Consolidate CPU utilization patterns into 'Match CPU capacity to average and peak utilization'

[russelltrow]

Title

Match CPU capacity to average and peak utilization

Version

1.0

Submitted By

dubrie (original author of both source patterns)

Published Date

2022-11-10 (carried over from the originals; to be confirmed by the patterns working group)

Intent

Narrow the gap between average and peak CPU utilization so less stand-by hardware is provisioned, without compromising peak performance.

Tags

compute, monitoring, size:medium

Problem

optimize-avg-cpu-utilization and optimize-peak-cpu-utilization share an author, date and a verbatim-identical Description, and differ only in whether they target average or peak. Both fail principles 5, 10 and 11 in the August 2026 Operations analysis.

Solution

One pattern with two named techniques: raise average utilization to an efficient level (target set against scaling speed), and reduce/flatten peaks (caching, less data, deferring non-urgent work), cross-referencing the shedding and queuing patterns instead of restating them. Description rewritten rather than reusing the shared paragraph.

SCI Impact

SCI = (E * I) + M per R

E decreases (less idle draw, less CPU work per request); M decreases (less stand-by hardware), unless offset by adding servers.

Assumptions

Utilization is monitored over a window that captures peaks; scaling time is known; system-caused spikes are fixed separately.

Pros & Cons

• PRO: Removes duplicate content; testable Assumptions; Considerations cover real trade-offs; adds Cost Impact.
• CON: Two URLs retire (redirected). Cost Impact is qualitative, with no quantified figures.

---

What changed

• Added docs/operations/capacity-management/match-cpu-capacity-to-average-and-peak-utilization.md
• Moved the two originals to archive/operations/capacity-management/ (same approach as Archive low quality patterns #426)
• docusaurus.config.js: old URLs redirect to the new page
• src/data/personas.ts: old rows replaced with one in the DevOps, Infrastructure and Solution Architect lists (edited by hand to avoid unrelated regeneration drift)

For reviewers

• Title is one of the two options suggested in the wiki analysis.
• The original average pattern recommended reducing average CPU in one place and a higher average if scaling is fast; the new text frames it as a target set against scaling speed.
• This PR touches the same files as Consolidate VM rightsizing patterns into 'Rightsize compute to match utilization' #436 (rightsizing), so expect a trivial merge conflict in docusaurus.config.js / src/data/personas.ts for whichever merges second.
• Local site build not verified (missing @docusaurus/plugin-client-redirects locally); please rely on CI / preview.
• Needs at least one SME review before merge (per Consolidate Operations category duplicate patterns #434).

Part of #434.

🤖 Generated with Claude Code

[russelltrow]

@franziska-warncke I've done another of the really simple Operations patterns consolidation tasks, ready for your review