Scaling
Instance sizing and scaling considerations.
Each SkyKoi user gets a dedicated EC2 instance. The current default is a t3.small (2 vCPU, 2 GB RAM), which handles most workloads well.
Current Architecture
SkyKoi uses a simple 1:1 model — one EC2 instance per user. This prioritizes isolation and simplicity over density.
- •t3.small — 2 vCPU, 2 GB RAM. Sufficient for the Koi Gateway, nginx, and tool execution.
- •Burstable performance — t3 instances earn CPU credits during idle periods and burst during active use.
- •Local workspace — All data stored on the instance's EBS volume.
Future Scaling Options (Coming Soon)
- •Instance size upgrades — Larger instances (t3.medium, t3.large) for users with heavy workloads
- •Auto-start/stop — Hibernate instances during inactive periods to reduce costs
- •Multi-region — Deploy instances in regions closer to users
- •Workspace backup — S3-based workspace backup and restore
💡The 1:1 instance model ensures strong isolation and predictable performance. Scaling optimizations are on the roadmap as the platform grows.