ecs-architect
Architect-level container platform design without Kubernetes — ECS/Fargate task definitions, service scheduling and placement, ALB integration, secrets and IAM task roles, blue-green deploys with CodeDeploy, and the operational surface a small team can actually run. Use when containers are needed but a full cluster is not.
- License 许可
- MIT
- 最近更新
- 2026-11-13
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- ecs-architect
- license
- MIT
- metadata
- { "crusor-path": "cloud-and-devops-architect/compute-architect/ecs-architect", "crusor-domain": "cloud-and-devops-architect", "crusor-origin": "original", "crusor-sources": "docs.aws.amazon.com/AmazonECS/latest/developerguide; docs.aws.amazon.com/AmazonECS/latest/bestpracticesguide; aws.github.io/copilot-cli", "crusor-triggers": "ecs, fargate, task definition, container service, copilot, aws container, blue green deploy, service discovery, container platform", "crusor-verified-on": "2026-10-30" }
- description
- Architect-level container platform design without Kubernetes — ECS/Fargate task definitions, service scheduling and placement, ALB integration, secrets and IAM task roles, blue-green deploys with CodeDeploy, and the operational surface a small team can actually run. Use when containers are needed but a full cluster is not.
- compatibility
- AWS ECS, Fargate; Cloud Run and Azure Container Apps analogues
03 SKILL.md 正文
ECS Architect
ECS is a container scheduler with a small surface. It wins when the team wants containers, not a cluster to operate — so the design should stay boring.
When to use
- AWS-native containers without Kubernetes expertise.
- Small platform team, few services, strong AWS alignment.
- Batch jobs and scheduled tasks alongside long-running services.
When NOT to use
- Rich scheduling needs, CRDs, or multi-cloud portability →
kubernetes-architect. - A single container with mild traffic → Cloud Run / App Runner / Lambda container.
Decision 1 — Fargate or EC2 capacity
| Factor | Fargate | EC2 capacity |
|---|---|---|
| ops burden | none | patching, scaling, AMIs |
| cost at high, steady utilisation | higher | lower |
| GPU / special hardware | no | yes |
| daemon-style agents | limited | yes |
| predictable burst | good | fine with capacity providers |
- Start on Fargate; move to capacity providers when sustained utilisation makes the price difference real.
Decision 2 — Task definition
- One process per container; a sidecar only for a genuine cross-cutting concern (proxy, log router).
- CPU/memory at task level and container level, sized from measurement.
- Image by digest, not
latest; immutable tags in every environment. - Secrets from Secrets Manager/SSM via
secrets, never in environment plaintext. - Task role (application permissions) separate from execution role (pull/log).
- Health check defined in the task, matching the app's real readiness.
- Read-only root filesystem where the app allows; non-root user.
Decision 3 — Services
- Service per deployable; ALB target group per service, path/host rules upstream.
- Deployment: rolling with circuit breaker, or blue-green via CodeDeploy with traffic shifting and automatic rollback on alarms.
desiredCountacross AZs with spread placement; no single-AZ pinning.- Service discovery via Cloud Map for internal service-to-service calls.
- Autoscaling on ALB request count, queue depth, or latency — not CPU alone.
Decision 4 — Networking
awsvpcmode; private subnets only; egress via NAT or VPC endpoints.- Security groups per service, referencing other services' groups.
- No public IPs on tasks; ingress only through the load balancer.
Decision 5 — Operations
- Logs to CloudWatch with structured JSON and a retention policy; traces via X-Ray/OTel.
- Deploy pipeline: build → scan image → push → register task definition → deploy with rollback alarms.
- Runbook: how to roll back a service, drain a task, and read the last deployment event.
- Cost: Fargate task size drift is the usual leak — a 4 vCPU task for a 200 ms handler.
Failure modes at scale
- Task definition with 4 GB memory for a service that needs 512 MB, ×40 services.
- Secrets in plaintext environment variables visible in the console.
latesttag deployed; two environments run different code with the same name.- No circuit breaker; a failing deploy replaces every task and stays broken.
- One cluster per service, doubling the operational surface for no isolation gain.
Verification receipts
- Task definition inventory: image digest, size, roles, secrets source.
- Blue-green deploy demonstrated with automatic rollback on a bad image.
- Networking check: no public IP, egress path documented.
- Autoscaling policy tested under a synthetic load increase.
- Cost per service at current task size, with right-sizing proposals.
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.