Reliability & Deployment
How the Loop Platform ensures reliability through staged deployments and operational practices.
Who this is for
Technical buyers evaluating the platform’s operational maturity, or developers who want to understand the deployment lifecycle.
Deployment Stages
The platform uses three permanent stages and ephemeral per-PR previews:
| Stage | Purpose | Deployment |
|---|---|---|
dev | Development and integration testing | Manual or on push |
staging | Pre-production validation | Auto-deploy on merge to main |
prod | Production traffic | Manual release from tagged version |
Infrastructure
- Compute — AWS Fargate (via SST). No Lambda cold starts.
- Database — Aurora Serverless v2 PostgreSQL. Auto-scaling.
- Events — AWS EventBridge. At-least-once delivery with deduplication.
- Observability — OpenTelemetry traces and logs shipped to Axiom.
SLA Commitments
Each service declares its SLA in service.yaml:
- Availability target — Percentage uptime commitment.
- Latency P95 — 95th percentile response time target.
Rollback
Every service supports immediate rollback to the previous deployment. Rollback procedures are documented in per-service runbooks and can be triggered via the deployment pipeline.