Video Transcoding Pipeline

This project is a backend-only video transcoding pipeline that handles video uploads, transcoding, and storage. It uses S3 for video storage, SQS for event-driven processing, and PostgreSQL for metadata and status tracking. The pipeline is designed to be scalable and efficient.

Architecture Overview

Workflow

1. Upload Server:

   • Accepts video uploads from clients.
   • Uploads the video to an S3 bucket (temp-video-bucket).

2. S3 (temp-video-bucket):

   • Stores the uploaded video.
   • Triggers an SQS event when a new video is uploaded.

3. Worker Server:

   • Listens to the SQS for new video upload events.
   • Adds the video metadata (bucket, key, status) to the database with status Pending.
   • Enqueues the video for transcoding in the task queue.

4. Transcoder:

   • Processes videos from the task queue.
   • Downloads the video from S3, transcodes it into multiple resolutions (480p, 720p, 1080p) using FFMPEG.
   • Uploads the output (.mp4 files) to a new S3 bucket (output-video-bucket).
   • Updates the database with the transcoded file locations and sets the status to Published.
   • Deletes the original video from the temp-video-bucket.

5. Database:

   • Stores video metadata, including:
     • id: Unique video ID.
     • bucket: S3 bucket for the raw video.
     • key: S3 key for the raw video.
     • status: Current status of the video (Pending, Processing, Published, Failed).
     • outputBucket: S3 bucket for the transcoded files.
     • outputKey: S3 key for the transcoded files (e.g., 480p.mp4, 720p.mp4, 1080p.mp4).
     • createdAt: Timestamp of video upload.
     • updatedAt: Timestamp of last status update.

Technologies Used

• AWS S3: For storing raw and transcoded video files.
• AWS SQS: For event-driven processing of video uploads and transcoding tasks.
• PostgreSQL: For storing video metadata and status.
• FFMPEG: For transcoding video into multiple resolutions.