Solution Architecture
Over the last several write-ups I deployed four open-source tools in my homelab, each solving one slice of how a company runs: Active Directory for identity, Faveo Helpdesk for IT support, OrangeHRM for human resources, and Mailcow for email. Taken together, on a single Proxmox host, they cover nearly everything a small business needs to operate — at a fraction of the cost of the commercial SaaS equivalents, and with the data kept in-house.
This is the architecture view: how the pieces fit, how identity ties them into one system, what it costs versus the alternative, and the trade-offs a business should weigh before going this route. It is the same instinct behind every one of these projects — start from a real problem and test whether open source genuinely solves it.
Strip away the jargon and every small business needs the same handful of things: a way to know who its people are and what they can access, a way to handle support requests, a way to manage employees, and a way to communicate. The default answer is to buy each of these as a separate SaaS subscription — one vendor for identity, another for the helpdesk, another for HR, another for email.
That works, but it adds up. Costs are per user, per month, per product, and they only grow. Just as important, the company data ends up scattered across four vendor clouds, each with its own account model, its own security posture, and its own lock-in. The architecture question is whether a small business can cover the same ground on infrastructure it owns — and the answer, with mature open-source tools, is yes.
Each tool maps to one pillar, and all of them run as VMs or containers on a single Proxmox node:
Three of the four (Faveo, OrangeHRM, Mailcow) deploy in well under an hour each — two as TurnKey appliances, one as a Docker stack — and the domain controller is a standard Windows install. The whole stack fits comfortably on one modest server.
Four apps on one host is convenient, but it is not yet a system — that only happens when they share an identity. This is why Active Directory goes in first and why it sits at the center of the diagram. The goal is for Faveo, OrangeHRM, and Mailcow to authenticate against AD over LDAP, so every person has exactly one account.
That single change is what turns a collection of tools into a platform. Onboarding becomes one action: create the account in AD, add the right group, and the new hire can log into the helpdesk, the HR portal, and their mailbox. Offboarding becomes one action too: disable the account and every door closes at once. No orphaned logins, no app-by-app cleanup, no wondering whether someone still has access somewhere. I am framing this as the roadmap rather than something already wired up — each app currently uses local accounts — but the design was built around it from the start.
The financial case is what makes this more than a hobby. Commercial equivalents — a hosted identity provider, a SaaS helpdesk, an HR platform, and business email — are each billed per user per month, and a small team can easily run into hundreds of dollars monthly once you add them up, climbing with every hire. The open-source stack replaces those recurring per-seat fees with one server and the time to run it.
The cost shifts from a growing subscription line to a largely fixed one: hardware (or a single VPS), a domain name, and the hours spent maintaining it. For a business that is cost-sensitive and willing to own a bit of operational work, that trade is compelling — and it gets better, not worse, as the headcount grows.
Self-hosting is a real commitment, and an honest architecture review names the costs. When you run this stack, you become the IT department: patching every system, managing TLS certificates, taking and testing backups, securing internet-facing services, and being the one who gets the call when mail stops flowing. Email in particular is unforgiving, and a domain controller is a high-value target that has to be protected.
So this is not the right answer for everyone. A business with no technical capacity and no appetite to build it is usually better served by SaaS. But for one with some in-house skill — or a willingness to grow it — the payoff is lower cost, full data ownership, and a genuine understanding of how its own systems work. That is exactly the kind of problem-to-solution evaluation I enjoy: not just whether something is technically interesting, but whether it actually fits.
The foundation is in place; the work ahead is what turns four deployments into a polished platform:
Identity, IT support, HR, and email — four problems every company has, solved with four open-source tools on one Proxmox host. Individually each is a useful project; together they are proof that a small business can run a real, professional operations stack without a stack of SaaS invoices. The tooling is mature, the cost is low, and the only real price is the willingness to own it. For a solutions-minded builder, that is a trade worth making — and a problem genuinely solved.