liveBook · Manning

Nonfunctional requirement

As explained in Decision Engineering

A requirement that specifies a system quality, constraint, or performance expectation rather than a function the system performs.

A nonfunctional requirement describes how well a system must work, or at what scale, rather than what function it performs. Latency, availability, throughput, data retention, and security may be invisible to users as features, but they can determine the architecture. The same feature list can describe very different systems when one serves ten users and another serves ten million. section 2.2 section 2.1.3

These expectations are often missing from product conversations. Rather than asking stakeholders to invent technical targets in isolation, engineers can turn business consequences into measurable options. Service tiers give stakeholders a shared vocabulary for choosing commitments before the team commits to a design. section 2.2 section 2.5

Different dimensions call for different conversations:

  • Discuss availability and durability together: both depend on topology decisions. Tiers can range from Basic to Critical for availability and Minimal to Very High for durability. section 2.5
  • Set latency expectations against human perception, from Instant to Background. Apply throughput tiers at the scale of each component, not necessarily the whole system. section 2.5
  • Account for volatility: a Foundational component and an Ephemeral one are expected to change at different rates. section 2.5
  • Handle compliance separately. Legal obligations are largely binary; security calls for risk-and-control discussion. Engage the relevant experts while architectural options remain open, and distinguish requirements from controls. section 2.5

When a target changes the design

An availability commitment can determine whether a single region is enough or multiple regions are needed. A latency target can determine whether a synchronous call across a service boundary is acceptable. A data-retention rule can affect storage tier, backups, and even database choice. These are architecturally significant requirements: discovering them late may require changing the architecture, not just an implementation detail. section 2.2 section 2.1.3

Use tiers to make assumptions and trade-offs explicit, not to impose the highest specification everywhere. Before moving to solutions, identify the requirements that could break the proposed design, assign tiers to the relevant nonfunctional dimensions, and put compliance constraints on the table. section 2.5 section 2.1.3

How Decision Engineering covers it

How deep it goes
Defines the category, traces several design consequences, and offers tiered discussion; detailed collection and definition are explicitly outside its focus.
Examples
Contrasts applications serving ten and ten million users, and connects availability, latency, and retention targets to concrete design choices.
What you can do after
Practice/perform tier-based discussions to surface architectural commitments and test whether missing requirements could invalidate a proposed design.
How it connects
Links nonfunctional requirements to architecturally significant requirements, service tiers, availability, durability, latency, throughput, volatility, and compliance.

Also explained in

Compare how 9 books explain nonfunctional requirement →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Decision Engineering. The section links point to the parts of the book it draws on.