How I Built a Private Cloud on ManageIQ: Experience, Pain, and Architecture

Introduction

At the end of 2021, I was offered a project to build a private cloud. The key interview question was simple: “Do you know Ruby on Rails?” — I did. I didn’t think long before saying yes.

Phase One: The Curious Beginning

From day one, we started working with an open-source solution from Red Hat — ManageIQ. The initial team was small: a system administrator, an architect, and a developer. We studied the system architecture, dug into the code, frameworks, and built-in features. I proposed creating a client-facing management panel — and the idea was approved.

I built an MVP of the portal, the integrator configured the interaction with VMware, and after a successful demo, the project got the green light. We hired a frontend developer and continued developing the product towards production.

Import Substitution and Organizational Struggles

When sanctions hit, we had to test a domestic virtualization system (oVirt). Meanwhile, the team was growing — but without proper technical screening. Some hires quickly left.

In summer 2023, a new project manager joined. His management style was purely formal: oversight without involvement, tasks with no structure, and communication without content. Jira took months to launch, and tasks came without description or logic.

Team Formation and Infrastructure Growth

The team expanded with people of various engagement levels. Some ignored feedback, others dodged responsibility. The sysadmin left, while the manager continued building the illusion of progress. A newly hired "architect" was a token figure, promised a CTO title but without decisions or accountability.

Meanwhile, the infrastructure kept evolving. I maintained and expanded these components:

• 2 ManageIQ instances (prod and pre-prod)
• Integration with three oVirt providers and one vCenter
• Support for over 5,000 virtual machines
• Automation Engine, LDAP, AWX, and Netbox integrations
• Custom client portal built on Laravel + Vue.js with REST interfaces
• Use of Service Dialogs, custom forms, and catalogs

The work continued, but team dynamics degraded.

Contract Ends — What’s Left?

By the end of the contract, less than half of the original requirements had been met. Team motivation dropped, tasks remained unfinished, and Jira reports replaced real progress. Some team members started sharing informal insights — including the manager’s favorite book, The 7 Habits of Highly Effective People.

New Year, Old Problems

After a New Year break, conflict became a management tactic. One example: an attempt to file an unsubstantiated HR complaint through delegation. It collapsed under the first fact-check. Work went on, but trust in the team hit rock bottom.

What Works: Architecture and Stack

Despite everything, the technical part succeeded:

• Two ManageIQ instances (HA not required if zones are logically isolated)
• Integrations: LDAP, AWX, Netbox, vCenter, oVirt
• Custom portal with auth, catalogs, and API interface
• Service Dialogs + Dynamic Fields + Automation Engine + Tags

The platform is in use, users rely on it, and the infrastructure is operational.

Key Takeaways

• You can build a private cloud on ManageIQ — but it requires technical independence and consistency.
• One toxic team member — especially if it’s the manager — can derail the entire process.
• A weak team can’t be saved even with a solid budget.
• HR processes must be transparent; otherwise, they can be used for pressure.

Recommendations

• Build your core team around technical competence.
• Maintain architectural integrity: define clear zones, providers, and isolation.
• Don’t be afraid to build your own frontend over MIQ — the API and Automate make it feasible.
• Always prepare for the worst-case scenario in management.

Want to Try This Yourself?

Think three times. This is a path for those who aren’t afraid to combine DevOps, engineering, and a minor in applied psychology. But if you succeed, you’ll earn unique experience — and a working product.