Designing Efficient Data Center Architectures: LAN, SAN, POD & ToR Explained

Data Center Architecture Design

Data center cabling includes internal and external wiring, with system architecture divided into Main Distribution Area (MDA), Horizontal Distribution Area (HDA), and Equipment Distribution Area (EDA). Connections follow a star topology from MDA to HDA and HDA to EDA.

LAN System Network Architecture

1) Uses a three‑tier architecture: core‑aggregation‑access.

2) In each floor’s equipment module, a core area (MDA) and a local aggregation area (IDA) are defined.

3) Each module has two aggregation (IDA) distribution zones, with dual links between HDA and IDA.

4) Core‑to‑core and aggregation‑to‑aggregation interconnections are established within the same module.

SAN System Network Architecture

1) Employs fiber‑direct, centralized distribution.

2) SAN switches are placed in the main distribution area.

KVM System Network Architecture

1) KVM considers servers and network devices.

2) KVM switches reside in the HDA of the column cabinet; servers connect via copper cables.

3) KVM switches interconnect through LAN switches.

Three Common Architectures

Distributed Architecture

In a typical distributed switching design, each server rack connects via copper or fiber cabling. Copper cabling often uses 6A/EA‑class 10G Base‑T cables, with quantities determined by client requirements. Fiber cabling follows similar client‑driven specifications.

Additional auxiliary cabling is required for core network switches and SAN systems, covering areas such as carrier external services, main network equipment, high‑density zones, and SAN core/storage.

In distributed designs, copper links run between equipment racks and main distribution frames, providing centralized cross‑connects that maximize switch and port utilization while remaining server‑agnostic and highly flexible.

POD Design Architecture

PODs are modular units comprising processing, storage, network, and application components, designed for repeatable, scalable, and manageable data‑center space.

Modular POD designs enable adjustable power and cooling per rack (e.g., 30 kW capacity with 4 kW increments), supporting varied device densities within a single POD.

ToR Switching Architecture

ToR architecture provides high port density and I/O consolidation, reducing the number of cables and switches needed per POD.

In a POD, three cabinets form a unit with a top‑of‑rack (ToR) switch in the middle cabinet, delivering local access to all three cabinets. 10 GbE fiber backbones connect to aggregation and core layers.

Each server uses a converged network adapter (CNA) over 6A‑class 10 GbE links for both LAN and SAN traffic, connecting to the middle cabinet switch.

Copper 10G Base‑T cabling between servers and ToR switches reduces intra‑rack cable counts, adapters, transceivers, power consumption, and cooling load.

Only a few fiber links are needed to extend to the aggregation layer, decreasing the total number of switches and saving rack space and operational costs.

TOR Switching Architecture

A unified I/O network merges Ethernet and Fibre Channel traffic onto single 10 GbE links, simplifying cabinet‑level connectivity. Deploying access switches at the cabinet level reduces required adapters, transceivers, and uplink ports.

ToR designs require understanding the number and density of network connections per compute resource; deployments exceeding 48 links per server typically need an additional access layer switch per cabinet.