Servers & Storage

The benefits

• Customised servers come with exactly the connections you need.
• And the hardware that is important for your situation. This prevents waste or underperformance.
• We work with you to optimise performance for the tasks the server needs to perform.
• Tailored to your environment, by which we mean: how many staff members use the machine simultaneously, how much data must the machine be able to process at peak times, has consideration been given to what happens if this machine fails, and which operating system is best suited to which task

What makes a server different

A server is not just a powerful computer. It is a system designed to run unattended, continuously and reliably. Servers are the central machines in your network, facilitating your work efficiently and productively.

What makes server hardware different:

• IPMI / BMC: remote management, even if the operating system is unresponsive
• ECC memory: automatic error correction, essential for long-running processes
• Redundant power supply: the system continues to run if one power supply fails
• Hot-swap drive bays: replace drives without shutting down the system
• Industrial components: motherboards and power supplies with a longer lifespan
• Bandwidth: server motherboards and CPUs such as AMD Epyc and Intel Xeon offer significantly higher bandwidth and throughput

What a NAS does well

A Synology or QNAP NAS is quick to set up, has a user-friendly interface and offers basic functionality for file sharing, backup and media. For a small office with 5 users and a few hundred gigabytes of data, it is often sufficient.

Where a NAS falls short

• Performance: the processor and memory of a consumer NAS are limited. With simultaneous use by 5+ users, or with large files (video, CAD), performance quickly drops.
• Expandability: a 4-bay or 8-bay NAS will eventually run out of space. Expanding means buying a new device.
• Flexibility: you are tied to the manufacturer’s operating system. If you want to run a specific service that isn’t available as an ‘app’, you’re out of luck.
• Data integrity: most NAS systems use ext4 or btrfs without end-to-end checksumming. Bit rot — silent data corruption — goes undetected.

What we build

Our file servers run on ZFS — a file system that guarantees data integrity right down to the core. Every bit is checked during reading and writing. Corruption is automatically detected and corrected using redundancy. Of course, a different file system can be used if desired.

The hardware is a true server: ECC memory, IPMI for remote management, hot-swap drive bays for easy maintenance, and a chassis that can hold 12, 24 or more drives. Similarly, the network card — a crucial factor for storage servers — can be selected and configured to suit your working environment.

The 3-2-1 rule

The gold standard for backups is simple:

• 3 copies of your data (the original + 2 backups)
• 2 different media (for example: server + external drive)
• 1 copy at a different location (offsite)

This protects against hardware failure, human error, ransomware and physical disasters (fire, water damage). Most companies we encounter have at most 1 backup, at the same location, on the same type of media.

What we build

• Primary backup server: on-site, ZFS-based, daily incremental snapshots. Hot-swap drives for easy replacement in the event of failure.
• Offsite backup: a second server at a different location, or encrypted backup to an external storage service. Automatic, unmanaged, monitored.
• Backup verification: a backup you don’t test isn’t a backup. We configure automatic restore tests so you know your data can actually be recovered.

How much space do you need?

Rule of thumb: allow for at least twice the amount of data you wish to protect. Incremental backups and snapshots grow over time. With compression on ZFS, you typically save 30–50%, but it is wise to leave some space for future growth.

HDD vs SSD

HDDs (hard disk drives) are inexpensive per terabyte and ideal for large amounts of data that are not accessed constantly: archives, backups, media libraries. A 10TB HDD costs a fraction of a 10TB SSD.

SSDs are faster, quieter and more reliable for actively used data. For databases, virtual machines and working files, SSD is the standard.

Most server systems we build combine both: SSD for active data, HDD for archives and backups. ZFS makes it easy to combine these tiers into a single coherent storage system.

HBA cards

An HBA (Host Bus Adapter) connects your drives directly to the operating system, without the intervention of a RAID controller. This is essential for the file system, which manages its own redundancy and data integrity (we prefer to use Linux for storage servers).

We use HBA cards from Broadcom (LSI) and Areca in IT mode — the industry standard. They are reliable, widely supported and allow the operating system to retain full control over the drives.

Why not use hardware RAID

Hardware RAID controllers have their own firmware, cache and logic. If the controller fails, you need exactly the same model to read your data. The controller is a single point of failure that you can avoid by using software RAID (ZFS). RAID controllers are used almost exclusively for Microsoft OS systems; we consider Linux to be the better operating system for storage systems.

With Linux file systems on an HBA, you have:

• End-to-end checksumming (no silent data corruption)
• Automatic scrubbing and repair
• Snapshots and clones without performance loss
• Full portability — the drives work in any other machine using, for example, ZFS

What is IPMI

IPMI (Intelligent Platform Management Interface) — also known as BMC (Baseboard Management Controller) — is a small, independent computer on the server’s motherboard. It is always running, even when the server itself is switched off. Via a separate network interface, you can:

• Switch the server on and off remotely
• Configure the BIOS as if you were sitting in front of it
• Install the operating system via a virtual console
• Monitor temperatures, voltages and fan speeds
• Diagnose hard drive and memory errors

This isn’t a luxury — it’s a basic requirement for any server that isn’t sitting next to your desk.

Monitoring

In addition to IPMI, we always recommend monitoring software that proactively alerts you to problems:

• Drive reporting SMART errors → replace before it fails
• Temperature exceeding threshold → check cooling
• Memory logging ECC corrections → replace module
• Backup that has not completed successfully → take action

Most problems give warning signs before they become critical. Monitoring picks up on these signals. Without monitoring, you will only notice when it is too late.

What we provide

Every server we build comes with IPMI as standard. We configure the IPMI interface, set the alert thresholds and document the access credentials. On request, we install monitoring software and configure email or webhook notifications.