Efficiency Practices — Quick Wins for Carbon + Cost

"Yeşil yazılım 'gelecekte düşünelim' işi değil — bu çeyrek uygulanan pratikler %20-50 emisyon azaltır + cost ↔ carbon dual'i çoğu durumda maliyet de azaltır. Kazan-kazan."

Bu rehber yeşil yazılım için hızlı uygulanabilen pratikleri — ARM/Graviton, spot instance, idle cleanup, compression, caching, right-sizing — somut komut + ROI ile anlatır.

---

🎯 Cost ↔ Carbon Dual

"Her FinOps kazancı çoğu zaman sustainability kazancıdır."

Pratik	Cost azaltma	Carbon azaltma
Idle cleanup	%10-30	%10-30
Right-sizing	%15-25	%15-25
Spot instance	%70	%30 (yeni hardware üretmiyor)
ARM/Graviton	%20-40	%20-40 (per-watt verim)
CDN	%20-40 (bandwidth)	%20-40
Compression	%5-15	%5-15
Cold tier (eski log)	%80 (storage)	%50

🔑 Quick wins: 4-8 hafta uygulanır, Q1'de %20-30 reduction normal.

---

1️⃣ Idle Resource Cleanup

Hedefler

• Kullanılmayan EC2, EBS, RDS, NAT Gateway
• Boş S3 bucket, eski snapshot'lar
• Zombie load balancer'lar
• Test environment'ları (haftasonu kapalı)

Tooling

• AWS Trusted Advisor → idle resource list
• Kubecost → K8s pod-level idle
• Cloud Custodian → policy-based cleanup
• awsweeper → idle taraması + delete

Custodian örneği

# cleanup-old-snapshots.yml
policies:
  - name: ebs-snapshot-old
    resource: ebs-snapshot
    filters:
      - type: age
        days: 90
        op: gt
      - "tag:Name": null
    actions:
      - delete

Idle EC2 finder

# CloudWatch CPUUtilization < %5 son 14 gün
aws cloudwatch get-metric-statistics \
  --namespace AWS/EC2 --metric-name CPUUtilization \
  --start-time $(date -d '14 days ago' +%FT%T) \
  --end-time $(date +%FT%T) \
  --period 86400 --statistics Average

---

2️⃣ Right-Sizing

Pattern

• Memory: %30+ headroom OK, %70+ bos = ölçek küçült
• CPU: peak %50 → 1 size küçült
• Disk: utilization < %20 → küçült veya tier düşür

Tooling

• AWS Compute Optimizer → otomatik öneri
• Kubecost → K8s right-sizing (önerilen pod resource)
• Vertical Pod Autoscaler (VPA) → otomatik

VPA örneği

apiVersion: autoscaling.k8s.io/v1
kind: VerticalPodAutoscaler
metadata:
  name: payments-vpa
spec:
  targetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: payments
  updatePolicy:
    updateMode: "Auto"   # veya "Off" (sadece öneri)
  resourcePolicy:
    containerPolicies:
      - containerName: payments
        minAllowed: {cpu: 100m, memory: 128Mi}
        maxAllowed: {cpu: 2000m, memory: 4Gi}

⚠️ HPA + VPA ikisi birlikte ihtilafa girebilir. VPA recommendation-mode'da kullanılır + HPA ölçek için.

---

3️⃣ ARM / Graviton Migration

Avantajlar

• Per-watt 2-4x verimli (Graviton 3 → 4)
• Maliyet: x86 instance'a göre %20 daha ucuz
• Performance: many workload'da daha hızlı

Uygun workload'lar

• ✅ Go (cross-compile kolay)
• ✅ Java (JVM Graviton optimized)
• ✅ Python (CPython arm64 wheel)
• ✅ Node.js (native arm64 v20+)
• ✅ Rust (cross-compile)
• ⚠️ Sayısal (NumPy, TensorFlow): kontrol et — bazı pip wheel arm64 yok

Migration adımı

# Multi-arch Dockerfile
FROM --platform=$BUILDPLATFORM golang:1.23 AS builder
ARG TARGETOS TARGETARCH
WORKDIR /src
COPY . .
RUN GOOS=$TARGETOS GOARCH=$TARGETARCH go build -o /app .

