A Practical Guide to Sourcing Enterprise Servers
Specifying a server for a company is a different exercise than buying a desktop PC — the priorities flip. A desktop is judged on peak performance; a server is judged on whether it stays up, whether it can be serviced without downtime, and whether it still has room to grow in three years. Here's how to think through the decision.
Rack vs. tower — start with your physical setup, not the spec sheet
A 2U rack server is built for a proper server room or data center rack — it's the right choice once you have more than a couple of servers, or if you're colocating in a data center at all. A tower server makes more sense for a small office with no dedicated server room — it can sit in a closet or under a desk without needing rack infrastructure. If you're not sure yet, the deciding question is simply: do you have (or are you planning) a server rack?
Xeon-class vs. EPYC-class — the CPU decision
Both are built for the same job — sustained, multi-core workloads running 24/7 — but they come from different design philosophies. Xeon-class CPUs have the longer track record in enterprise deployments and broad compatibility with existing server management tools. EPYC-class CPUs typically offer a higher core count and more memory channels per socket, which matters most for virtualization-heavy workloads or database servers that are memory-bandwidth constrained. For most general-purpose business servers, either is a defensible choice — the deciding factor is usually which one fits your existing infrastructure and admin team's familiarity, not raw benchmarks.
How much memory, and why ECC matters
Server memory should always be ECC (error-correcting code), not standard desktop memory — ECC memory detects and corrects the small memory errors that inevitably occur over months of continuous uptime, which is exactly the failure mode a server can't afford. On capacity: undersizing memory is the single most common server-sizing mistake, because the cost of adding RAM later (potential downtime, matching existing modules) is much higher than specifying generously up front.
Power supply redundancy — this is not optional for anything business-critical
A redundant PSU setup (two power supplies, either one capable of running the server alone) means a single power supply failure doesn't take the server down — it just needs replacing at a convenient time instead of during an emergency. For anything running a business-critical service, this is worth the modest extra cost. Titanium and Platinum efficiency ratings also matter over the server's life: a more efficient PSU means a measurably lower electricity bill for a system running continuously for years.
The question to answer before requesting a quote
Before reaching out for pricing, have rough answers to: how many physical servers, rack or tower, expected workload (virtualization, database, file/application server), and your realistic growth window. You don't need exact numbers — even directional answers let us narrow down a configuration fast instead of guessing.
Browse NAD's IT hardware range — including Xeon-class and EPYC-class server CPUs, ECC server memory, and redundant server power supplies — or get in touch with what you're building and we'll help you land on the right configuration.
