Chapter 10: Advanced Scheduling

The Problem This Chapter Solves

Not every bus can go to every depot. Some depots have special requirements. Some buses need special facilities. Some routes work better when buses are spread across multiple depots (so one depot failure does not kill the entire route).

Kubernetes has a rich system for controlling exactly where Pods are placed.

---

Part 1: Resource Management

Resource Requests

When you create a Pod, you can tell Kubernetes: "This Pod needs at least 2 CPUs and 4GB of RAM to run properly."

This is a Resource Request. Kubernetes uses it when scheduling — it will only place the Pod on a Node that has at least that much available.

BMTC Analogy: Reserved resources for a bus.

"Bus route 500D requires: 1 dedicated bay at the depot, fuel allocation for 200km, 1 assigned driver."

The operations manager only assigns this bus to a depot that can meet these requirements.

Resource Limits

A Resource Limit sets the maximum a Pod can use. Even if more is available, the Pod cannot exceed this limit.

BMTC Analogy: Maximum allowed fuel consumption.

A bus is allocated fuel for 200km. Even if there is more fuel available at the depot, this bus cannot take more than its limit. This ensures resources are fairly distributed.

Resource Request = Minimum needed (guaranteed)
Resource Limit   = Maximum allowed (cannot exceed)

# resources.yaml
apiVersion: v1
kind: Pod
metadata:
  name: bus-app
spec:
  containers:
    - name: app
      image: my-app:latest
      resources:
        requests:
          memory: "512Mi"
          cpu: "500m"
        limits:
          memory: "1Gi"
          cpu: "1000m"

# Check resource usage of Nodes
kubectl top nodes

# Check resource usage of Pods
kubectl top pods

# Describe a Node to see its capacity and available resources
kubectl describe node <node-name>

---

Part 2: Controlling Pod Placement

Taints and Tolerations

Taints are marks on a Node that say: "Normal Pods cannot be scheduled here."

Tolerations are marks on a Pod that say: "I can handle this taint. I am allowed to go there."

BMTC Analogy:

Taint: Yeshwanthpur Charging Depot puts up a sign: "Only Electric Buses Allowed." Regular diesel buses cannot enter.

Toleration: The electric bus has the right credentials. It can enter the charging depot. The diesel bus does not have a toleration for this taint — it stays out.

Taint on Node       =  "Electric buses only" sign on depot
Toleration on Pod   =  Electric bus credentials that allow entry

# Add a taint to a Node
kubectl taint nodes node1 bus-type=electric:NoSchedule

# View taints on a Node
kubectl describe node node1 | grep Taint

# Remove a taint
kubectl taint nodes node1 bus-type=electric:NoSchedule-

# toleration.yaml — Pod that tolerates the taint
apiVersion: v1
kind: Pod
metadata:
  name: electric-bus
spec:
  tolerations:
    - key: "bus-type"
      operator: "Equal"
      value: "electric"
      effect: "NoSchedule"
  containers:
    - name: app
      image: my-app

Node Selector

The simplest placement rule. You say: "Put this Pod only on Nodes with this specific label."

BMTC Analogy: "Assign this bus to Whitefield Depot only." The operations manager does not consider any other depot for this bus.

# node-selector.yaml
apiVersion: v1
kind: Pod
metadata:
  name: whitefield-bus
spec:
  nodeSelector:
    depot: whitefield
  containers:
    - name: app
      image: my-app

Affinity and Anti-Affinity

Affinity: "Try to place these Pods near each other (on the same Node or nearby Nodes)."

Anti-Affinity: "Try to keep these Pods away from each other."

BMTC Analogy:

Affinity: "Keep the ticket booking buses and the payment processing buses in the same depot so they can communicate quickly."

Anti-Affinity: "Do not park all Route 500D buses in the same depot. Spread them across Shivajinagar, Yeshwanthpur, and Bannerghatta depots. If one depot has a problem, we still have buses at the other two."

Anti-Affinity is especially important for high availability — spreading your application across multiple Nodes so one Node failure does not take down your entire application.

---

Advanced Scheduling Summary

Term	BMTC Meaning	Kubernetes Meaning
Resource Request	Minimum fuel, bay, driver needed	Minimum CPU/Memory guaranteed
Resource Limit	Maximum fuel allowed	Maximum CPU/Memory the Pod can use
Taint	"Electric buses only" depot sign	Node restriction marker
Toleration	Electric bus credentials	Pod's permission to enter tainted Node
Node Selector	"Whitefield Depot only"	Place Pod on specific labeled Node
Affinity	Keep buses together in same depot	Place Pods near each other
Anti-Affinity	Spread buses across depots	Keep Pods apart for high availability

---

❓ Quick Quiz

4 questionsStudy Mode

Question 1: A Pod requests 512MB of memory. What does this mean?

Question 2: A Node has a taint 'gpu=true:NoSchedule'. What happens to Pods without a matching toleration?

Question 3: Why would you use Anti-Affinity for your application Pods?

Question 4: What is the difference between a Resource Request and a Resource Limit?