FROM gcr.io/distroless/static-debian12:nonroot
COPY --from=builder /app /app
ENTRYPOINT ["/app"]

# CI: multi-platform build
- uses: docker/build-push-action@<VERSION>
  with:
    platforms: linux/amd64,linux/arm64
    push: true
    tags: <REGISTRY>/<APP>:<TAG>

K8s deployment

spec:
  template:
    spec:
      nodeSelector:
        kubernetes.io/arch: arm64

Karpenter ile mixed nodepools

apiVersion: karpenter.sh/v1
kind: NodePool
metadata:
  name: graviton
spec:
  template:
    spec:
      requirements:
        - key: kubernetes.io/arch
          operator: In
          values: [arm64]
        - key: karpenter.k8s.aws/instance-family
          operator: In
          values: [c7g, m7g, r7g]   # Graviton 3

---

4️⃣ Spot Instance / Preemptible

Avantajlar

• %70 daha ucuz (AWS Spot, GCP Preemptible)
• Existing hardware → ek üretim gerekmez (carbon)

Uygun workload'lar

• ✅ Stateless (HTTP API)
• ✅ Batch / ML training
• ✅ CI runner
• ⚠️ DB (genelde değil — state)
• ⚠️ Cache (Redis primary değil — replica OK)

Mixed pod scheduling

# Critical: on-demand
spec:
  nodeSelector:
    karpenter.sh/capacity-type: on-demand

# Tolerant: spot
spec:
  tolerations:
    - key: karpenter.sh/capacity-type
      operator: Equal
      value: spot
      effect: NoSchedule

Karpenter spot fleet

spec:
  template:
    spec:
      requirements:
        - key: karpenter.sh/capacity-type
          operator: In
          values: [spot, on-demand]
      # Spot interruption koruması
      disruption:
        consolidationPolicy: WhenEmpty

---

5️⃣ Dev Cluster Gece Kapatma

# CronJob: 19:00 scale to 0
apiVersion: batch/v1
kind: CronJob
metadata:
  name: scale-down-dev
  namespace: kube-system
spec:
  schedule: "0 19 * * 1-5"   # weekday 19:00
  jobTemplate:
    spec:
      template:
        spec:
          serviceAccountName: scaler
          containers:
            - name: scaler
              image: bitnami/kubectl:latest
              command:
                - kubectl
                - scale
                - deployment
                - --all
                - --replicas=0
                - -n
                - dev

# CronJob: 09:00 scale up
spec:
  schedule: "0 9 * * 1-5"
  # ... --replicas=1

🔑 %60 dev cost reduction — 5 gün × 14 saat / 7 gün × 24 saat ≈ %58.

---

6️⃣ Compression (HTTP + Storage)

HTTP gzip / brotli

# Ingress-NGINX
gzip on;
gzip_types text/plain application/json application/javascript;
gzip_min_length 1000;
brotli on;
brotli_types text/plain application/json application/javascript;

→ %20-40 bandwidth tasarrufu.

Storage compression

# S3 lifecycle: gzip eski log
- Filter: prefix=logs/
- Transitions:
    - Days: 30
      StorageClass: STANDARD_IA
    - Days: 90
      StorageClass: GLACIER

DB compression

-- Postgres: column compression (TOAST)
ALTER TABLE events ALTER COLUMN payload SET COMPRESSION lz4;

---

7️⃣ CDN — Edge Caching

Origin'den uzaklaştırma = path tasarrufu = enerji tasarrufu.

Cloudflare / CloudFront

• Static asset (JS, CSS, image) → CDN
• API cache (TTL 60s) → POST cache hard
• HTML page (5min TTL acceptable)

Cache headers

Cache-Control: public, max-age=3600, s-maxage=86400, immutable

---

8️⃣ Cold Storage — Eski Log

S3 Lifecycle

LifecycleRules:
  - Filter: prefix=logs/
    Transitions:
      - Days: 30
        StorageClass: STANDARD_IA   # Infrequent Access
      - Days: 90
        StorageClass: GLACIER        # Archive
      - Days: 365
        StorageClass: DEEP_ARCHIVE
    Expiration: {Days: 2555}         # 7 yıl (compliance retention)

→ STANDARD_IA: %40 daha ucuz. GLACIER: %80 daha ucuz.

