What it is
Object storage stores blobs such as images, videos, documents, backups, and model artifacts. Each object is addressed by a key and stored with metadata.
Why it matters
Large files do not belong in application databases. Object stores provide durable, cheap, scalable storage with lifecycle rules and integration with CDNs and processing pipelines.
How it works
Applications upload objects directly or through signed URLs. Metadata lives in a database, while the blob lives in storage. Background jobs scan, transform, or replicate objects.
When to use it
Use object storage for user uploads, media, logs, exports, backups, static assets, and large generated artifacts.
Tradeoffs / failure modes
Object stores are not relational databases. Listing can be expensive, overwrites need care, and authorization must protect signed URLs and object keys.
Tiny example
Google Drive stores file bytes in object storage and keeps permissions, names, folders, and version metadata in separate databases.
Related guides and systems
See File Upload Pipeline, Video Transcoding, CDN & Asset Delivery, Google Drive, Google Photos, YouTube, Netflix, and Instagram.
Seen in deeper dives
Capability
CDN & Asset Delivery
Origin shielding, cache hierarchies, purging strategies, and global content distribution
Capability
File Upload Pipeline
Chunked and resumable uploads, server-side processing, and CDN delivery at scale
Capability
Video Transcoding Pipeline
Adaptive bitrate encoding, queue-based processing, and HLS/DASH delivery for streaming
Case Study
Google Drive
Cloud file storage and real-time collaboration at planetary scale
Case Study
Google Photos
AI-powered photo storage, search, and memories for billions of images
Case Study
Photo and video sharing at massive scale with real-time feeds
Case Study
Netflix
Global video streaming with personalized content delivery
Case Study
YouTube
The world's largest video platform serving billions of hours daily
Related concepts
Bloom Filter
Quickly test whether an item is definitely absent or maybe present.
Cache-Aside
Let the application read through the cache and repopulate missing data on demand.
Eventual Consistency
Allow replicas to temporarily disagree while guaranteeing they converge later.