⚠  These docs are a work in progress. Content is being written actively. Structure is stable; detail pages are placeholders.

Nyxever Documentation

Architecture guidance and best practices for building resilient infrastructure, organized around the Nyxever reference foundation: compute, networking, storage, identity, AI and observability. Concept-first — architecture before technology, understanding before configuration.

Written for operators: people who run systems, read logs, and need to understand what is actually happening, as well as whether things are working. Every layer is documented and independently replaceable.

Design principle Every component in Nyxever can be understood from first principles. If a section of this documentation doesn't explain the why as well as the how, that's a documentation bug worth filing.

Requirements

Nyxever runs on commodity server hardware. No vendor support contracts, specialized hardware, or cloud accounts required to start. A single physical machine is enough for development. Three nodes is the minimum for a production HA cluster.

Single node (development) x86-64 with hardware virtualization (VT-x or AMD-V), 32 GB RAM, two drives (OS + storage), 1 GbE network.
Three-node cluster (production) 3× servers, 64 GB+ RAM each, dedicated SSD for storage metadata, 10 GbE interconnect, out-of-band management (IPMI/iDRAC/iLO).

Installation

Nyxever layers on top of Proxmox VE as the base hypervisor. The installation process starts with a standard Proxmox VE install, then applies the Nyxever configuration layer via infrastructure-as-code (tooling under evaluation). Each layer is applied independently and in order.

⚠  Detailed installation guide in progress. Infrastructure-as-code tooling is under evaluation.

General sequence: install Proxmox VE → configure networking → apply Nyxever infrastructure-as-code → bootstrap distributed storage (Ceph) → deploy identity services → bring up observability → optionally deploy AI inference.

Quick Start

The quick start gets a single-node environment running on one physical machine or a cloud VM. Suitable for learning and testing; it has no redundancy, so production deployments use the three-node cluster.

⚠  Quick start guide in progress.

Sections