What Every Engineer Should Know About Server Hardware, RAID, and Network Cards

Chapter 1: Server Fundamentals

In networking, a server is a computer that provides services to client machines, characterized by high performance, high reliability, and strong I/O capabilities, and is essential for enterprise applications, big‑data services, HPC, and internet workloads.

Typical server hardware includes CPU, memory, storage, RAID controller, power supply, motherboard, and chassis.

CPU Architectures

CISC : Primarily Intel (non‑Itanium) and AMD processors.

RISC : IBM Power series, Sun SPARC series, and ARM‑based CPUs for servers.

In July 2017 Intel launched the “Purley” platform with the Xeon Scalable family, replacing the E5/E7 naming with Platinum, Gold, Silver, and Bronze tiers; Skylake is the first generation, followed by Cascade Lake.

Server Categories

Mainframe : Large‑scale systems used by governments, banks, and large manufacturers, offering extreme processing power, stability, and security.

Minicomputer : Common in finance, power, and telecom, valued for Unix OS support, high availability, vertical scalability, and strong concurrency handling.

x86 Servers : Based on CISC processors, tracing back to Intel’s 8086 and evolving into modern Xeon CPUs.

ARM Servers : Energy‑efficient RISC solutions.

Typical deployment models include Client/Server (C/S) with high‑performance PCs or workstations and enterprise databases (Oracle, SQL Server, etc.), and Browser/Server (B/S) where the client uses a web browser and the server hosts the database and application logic.

Chapter 2: Disk and RAID Basics

Servers employ various storage devices and RAID configurations to balance performance, redundancy, and capacity. Common RAID levels (0, 1, 5, 10) are explained along with their trade‑offs, and typical enterprise HDD/SSD specifications are highlighted.

Chapter 3: Network Card Fundamentals

Network Interface Cards (NICs) operate at the physical and data‑link layers of the TCP/IP model, handling transmission and reception of frames.

Key Functions

Provide a fixed MAC address for addressing.

Encapsulate and decapsulate data into Ethernet frames.

Manage link status to avoid collisions on shared Ethernet.

Encode/decode binary data into electrical or optical signals.

Handle send and receive operations.

Classification

Bus type : PCIe, USB, ISA, PCI (legacy).

Form factor : Integrated (LOM), PCIe add‑in cards, mezzanine cards.

Application type : Workstation‑grade NICs vs. server‑grade NICs.

Connector Types

Electrical (RJ45) : Standard copper Ethernet ports.

Optical : Ports for SFP+, SFP28, QSFP+ modules.

Optical modules differ in speed: SFP+ (up to 10 Gbps), SFP28 (up to 25 Gbps), QSFP+ (40 Gbps and above). Direct‑attach copper (DAC) cables combine transceiver and cable for short runs (1‑5 m) at low cost, while active optical cables (AOC) provide fiber‑like performance in a single cable but at higher price.