Service
Server
Windows Server, Linux, Proxmox, Hyper-V.
Scope of work
- 01Server infrastructure consulting and architecture design
- 02Physical server installation, RAID configuration and remote management (iDRAC/iLO)
- 03Proxmox VE and Hyper-V virtualisation deployment
- 04Windows Server deployment: Active Directory, DNS, DHCP, File Server
- 05Linux server deployment and administration (Debian, Rocky Linux)
- 06Backup, snapshot and disaster recovery planning
- 07System health monitoring and alerting
- 08Infrastructure documentation: IP tables, VM inventory, service maps
An incident I have seen many times: a single server running everything, from file shares and accounting software to the camera system. It runs fine for years, then one morning a disk fails and the whole company stops working. Nobody knows how that server was configured or when it was last backed up. Servers are not the hard part. The hard part is designing so the system can recover when something goes wrong.
What is a server — and how does it differ from an ordinary computer?
A server uses hardware architecture similar to a personal computer, but is designed to run continuously 24/7, serve many users at once, protect critical data and allow components to be replaced while running. Redundant power supplies, hot-swap drives, RAID, ECC memory and remote hardware management are what make the real difference.
Choosing architecture: physical, virtualised or hybrid?
There is no universal answer. A workload needing dedicated, consistent performance, such as a large database, may suit its own physical server. Most other services, such as AD, file servers, internal applications and monitoring, should be virtualised to use resources efficiently and simplify backup and migration. I usually recommend a hybrid architecture based on the number of services, the downtime tolerance and the budget.
Virtualisation with Proxmox VE
Virtualisation runs multiple virtual machines (VMs) on one physical server. Instead of buying ten servers, one sufficiently powerful server runs ten independent VMs, each with its own operating system, resources and services. Proxmox VE is an open-source virtualisation platform supporting both VMs (KVM) and containers (LXC). Containers are lighter and start faster, suiting small services such as DNS, reverse proxy or monitoring.
Proxmox's strengths are its intuitive web interface, built-in snapshots and backup, and the ability to join multiple servers into a cluster. With a cluster, VMs can move between nodes without interruption and restart automatically on another node if one fails (High Availability). For organisations wanting to reduce licensing cost, it is a very practical option. For environments tightly bound to the Microsoft ecosystem, Hyper-V remains a sensible choice.
Windows Server — the centre of the Microsoft environment
Active Directory Domain Services (AD DS) centrally manages user accounts, computers and security policy (Group Policy) across the whole organisation. A new employee needs one account to sign in to a computer, reach files and use applications. When someone leaves, disabling one account revokes all access. Internal DNS and DHCP usually run alongside AD: DNS resolves internal names, DHCP assigns IP addresses automatically. File Server shares data with granular permissions by group and department.
Linux — stable, flexible, no licensing cost
Linux suits services that do not need the Microsoft ecosystem: web servers (Nginx, Apache), databases (MariaDB, PostgreSQL), monitoring (Zabbix, Grafana), IT asset management (GLPI) and containers (Docker). Debian and Rocky Linux both offer long support lifecycles and rich documentation. Administration is via SSH and scripting, and in multi-server environments Ansible keeps every machine configured consistently.
Remote hardware management — iDRAC and iLO
Dell iDRAC and HPE iLO are hardware management systems independent of the operating system. When the OS has completely hung, I can still reach the server over the network: restart it remotely, mount an ISO to reinstall the OS, read hardware error logs, monitor temperature and power. For servers in remote sites or data centres, this is indispensable.
Backup and recovery — the most important part and the most often neglected
A system without backup is not a complete system. The principle I apply is 3-2-1: at least three copies of data, on two different media types, with one copy off-site. More important still is testing recovery regularly. A backup that has never been restored is a belief, not a guarantee. A disaster recovery plan must answer two questions: how much data loss is acceptable, and how quickly the system must be running again.
When does server infrastructure need rebuilding?
When one server carries too many services, when hardware is old and out of warranty, when there is no reliable backup, when the company is growing and needs to expand services, or when nobody fully knows what the system is running any more.