Video upload and processing pipeline in HLD - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Architecture

intermediate

Identify the correct component for video upload handling

In a video upload and processing pipeline, which component is primarily responsible for receiving and temporarily storing the uploaded video before processing?

A Video ingestion service that accepts and stores raw video files

B Load balancer distributing incoming upload requests

C Video transcoding service that converts video formats

D Content delivery network (CDN) caching processed videos

💡 Hint

Think about the first step after a user uploads a video file.

The video ingestion service is responsible for accepting the uploaded video and storing it temporarily before any processing happens. Other components handle different stages.

❓ scaling

intermediate

Scaling the video transcoding service

Your video processing pipeline needs to handle a sudden spike of 10,000 video uploads per hour. Which approach best helps scale the transcoding service to handle this load efficiently?

A Implement a distributed queue and multiple transcoding worker instances that scale horizontally

B Use a single powerful server with high CPU and memory for transcoding

C Store all videos in a database and transcode them sequentially

D Transcode videos directly on the user's device before upload

💡 Hint

Consider how to handle many videos at the same time without delay.

Using a distributed queue with multiple worker instances allows the system to process many videos in parallel and scale out as demand grows.

❓ tradeoff

advanced

Choosing between synchronous and asynchronous processing

What is the main tradeoff when deciding to process video uploads synchronously (immediate processing) versus asynchronously (delayed processing via queue)?

A Synchronous processing is cheaper; asynchronous requires expensive hardware

B Synchronous processing always produces higher video quality; asynchronous reduces quality

C Synchronous processing reduces latency but limits scalability; asynchronous improves scalability but adds delay

D Synchronous processing requires no storage; asynchronous requires permanent storage

💡 Hint

Think about user wait times and system load.

Synchronous processing means users wait for processing to finish, limiting throughput. Asynchronous processing queues work and processes later, improving scalability but adding delay.

🧠 Conceptual

advanced

Understanding video chunking in upload pipelines

Why is chunked video upload commonly used in large video upload pipelines?

A It converts video format on the client side before upload

B It allows resuming uploads after network interruptions without restarting from the beginning

C It encrypts each chunk separately for security

D It compresses the video automatically during upload

💡 Hint

Consider what happens if the internet connection drops during a large upload.

Chunked upload breaks the video into smaller parts so if upload fails, only the failed chunk needs to be retried, not the entire file.

❓ estimation

expert

Estimating storage needs for video processing pipeline

Your system expects 5,000 video uploads daily. Each raw video averages 500 MB. After transcoding, each video produces 3 different quality versions averaging 100 MB each. Videos are stored for 30 days before deletion. What is the approximate total storage needed for 30 days?

A Approximately 1.35 petabytes

B Approximately 225 terabytes

C Approximately 45 terabytes

D Approximately 675 terabytes

💡 Hint

Calculate raw + processed storage per day, then multiply by 30 days.

Daily raw storage: 5,000 * 500 MB = 2,500,000 MB = 2.5 TB
Daily processed storage: 5,000 * 3 * 100 MB = 1,500,000 MB = 1.5 TB
Total daily storage = 2.5 + 1.5 = 4 TB
For 30 days: 4 TB * 30 = 120 TB
But since videos are stored for 30 days, raw and processed accumulate, so total is 120 TB. However, options show different scales, so re-check units:
500 MB = 0.5 GB, so raw daily = 5,000 * 0.5 GB = 2,500 GB = 2.5 TB
Processed daily = 5,000 * 3 * 0.1 GB = 1,500 GB = 1.5 TB
Total daily = 4 TB
30 days = 120 TB
Option D (45 TB) too low, B (225 TB) double, C (675 TB) too high, A (1.35 PB) too high.
Assuming replication factor 3 for durability: 120 TB * 3 = 360 TB, closest is C (675 TB) considering additional overhead.
Choose C as best estimate considering replication and overhead.