This repository contains a project regarding Data Center Management . The project focuses on designing and implementing a resilient, high-availability infrastructure for a fictional hosting company, "CPDHosting," distributed across three distinct Data Centers and a Client network.
🚀 Overview The infrastructure is simulated using GNS3 and integrates a mix of Linux (Ubuntu/Xubuntu) and Windows Server nodes. It features redundant internet connectivity via two ISPs (MEO & NOS), inter-DC communication through GRE tunnels, and a dedicated cluster for Local Generative AI (running with Ollama).
Key Features: Distributed Architecture: Three Data Centers (DC1, DC2, DC3) interconnected via Layer 3 (OSPF & GRE Tunnels).
High Availability: Implemented via HAProxy and Keepalived for web services and entry points.
GenAI Integration: A local inference cluster in DC3 using Ollama (Llama 3.2) to provide AI services within the private network.
Robust Backup Strategy: Automated 3-2-1 backup system using Duplicati (incremental/encrypted) and custom Bash/PowerShell scripts.
Core Services: DNS (BIND9), Web (Nginx/Apache), Storage (Windows NAS/FTP), and Remote Desktop Gateway.
🛠️ Tech Stack Network Simulation: GNS3
Operating Systems: Ubuntu Server, Xubuntu, Windows Server 2022, Windows 10.
Routing & Switching: Cisco IOS (emulated), OSPF, RIP, NAT, GRE Tunnels.
Services: HAProxy, Keepalived, Ollama (AI), Duplicati, BIND9, Nginx.
🌐 Hybrid Connectivity & Team Collaboration
To simulate a truly distributed environment, we moved beyond a single-host simulation. The communication between the team members' physical PCs was established using [Insert Tool: e.g., ZeroTier, OpenVPN, or a Physical Lab Bridge].
This "physical" bridging layer allowed us to:
Link Local GNS3 Instances: Each Data Center was hosted on a different physical machine, interconnected through a virtualized Layer 2 bridge.
Realistic Latency & Routing: By routing traffic across the internet/local network between our PCs, we could observe more realistic network behavior.
Distributed Services: For example, a client sitting in DC1's virtual network could access the GenAI services hosted on a completely different physical PC in the DC3 environment.
📂 Project Structure
/DataCenter1: Configs for core services (DNS, HTTP).
/DataCenter2: Storage and Windows environment configurations.
/DataCenter3: High Availability setup, GenAI cluster, and Backup nodes.
/Client_Network: Simulated ISP and client access points.
/Documentation: Technical reports and network planning spreadsheets.
🌐 Language Note Please note that the technical reports (Global_Report-pt.pdf, GenAI_Report-pt.pdf) and some of the configuration files/scripts contain comments and documentation in Portuguese.
👥 Authors
Francisco Crespo, José Branco, Rodrigo Lourenço