Day 4: Scaling and Resource Management | Kubernetes in 5 Days

Resource Requests and Limits

resources in a Deployment

containers:
- name: web
  image: nginx:alpine
  resources:
    requests:
      memory: '64Mi'   # guaranteed minimum
      cpu: '100m'      # 100 millicores = 0.1 CPU
    limits:
      memory: '128Mi'  # hard cap
      cpu: '500m'      # throttled if exceeded

# CPU units: 1000m = 1 core. 100m = 0.1 core
# Memory: Mi = mebibytes, Gi = gibibytes

Horizontal Pod Autoscaler

# HPA scales Pods based on CPU or memory usage
kubectl autoscale deployment web-app \
  --min=2 --max=10 --cpu-percent=50

# Or as YAML:
apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: web-hpa
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: web-app
  minReplicas: 2
  maxReplicas: 10
  metrics:
  - type: Resource
    resource:
      name: cpu
      target:
        type: Utilization
        averageUtilization: 50

Check HPA status

kubectl get hpa
kubectl describe hpa web-hpa
# Columns: TARGETS shows current/target CPU usage
# REPLICAS shows current pod count

💡

HPA requires the Metrics Server to be running. On minikube: minikube addons enable metrics-server. In production clusters, deploy it separately: kubectl apply -f https://github.com/kubernetes-sigs/metrics-server/releases/latest/download/components.yaml.

📝 Day 4 Exercise

Set Up Autoscaling

A
d
d
r
e
s
o
u
r
c
e
r
e
q
u
e
s
t
s
a
n
d
l
i
m
i
t
s
t
o
y
o
u
r
D
e
p
l
o
y
m
e
n
t
.
E
n
a
b
l
e
m
e
t
r
i
c
s
-
s
e
r
v
e
r
o
n
m
i
n
i
k
u
b
e
.
C
r
e
a
t
e
a
n
H
P
A
t
a
r
g
e
t
i
n
g
5
0
%
C
P
U
.
U
s
e
a
l
o
a
d
t
o
o
l
(
h
e
y
,
v
e
g
e
t
a
,
o
r
a
s
i
m
p
l
e
b
a
s
h
l
o
o
p
)
t
o
g
e
n
e
r
a
t
e
t
r
a
f
f
i
c
a
n
d
w
a
t
c
h
t
h
e
H
P
A
s
c
a
l
e
u
p
.

Day 4 Summary

Always set resource requests. Without them, the scheduler can't make good placement decisions.
Requests = guaranteed resources. Limits = hard cap. CPU throttles at limit; memory gets OOMKilled.
HPA watches metrics and adds/removes Pods to meet the target utilization.
Scaling requires the Metrics Server. Don't forget to enable it.

→

← Previous

Day 3: Services and Networking

Day 5: Helm: Package Manager for K8s

→