release-strategy-architect
Architect-level release practice — small-batch delivery, feature flag and canary strategy, expand/contract database changes, rollback versus roll-forward decisions, and the versioning and communication that keep releases boring. Use when deciding how code reaches users.
- License 许可
- MIT
- 最近更新
- 2026-11-13
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- release-strategy-architect
- license
- MIT
- metadata
- { "crusor-path": "website-architect/testing-and-deployment-architect/release-strategy-architect", "crusor-domain": "website-architect", "crusor-origin": "original", "crusor-sources": "12factor.net; martinfowler.com/articles/feature-toggles.html; sre.google/workbook/canarying-releases", "crusor-triggers": "release strategy, feature flags, canary, blue green, rollback, rollout, deployment frequency, versioning", "crusor-verified-on": "2026-10-30" }
- description
- Architect-level release practice — small-batch delivery, feature flag and canary strategy, expand/contract database changes, rollback versus roll-forward decisions, and the versioning and communication that keep releases boring. Use when deciding how code reaches users.
- compatibility
- Any stack; feature flag service optional
03 SKILL.md 正文
Release Strategy Architect
A release is a reversible step, not an event. If it cannot be undone in one action, it is not a release — it is a gamble.
When to use
- Planning how changes reach users, or deploys feel risky.
- Adding flags, canaries, or multi-environment rollout.
- A rollback is impossible because a migration already ran.
When NOT to use
- Choosing the pipeline stages themselves (see
ci-cd-architect).
Decision 1 — Small batches
- Deploy often and small; batch size is the strongest predictor of incident duration.
- Separate deploy (code present, switched off) from release (users affected).
- A change that needs a maintenance window is a design that skipped this rule.
Decision 2 — Flags
| Flag type | Lifetime | Owner |
|---|---|---|
| release flag | days | the team shipping |
| experiment flag | weeks | product |
| ops/kill switch | permanent | on-call |
| permission flag | permanent | entitlements (not a flag) |
- Every flag has an owner and a removal date; stale flags are debt with a blast radius.
- Flag evaluation must be fast and fail-safe: on error, serve the last known good state.
- Never use flags to choose a database schema shape you cannot migrate.
Decision 3 — Rollout mechanics
| Strategy | Use when |
|---|---|
| all-at-once | low risk, easy rollback, non-critical internal services |
| canary (1% → 10% → 50% → 100%) | user-facing changes with metrics to judge |
| blue/green | need instant cutover and instant revert |
| ring/cohort | enterprise customers and staged exposure |
- Canary needs a judge: error rate, latency, business metric — defined before rollout, not eyeballed.
- Automated abort on SLO burn; a human watching a graph is not a control.
Decision 4 — Data is the hard part
- Schema changes are expand/contract: add compatible structure, dual-write, backfill, switch reads, remove old — across releases.
- Never drop or rename a column in the same release that stops using it.
- Backfills are batched, idempotent, resumable, and monitored.
- "Rollback" of a data migration is a forward migration, not a restore from backup — plan it.
Decision 5 — Rollback vs roll-forward
- Decide the trigger in advance: error rate above X for Y minutes → revert.
- Rollback must not depend on a schema reversal; that is why data changes ship separately.
- Roll-forward is acceptable when the fix is small and known; document which path you chose and why.
- Post-release verification: smoke journey plus the metric that justified the release.
Decision 6 — Communication
- Version numbers or changelogs for anything external; internal services use commit-based deployment records.
- Release notes name user-visible changes, not commits.
- Announce risky releases in the team channel with the rollback command in the message.
Failure modes at scale
- One deploy containing 40 changes, one of which breaks the site — bisecting in production.
- 300 permanently-on flags; nobody knows which code is live.
- A migration dropped a column, and rollback means restoring a backup.
- Canary deployed to 1% but the metric chosen cannot detect the bug.
- Friday deploy with the one person who knows the rollback on holiday.
Verification receipts
- Deployment frequency and change failure rate over the last quarter.
- Flag inventory with owner and expiry; count of flags older than 90 days.
- Canary run with an automated abort demonstrated.
- Migration plan for a live table (expand/contract, separately deployable).
- Rollback drill executed and timed.
04 同仓库更多 Skills
interaction-patterns-architect
Architect-level interaction design in code — choosing established patterns over invented ones, feedback and latency behaviour, destructive-action safeguards, navigation and overlay conventions, and the state model behind every interactive surface. Use when a screen has to behave, not just look.
design-system-ops-architect
Architect-level operation of a design system in code — token pipelines and versioning, component API and variant policy, the catalogue as the contract, adoption metrics, and the governance that stops a system from forking into three. Use when a component library must survive several teams.
design-handoff-architect
Architect-level design handoff — what must be specified before implementation starts, tokens over raw values, responsive and content-limit contracts, state matrices, and the review loop that catches drift between design and code. Use when a design is about to become code.
content-microcopy-architect
Architect-level UI content — where strings live, voice and terminology consistency, error and empty-state copy as part of the interface, translation-ready authoring, and the content model a product needs before it multiplies screens. Use when the words in the interface matter as much as the pixels.
ui-ux-implementation-architect
Architect-level direction for turning a design into working UI — the handoff contract, component ownership versus screen-level layout, interaction and state conventions, design QA gates, and the review habits that keep an implementation faithful without frozen pixels. Use before any specific UI implementation skill.
test-strategy-architect
Architect-level test strategy — allocating assertions across unit, integration, contract and end-to-end layers, designing for testability through dependency boundaries, fixture and data strategy, and the flake and coverage policies that keep a suite trustworthy. Use when defining how a codebase is tested.