---

9️⃣ Database Cache + Read Replica

Redis önünde Postgres

• Read query'lerin %80'ini Redis cache
• Postgres CPU %30 azalır → daha küçük instance

Read replica

• Reporting / analytics → replica
• Primary CPU %20-40 azalır

---

🔟 ML Training Optimization

Daha küçük model = az enerji

• Distillation (büyük model → küçük model)
• Quantization (FP32 → INT8)
• LoRA fine-tuning (full retrain yerine)

GPU efficiency

• Mixed precision (FP16) — 2x throughput
• Multi-instance GPU sharing (MIG)
• Spot GPU (eğer interrupt OK)
• Carbon-aware scheduling (low-intensity saat)

---

📊 Quick Wins ROI Hesabı

Aksiyon                  | Effort  | Cost Save | CO₂ Save
─────────────────────────|─────────|───────────|──────────
Idle cleanup             | 1 hafta | %10-30    | %10-30
Right-sizing             | 2 hafta | %15-25    | %15-25
ARM migration            | 4 hafta | %20-40    | %20-40
Spot adoption            | 2 hafta | %30-50    | %20
Dev cron scaler          | 3 gün   | %60 (dev) | %60 (dev)
CDN                      | 1 hafta | %20-40    | %20-40
Compression              | 1 gün   | %5-15     | %5-15
Cold tier                | 1 hafta | %50-80    | %30-50
                         |         |           |
İlk 3 ayda toplam:       |         | %30-50    | %30-50

---

🚫 Anti-Pattern Tablosu

Anti-pattern	Niye kötü	Doğru
"Performance her şey, cost / carbon sonra"	Bütçe + iklim ihmal	Quarterly review
Idle resource ihmal	Aylık $X kayıp	Custodian + alarm
Spot kullanılmıyor	%70 maliyet kaybedildi	Mixed nodepool
Tüm workload x86	%30 fazla watt	ARM uygunları migrate
Dev 24/7 açık	Hafta sonu israf	Cron scaler
Log retention sonsuz	Storage cost + carbon	Lifecycle policy
ML training peak hour	Yüksek karbon	Carbon-aware scheduling
Compression kapalı	Bandwidth + enerji	gzip/brotli enable
CDN yok / yetersiz	Origin overload	CloudFront / Cloudflare
HPA agresif scale up + scale down hızlı	Pod thrashing	Stabilization window

---

📋 Efficiency Quick Wins Checklist

[ ] Idle resource scan (haftalık cron)
[ ] Right-sizing (VPA recommendation mode)
[ ] ARM/Graviton migration plan + uygun workload list
[ ] Spot instance: %30+ workload
[ ] Dev cluster cron scaler (gece + hafta sonu)
[ ] HTTP compression (gzip + brotli)
[ ] CDN: static asset + API cache
[ ] S3 lifecycle: 30/90/365 gün tier
[ ] Redis cache (read-heavy DB önünde)
[ ] Read replica (reporting / analytics)
[ ] ML mixed precision (FP16)
[ ] Carbon-aware batch scheduler
[ ] Quarterly: efficiency report (cost + CO₂)
[ ] FinOps + Sustainability shared dashboard

---

📚 Referanslar

• AWS Cost Optimization Pillar — aws.amazon.com/architecture/well-architected
• GCP Cost Optimization — cloud.google.com/architecture/framework/cost-optimization
• Cloud Custodian — cloudcustodian.io
• Karpenter — karpenter.sh
• VPA — github.com/kubernetes/autoscaler
• Kubecost — kubecost.com
• Green-Software-Principles.md
• Carbon-Aware-Computing.md
• Region-Selection.md
• Measuring-Software-Carbon.md
• 12-FinOps/Cloud-Cost-Allocation.md

---

"Efficiency 'yarın' işi değil — bu çeyrek'in kazancı. Cost ↔ carbon dual'inde her dolar tasarrufu çoğu zaman kg CO₂ tasarrufudur. Hem bütçe hem yeşil rapor."