The Multi-Project Architecture: Running Your Entire Portfolio on One Server

*What if you could run 10+ production projects — across different tech stacks — on a single affordable VPS with shared CI, email, and DNS? Here's the architecture.*

---

The Portfolio Problem

If you're a technical founder building multiple products, each project needs: hosting, CI/CD, a database, email, DNS, and SSL certificates. Multiply by 10 projects and you're managing 60+ services across a dozen providers.

Or you could put it all on one machine.

The Architecture

**Multiple projects** across any tech stack — Go, Rails, Next.js, Elixir, Python — all deploying to the same infrastructure:

Deployment

Every project follows the same pipeline:

```

git push → CI runs → Docker build → Push to registry → Deploy → Health check

```

CI runs on self-hosted runners (free — no hosted CI minutes billing). Deployment uses container orchestration with zero-downtime rolling updates.

A single reusable workflow handles all projects. Per-project config is just: service name, port, health check path, and optional migration command.

DNS & Email

Each project gets:

- **Subdomain**: `project.yourdomain.com` (wildcard A record)

- **Custom domain** (optional): pointed to the same server IP

- **Email**: Self-hosted mail server with per-project DKIM, SPF, DMARC — auto-configured

Database

- **PostgreSQL** with pgvector for AI embeddings

- Each project gets its own database on the shared instance

- High-traffic projects can be split to dedicated containers when needed

Monitoring

One dashboard shows health across all projects: deployment metrics, CI pipeline status, runner health, resource usage.

The Numbers

| Metric | Value |

|--------|-------|

| Monthly infrastructure cost | ~$50 |

| Projects hosted | 10+ |

| Custom domains | Unlimited |

| Email domains | Auto-configured per project |

| CI runners | Shared pool (self-hosted) |

| Average deploy time | < 3 minutes |

**Cost per project**: under $5/month. Compare that to $20-50/month per project on managed platforms.

What Makes This Sustainable

**Shared infrastructure, isolated projects.** Container orchestration provides service isolation. Each project runs in its own container with defined memory limits. A misbehaving project can't take down the others.

**AI-managed operations.** CI failures auto-diagnose. Deployments verify themselves. Email DNS configures through tool calls. The platform reduces operational overhead to near zero.

**Incremental onboarding.** New projects onboard in minutes: register, configure CI workflow, first deploy. No new infrastructure to provision.

When to Scale Out

One server handles 10+ projects because most are moderate-traffic. When a project grows:

1. **Vertical scaling**: Upgrade the server (cheap, immediate)

2. **Horizontal split**: Move high-traffic projects to dedicated nodes

3. **Hybrid**: Keep management on the hub, deploy compute-heavy projects to separate machines

The architecture supports all three without re-architecting.

Who This Is For

- **Solo founders** with a portfolio of products

- **Agencies** managing client projects

- **Small teams** who want one infrastructure to rule them all

- **Anyone** tired of paying per-project for commodity infrastructure

The economics are simple: one server replacing a dozen managed services pays for itself in the first month.

---

*FlukeBase powers multi-project portfolios on commodity hardware with AI-native orchestration. [See how at flukebase.me](https://flukebase.me)*