Chapter 16 High Availability and Disaster Recovery

Fault Tolerance

The ability of a system to continue operating properly in the event of a failure of one or more components.

High Availability

Techniques and configurations designed to ensure a system or service remains accessible and operational with minimal downtime.

Load Balancing

Distributes network or application traffic across multiple servers to improve responsiveness and redundancy.

NIC Teaming

Combines multiple network interface cards into one logical interface to improve bandwidth and provide redundancy.

Multipathing

Uses multiple physical paths between a computer and a storage device to ensure fault tolerance and load balancing.

Switch/Router/Firewall Clustering

Groups multiple devices to work together for redundancy and high availability.

Active-Active Configuration

All systems are operational and handle traffic simultaneously.

Active-Passive Configuration

One system is active while the others are on standby and become active only during failure.

FHRP (First Hop Redundancy Protocol)

Protocol group that enables a virtual router to provide gateway redundancy (e.g., HSRP, VRRP).

HSRP (Hot Standby Router Protocol)

Cisco proprietary FHRP that allows a backup router to take over if the primary fails.

VRRP (Virtual Router Redundancy Protocol)

Open standard alternative to HSRP (RFC 2338).

HSRP MAC Address Group Number

Determined by the last two hexadecimal digits (e.g., 0000.0c07.ac0a = group 10).

Uninterruptible Power Supply (UPS)

Provides temporary power during outages to maintain uptime until generators activate or systems shut down properly.

Standby UPS

Switches to battery power when utility power fails.

Line-Interactive UPS

Maintains charge and conditions power, better for minor fluctuations.

Online UPS

Provides constant, clean power from battery-powered inverter; best for critical data centers.

Generator

Provides long-term backup power after UPS battery is exhausted.

Power Distribution Unit (PDU)

Distributes and manages power in server racks.

Power Load

The total power drawn by all connected devices.

Environmental Controls

Include temperature, humidity, and fire suppression systems to maintain optimal conditions in data centers.

HVAC

Heating, ventilation, and air conditioning system critical for temperature control.

Humidity

Must be controlled to prevent electrostatic discharge or condensation.

Fire Suppression (Clean Agent)

Gas-based systems (like FM-200) that extinguish fire without damaging electronics.

Deluge System

Water-based system; not suitable for electronics.

Dry Pipe / Preaction Systems

Triggered only under specific conditions; reduces risk of accidental water discharge.

Disaster Recovery Site

Alternate location used to restore operations in case the primary site becomes unusable.

Cold Site

Basic infrastructure; takes longest to set up after disaster.

Warm Site

Preconfigured with some systems and data; faster recovery than cold site.

Hot Site

Fully operational duplicate of primary site with live data and systems; fastest recovery.

Cloud Site

Virtual DR site hosted in cloud, mimics on-premises network.

RPO (Recovery Point Objective)

Maximum tolerable amount of data loss (i.e., how far back in time the recovery point can be).

RTO (Recovery Time Objective)

Maximum tolerable downtime; time needed to recover after disruption.

MTTR (Mean Time to Repair)

Average time to restore a failed system.

MTBF (Mean Time Between Failures)

Expected time between system failures.

Tabletop Exercise

A discussion-based simulation where team members walk through a hypothetical disaster scenario to test response procedures.

Validation Test

Involves running procedures and configurations in a test environment to confirm readiness.

Testing

Includes exercises and system tests to ensure DR plans are functional and complete.

Stacking

Combines multiple physical switches into a single logical switch to simplify management and increase reliability.

Diverse Paths (ISP Redundancy)

Using multiple ISPs or physical paths to avoid a single point of failure in connectivity.