Skip to main content
Recommended for: Production deployments with 50+ users

Architecture Overview

Teable uses a stateless application tier with externalized state services. This design enables horizontal scaling and high availability.

Components

ComponentPurposeManaged Service Examples
Teable AppStateless web applicationK8s, AWS ECS, Azure App Service, GCP Cloud Run
PostgreSQLPrimary data storageAWS RDS, Azure PostgreSQL, Google Cloud SQL
Redis (Queue)Background jobs & cacheElastiCache, Azure Cache, Memorystore
Redis (Perf)Performance cache for readsSame or separate Redis instance
Object StorageFiles & attachmentsAWS S3, MinIO, any S3-compatible
Key insight: The Teable application holds no persistent state. You can scale horizontally by adding pods, and any pod can handle any request.

Resource Recommendations

These are production recommendations from the Teable engineering team. For evaluation/testing, see Docker Quick Start.

Application Tier (Teable Pods)

SettingRecommended Value
CPU per pod2 vCPU
Memory per pod4 GB
Minimum replicas2
Auto-scale trigger50% CPU utilization
Use Kubernetes with Horizontal Pod Autoscaler (HPA), or equivalent auto-scaling on AWS ECS / Azure App Service / GCP Cloud Run.

Database (PostgreSQL)

SettingRecommended Value
CPU4 vCPU
Memory16 GB
Topology1 Master + 1 Replica
BackupDaily automated
Version15+ recommended
Always deploy with a replica for high availability. Use managed database services for automatic failover.

Cache (Redis)

Teable uses two separate Redis instances for different purposes:
PurposeEnvironment VariableCPUMemory
Queue & CacheBACKEND_CACHE_REDIS_URI1 vCPU2 GB
Performance CacheBACKEND_PERFORMANCE_CACHE2 vCPU4 GB
The Performance Cache (BACKEND_PERFORMANCE_CACHE) is optional but strongly recommended for multi-user collaboration. It significantly improves read performance in busy workspaces.

Object Storage

  • Type: S3-compatible (AWS S3, MinIO, etc.)
  • Buckets: 2 required (public + private)
  • Capacity: Scales automatically with managed S3
See Object Storage Guide for bucket configuration details.

Scaling Strategy

ComponentScaling TypeWhen to Scale
Teable AppAutomatic (horizontal)CPU > 50% (add pods)
PostgreSQLManual (vertical)CPU > 70% sustained
RedisManual (vertical)Memory > 80% or CPU > 60%
Application tier: Stateless design allows automatic horizontal scaling. Kubernetes HPA or cloud auto-scaling handles this automatically based on CPU metrics.Database & Redis: These are stateful services where vertical scaling (bigger instance) is simpler and safer than horizontal scaling. Monitor metrics and resize when thresholds are exceeded.

Minimum vs. Production Comparison

Minimum (PoC)Production
Teable App1 instance, 2c4g2+ pods, 2c4g each, auto-scaling
PostgreSQL2c4g, single instance4c16g, 1 master + 1 replica
Redis1 instance, 1c2g2 instances (queue + performance)
StorageS3-compatible object storageS3-compatible object storage
AvailabilityNone (single point of failure)High availability with redundancy
The “Minimum” setup is for evaluation only. It has no redundancy—data loss is possible if the server fails.

Next Steps: Choose Your Platform


Last modified on January 23, 2026