Setting the file. One moment.
Subchapter 25.47
references/phases/workshop/workshop-compare.mdMarkdown3 KBView on GitHub
Fragment unit. See
workshop.mdfor how it is composed into the sidebar.
Renders the priced scenarios side by side so the tradeoff is visible: what each knob change cost or saved, and what it changed about the architecture. Comparison is capped at five scenarios — beyond that the table stops informing a decision.
scenarios/index.json (if missing, tell the user to Apply once for baseline).created_at), build columns:
global.target_region.valuedata.availability.valuedesign_constraints.compute_target.value when presentdesign_constraints.cpu_architecture.value when presentestimation_summary tiers (per
estimate-infra.md § Scenarios — Balanced is the anchor, Premium and
Optimized are stated adjustments off it)estimation_summary.complexity_tierestimation_summary.recommendation_outcome when present (omit the column when no scenario carries it)| Scenario | Region | HA | Compute | Arch | Premium $/mo | Balanced $/mo | Optimized $/mo | Complexity | Outcome |
|---|
Outcome flip callout: when any scenario’s recommendation_outcome differs
from the baseline’s, add one line under the table naming the flip and the knob
that drove it — e.g. “single-az scenario flips conditional_go → go: the
availability-downgrade condition no longer applies.” This is the compare view’s
most decision-relevant line; never bury it.
Forced-architecture note: if any scenario carries (architecture forced by <resource>) in its label (see workshop-refresh.md § Snapshot), keep that
parenthetical visible in the Scenario column — it explains why that row’s Arch
is x86_64 even when the user asked for Graviton on the sheet.
Under the table: active vs baseline preferences_subset; any region_note;
reminder that inventory is frozen. Keep under 25 lines.
estimation_summary.calculator_url is always null — no shareable calculator
link is produced, so do not emit one.
Implemented, ported from gcp-to-aws‘s equivalent and adapted to this skill’s
knobs (region / HA posture / compute target / cpu_architecture — GCP’s compare
table instead shows region / availability / kubernetes / cpu_architecture) and
its Premium/Balanced/Optimized scenario schema. Added the forced-architecture
note, which has no GCP analogue — GCP’s Graviton risk-signal caveat travels as
workshop.graviton_note rather than a hard per-scenario label suffix, because
Azure’s forced_by is a harder constraint than GCP’s risk-signal tiers (see
workshop-refresh.md § Status).