RETURN TO SYSTEM DESIGN VAULT (55)BLUEPRINT #04 // Medium
#04MediumMedia Streaming & Storage

Design Dropbox or Google Drive

ASKED IN INTERVIEWS AT:DropboxMetaGoogleAmazon

EXECUTIVE SUMMARY & PROBLEM STATEMENT

Design a cloud file synchronization service that supports multi-device chunked uploads, delta sync, deduplication, and cross-client real-time synchronization.

SCALE ESTIMATION & TRAFFIC ENVELOPE

DAILY ACTIVE USERS / TPS100 Million DAU
READ / WRITE TRAFFIC RATIO1:1 balanced sync operations
5-YEAR STORAGE CAPACITYHundreds of Petabytes across distributed chunk stores

FUNCTIONAL REQUIREMENTS

  • [1]Upload, download, and sync files across devices
  • [2]Automatic delta sync (only upload modified chunks)
  • [3]File versioning & revision history

NON-FUNCTIONAL REQUIREMENTS

  • •100% data durability (11 9s)
  • •Bandwidth-efficient sync
  • •Conflict resolution on concurrent writes

CORE ARCHITECTURAL COMPONENTS

Client Desktop Agent
Block Chunker
Chunk Storage Service (S3)
Metadata Database (Spanner/PostgreSQL)
Notification Service (WebSockets)

DATA SCHEMA & PERSISTENCE

FileMetadata (file_id, user_id, name, version), FileChunks (chunk_hash, byte_size, storage_path), FileChunkMap (file_id, version, chunk_index, chunk_hash)

DISTRIBUTED TRADE-OFFS & FAILURE MODES

  • ⟡Client-side chunking reduces network bandwidth at the cost of client CPU.