on write' mean?

Accepted Answer

On publishing, the post is written into every follower's feed cache

Answer

On opening the feed, posts from all followees are gathered

Answer

Posts are stored only in the author's database

Answer

Every post is sent to all users in the system

Question 22

What is the main advantage of 'fan-out on read' (the pull model)?

Accepted Answer

Writes are cheap and no work is wasted on inactive users

Answer

Reading the feed is always as fast as possible

Answer

There is no need to store posts in a database

Answer

Post ordering is guaranteed automatically

Question 23

What does the 'celebrity problem' in a news feed refer to?

Accepted Answer

Fan-out on write for a user with millions of followers is very expensive

Answer

Celebrities delete posts more often than others

Answer

Celebrity posts require manual moderation

Answer

Celebrities cannot follow anyone

Question 24

Which pagination approach suits a feed that constantly receives new posts?

Accepted Answer

Cursor-based pagination using the last item's ID or timestamp

Answer

Offset pagination with `LIMIT` and `OFFSET`

Answer

Loading the entire feed in a single request

Answer

Pagination by random page number

Question 25

Why is WebSocket commonly used for a chat application?

Accepted Answer

A persistent bidirectional connection lets the server push messages instantly

Answer

Because WebSocket automatically end-to-end encrypts messages

Answer

Because WebSocket does not need a server

Answer

Because WebSocket persists messages to disk

Question 26

How does long polling work?

Accepted Answer

The server holds the request open until new data arrives or it times out

Answer

The client sends a new request every second and gets an immediate reply

Answer

The server opens a raw TCP connection to the client

Answer

The client receives data via email

Question 27

How is user online/offline status (presence) usually detected?

Accepted Answer

With periodic heartbeats; if none arrive for a while the user is offline

Answer

By checking the user's IP address every hour

Answer

By sending an email and waiting for a reply

Answer

By counting messages sent per day

Question 28

What mechanism are 'sent / delivered / read' ticks in a chat app built on?

Accepted Answer

Acknowledgments from the server and recipient device for each stage

Answer

Guessing based on time elapsed since sending

Answer

Checking the recipient phone's battery status

Answer

Counting the message's characters

Question 29

If a chat message recipient is offline, what should the system do?

Accepted Answer

Store it and deliver on reconnect or via push notification

Answer

Drop the message and return an error to the sender

Answer

Keep it in the sender's RAM until the recipient is online

Answer

Send the message to all online users

Question 30

Which set lists common channels of a notification system?

Accepted Answer

Mobile push, SMS, email, and in-app

Answer

DNS, FTP, SSH, and Telnet

Answer

CPU, RAM, disk, and network

Answer

REST, GraphQL, gRPC, and SOAP

Question 31

Why is a message queue placed between services and providers in a notification system?

Accepted Answer

For decoupling, absorbing bursts, and retrying on provider failures

Answer

So notifications are encrypted faster

Answer

To remove the need for a user database

Answer

So each notification is sent only once a month

Question 32

In a web crawler, what is the 'URL frontier'?

Accepted Answer

The structure holding URLs waiting to be crawled

Answer

The list of sites that are forbidden to crawl

Answer

The cache of downloaded HTML pages

Answer

The crawler's dedicated DNS server

Question 33

Which graph traversal does a web crawler typically use to explore the web?

Accepted Answer

BFS, because DFS can get stuck deep within one site

Answer

DFS, because it is always the fastest traversal

Answer

Dijkstra, because links have negative weights

Answer

Random traversal with no queue at all

Question 34

Which file should a polite crawler respect?

Accepted Answer

`robots.txt`

Answer

`package.json`

Answer

`.htaccess`

Answer

`manifest.json`

Question 35

Why use a Bloom filter in a crawler to check seen URLs?

Accepted Answer

It uses very little memory; it has false positives but no false negatives

Answer

It is never wrong and behaves exactly like a hash set

Answer

It can delete and sort URLs

Answer

It has false negatives but no false positives

Question 36

Which data structure fits prefix-based search autocomplete?

Accepted Answer

Trie (prefix tree)

Answer

Stack

Answer

A plain min-heap

Answer

Linked list

Question 37

What is a common client-side way to reduce autocomplete requests?

Accepted Answer

Debouncing input and caching results per prefix

Answer

Sending a request for every key including Shift

Answer

Disabling autocomplete for words under 10 letters

Answer

Sending the full search history with every request

Question 38

Why is consistent hashing used in a distributed key-value store?

Accepted Answer

Adding/removing a node moves only a small fraction of keys

Answer

It guarantees all keys stay on a single node

Answer

It automatically encrypts the data

Answer

It completely removes the need for replication

Question 39

Why is each key replicated on several nodes in a distributed key-value store?

Accepted Answer

For availability and durability when a node fails

Answer

To reduce the storage space used

Answer

So that writes become faster than reads

Answer

To remove the need for networking between nodes

Question 40

Why are uploaded videos transcoded in a streaming service?

Accepted Answer

To produce versions at different resolutions, bitrates, and formats

Answer

To strip the audio and reduce size

Answer

To avoid storing the video on a CDN

Answer

So the video only plays on desktop

Question 41

What does adaptive bitrate streaming (like HLS or DASH) do?

Accepted Answer

The player switches segment quality based on current bandwidth

Answer

Downloads the whole video at once before playback

Answer

Randomly changes the video bitrate on the server

Answer

Only adapts the video's audio

Question 42

What is the role of a CDN in a video streaming service?

Accepted Answer

Serving video segments from edges close to the user

Answer

Transcoding videos during upload

Answer

Storing user account info and passwords

Answer

Running the recommendation algorithm

Question 43

Why do services like Dropbox split files into small chunks?

Accepted Answer

For resumable uploads and sending only changed chunks

Answer

To make the files more human-readable

Answer

To remove the need for metadata

Answer

So each file lives on only one disk

Question 44

How is deduplication usually done in file storage?

Accepted Answer

By hashing each chunk's content; a chunk with an existing hash is not stored again

Answer

By comparing file names

Answer

By comparing file creation dates

Answer

By checking file extensions

Question 45

What key property of geohash makes it useful for location search?

Accepted Answer

Nearby points usually share a common prefix

Answer

Each city has exactly one geohash

Answer

Geohash length always equals distance from the equator

Answer

Geohash securely encrypts coordinates

Question 46

How does a quadtree divide space in a proximity service?

Accepted Answer

It splits each node into four quadrants until each leaf has few points

Answer

It splits space into equal-sized hexagons

Answer

It sorts points by place name

Answer

It only indexes latitude coordinates

Question 47

Which Redis structure is suitable for a real-time leaderboard?

Accepted Answer

Sorted Set (`ZADD`, `ZREVRANGE`, `ZREVRANK`)

Answer

List (`LPUSH`, `LRANGE`)

Answer

String (`SET`, `GET`)

Answer

Pub/Sub (`PUBLISH`, `SUBSCRIBE`)

Question 48

Why is an idempotency key used in a payment API?

Accepted Answer

So retrying a request does not cause a double charge

Answer

So the payment amount is encrypted

Answer

So the payment is processed faster

Answer

So user authentication is not needed

Question 49

In a ticket booking system, what does 'oversell' mean?

Accepted Answer

Selling more tickets than real capacity due to race conditions

Answer

Selling tickets above the official price

Answer

Selling all tickets in the first few seconds

Answer

Selling tickets to unregistered users

Question 50

What kind of database best fits storing monitoring system metrics?

Accepted Answer

A time-series database like Prometheus or InfluxDB

Answer

A graph database like Neo4j

Answer

A document store for Word files

Answer

A CSV file on each server