Creating exact performance specifications for a new network design is a challenging task. There can be many unknowns that force the designer to make assumptions and guesses. The level of uncertainty also grows as the number of design variables increases. Engineers use factors of safety to compensate for any errors that are caused by the unknowns (Some fields of engineering use well-known safety factors).
In network design, adding a safety factor could mean doubling the amount of the WAN bandwidth needed between sites to accommodate unexpected traffic demands, or increasing the link budget for an optical link to offset attenuation caused by unexpected/unknown splicing, or using a firewall with higher VPN throughput than initially estimated.
The use of safety factor comes at additional cost; therefore, the more accurate the performance estimates are the smaller the factor needs to be. The safety factor is also higher in the design of new networks, when little is known about how the network will be perform in production. Upgrades and or designs of similar types of networks need lower safety factor because of the greater knowledge gathered about the network’s performance.
A safety factor reduces the probability of failure by exceeding the design specifications at the expense of higher cost. The factor of safety (or uncertainty) should not be confused with scalability provisions. The former protects against design errors (under estimation), uncertainty, and variability, whereas the latter is used to accommodate future growth.
Read about other Network Design Principles.