# Deployment

## Runs inside your environment

GraphReplica deploys as a container in your environment, on-prem or in your VPC. Raw data never leaves your data lake. You stay in control and your security review moves faster.

### Key Metrics

- **Raw data shared with Secludy:** 0%
- **Typical setup time:** ~1 hr
- **Records held consistent:** 100M+
- **Run modes supported:** 3

Your data never leaves.

## See the difference self-hosting makes

Toggle between a typical SaaS vendor and Secludy self-hosted. Watch where your data goes and what crosses the boundary of your environment.

| Typical SaaS vendor | Secludy self-hosted |
|----------------------|---------------------|
| Data stays in your environment | Your VPC |
| Your data lake | Sensitive raw records |
| Secludy container | Runs in place, self-managed |
| Environment boundary | Only safe synthetic data and an audit report leave |
| Deployment mode | On-prem / VPC / AWS Marketplace |
| Run mode | Batch / Streaming / One-time |

### What you get

- Raw data stays inside Your VPC
- A leakage report and red-team results for every run
- Faster security and compliance review

## Self-hosted, on your terms

GraphReplica meets your infrastructure and your reviewers where they are.

### Runs in your environment

GraphReplica ships as a container that runs inside your environment. Deploy it on-prem, into your VPC, or from the AWS Marketplace into a VPC you own. Run it self-managed when your policy requires it.

Run where your data already lives. No raw data leaves your data lake.

### Fits your current stack

GraphReplica plugs into standard container and MLOps workflows. It maps onto the pipelines you already operate, so there are no changes to existing tooling and no new system to learn.

Ship faster with no rebuild and no vendor data movement.

### Evidence on every run

Every run produces a leakage report, the privacy parameters used, and red-team results. Hand that package straight to your GC, CISO, and risk committee.

Clear security and compliance review faster with proof in hand.

### Scales with Kubernetes

GraphReplica scales across GPU clusters with Kubernetes. It supports batch processing, real-time streaming, and one-time submissions, so it grows with your workload.

Hold consistency across 100M+ records as volume climbs.

### Raw data stays put

No sensitive data needs to leave your data lake to use Secludy. There is no offshore step and no external exposure.

### Self-managed option

Run the container fully self-managed when your security posture calls for it. You hold the keys and the controls.

### Live in about an hour

Setup takes about an hour. Your first safe dataset lands in under a week.

### Run where your data already lives

No vendor data movement. No external exposure. See it in your own environment.

## Deploy GraphReplica where your data lives

Stand it up in about an hour and ship your first safe dataset in under a week. Raw data stays inside your environment the whole time.