Back to Library
gcp
System Design
intermediate
Real-time search suggestions

Search Autocomplete System

System Design Classic

Search autocomplete suggests query completions as users type, requiring sub-100ms response times and frequency-based ranking. This GCP-native design uses a trie (prefix tree) stored in Memorystore (Redis) for fast prefix lookups, with Firestore tracking query frequencies. Perfect for search teams needing sub-100ms typeahead suggestions with frequency-based ranking and user-specific personalization.

Data Flow

Autocomplete API
Search Log Stream
Suggestion Lookup
Frequency Aggregator
Memorystore (Redis)
Query Frequencies
Full-Text Fallback

Share this architecture with your network

Service Breakdown (7 services)

Other7 services
Autocomplete API
  • •Routes API traffic and enforces policies
  • •Manages authentication and rate limiting
  • •Provides a unified API endpoint
Suggestion Lookup
  • •Runs event-driven code without servers
  • •Scales instantly from zero to peak load
  • •Cost-effective for sporadic workloads
Memorystore (Redis)
  • •Caches data in-memory with sub-millisecond latency
  • •Supports Redis protocol for broad compatibility
  • •Scales vertically without downtime
Query Frequencies
  • •Tracks how often each query prefix is searched
  • •Updates frequency counts from aggregated logs
  • •Drives popularity-based suggestion ranking
Search Log Stream
  • •Delivers messages between decoupled services reliably
  • •Supports millions of messages per second
  • •Guarantees at-least-once delivery to all subscribers
Frequency Aggregator
  • •Runs event-driven code without servers
  • •Scales instantly from zero to peak load
  • •Cost-effective for sporadic workloads
Full-Text Fallback
  • •Runs stateless containers with auto-scaling to zero
  • •Handles HTTPS requests with managed SSL
  • •Scales instantly from zero to thousands of instances

Scaling Strategy

The trie is partitioned by prefix ranges across Memorystore shards for distributed lookups. Each keystroke triggers a Cloud Function that retrieves suggestions from the appropriate shard. Query frequency updates flow through Pub/Sub and are aggregated by Cloud Functions into Firestore, then periodically synced back to the trie. A Cloud Run service handles full-text fallback for long queries.

Related Architectures

Multi-Tenant SaaS Platform

Generic SaaS

intermediate
System Design

Notification System

System Design Classic

intermediate
System Design

Dropbox File Storage System

Dropbox

intermediate
System Design

Chat Application System

System Design Classic

intermediate
System Design

News Feed System

System Design Classic

intermediate
System Design

Instagram Photo Sharing System

Instagram / Meta

intermediate
System Design