Internal redundancy refers to the duplication of critical components within a data center's own infrastructure to prevent a single point of failure. Chillers are part of the data center's internal cooling system (HVAC). Implementing N+1 or 2N chiller configurations ensures that the failure of one unit does not compromise the environmental controls necessary to keep equipment operating within safe temperature limits. This is a classic example of building resiliency into the facility's internal mechanical plant.

A. Power feeds are typically external connections from a utility provider. While multiple feeds provide redundancy, they are external dependencies, not internal components.

C. Network circuits are external connections from telecommunication providers. Redundancy is achieved by using multiple carriers, which is an external dependency.

D. Generators are on-site but serve as a backup for the failure of an external resource (utility power), rather than providing redundancy for a constantly operating internal system.

1. Uptime Institute. (2018). Data Center Site Infrastructure Tier Standard: Topology. Section 7.3.2 Cooling System (N+1). The standard details the requirement for N+1 concurrently maintainable critical mechanical components, such as chillers, for Tier III certification.

2. Gao, J. (2012). Digital Utility: A New Infrastructure for the Enterprise and the Cloud. Morgan Kaufmann. Page 136 discusses the N+1 redundancy model for both power and cooling systems, stating, "For example, if a data center needs 10 chillers to provide the required cooling, an N+1 design would have 11 chillers."

3. Barroso, L. A., & Hölzle, U. (2009). The Datacenter as a Computer: An Introduction to the Design of Warehouse-Scale Machines. Morgan & Claypool Publishers. Page 21 discusses the need for redundancy in cooling systems, noting that "a single failure in the cooling system can take down the entire WSC [Warehouse-Scale Computer]."