​
Isofold Deployment Topology

Isofold is designed to run efficiently in cloud, edge, and self-hosted environments. This page describes how traffic flows through the system and how components interact.

​
Request Flow

At a high level, a typical request path looks like this: 
Client → Isofold Proxy → Rewrite Engine → Cost/Verify Layer → Data Warehouse → Response
Each layer can be tuned or observed independently depending on deployment mode.

​
Hosted Mode (Isofold Cloud)

In hosted mode, traffic flows through an edge-deployed proxy (e.g. Fly.io) that routes to the appropriate warehouse:
  • Minimal added latency (typically < 10ms)
  • Global traffic routed to nearest edge
  • Internal routing and queueing within Isofold’s infrastructure
[ Client ]
    ↓
[ Fly Edge Proxy ]
    ↓
[ Rewrite Engine + Cost Model ]
    ↓
[ BigQuery / Snowflake / Aurora ]
Each tenant has an isolated proxy URL and scoped configuration.

​
Self-Hosted Mode (VPC / Local)

When running inside your own network:
  • The proxy is deployed in your VPC or on-prem
  • Rewrite + verification happens inside the container
  • You control outbound access, logging, and observability
[ Client ]
    ↓
[ Local Isofold Proxy ]
    ↓
[ Internal Warehouse (e.g. Aurora in VPC) ]
This model is ideal for:
  • Compliance-sensitive workloads
  • Low-latency, high-throughput environments
  • Custom warehouse integrations

​
Multi-region Support

When deployed with Fly.io or GCP global load balancers:
  • Requests are routed to the closest live instance
  • All components are stateless and horizontally scalable
  • Proxy-level metrics can be pushed to Prometheus, GCP, or Datadog
Next: Learn more about Security & Compliance