Design – Build – Run: Applied Practices and Principles for Production-Ready Software Development

5.2. The Hardware Topology

This section takes a look at the hardware architecture — specifically, the typical server configuration, high availability and fault-tolerant features, shared storage and the storage area network (SAN), and, finally, network segmentation.

5.2.1. Individual Server Configuration

In keeping with the fault-tolerant theme, every "enterprise-level" server usually has internal fault-tolerant and redundancy characteristics to provide a robust environment and to reduce single points of failure. These include:

• Dual power supplies — Protects against hardware failure in one of the power supplies.

• Dual fans — Protects against hardware failure in one of the cooling fans.

• Dual network interface cards (NICs) — Protects against hardware failure in one network card. Depending on the number of network connections required, the server will have dual network cards for each connection — one acting as the primary and the other acting as the backup. This is generally referred to as teaming.

• Dual fiber cards — This is usually required only for the database servers that are connected to the SAN or other fiber channel resources. Doubling up the network cards protects against a failure in one. However, four network cards and slots are required to support both dual NICs and dual fiber cards.

Fault tolerance is paramount in a production system; however, memory and local disk still remain problematic. These issues can cause an outage with an individual server, which may require you ...