AKS with Azure Load Balancer - Time & Space Complexity

Time Complexity: AKS with Azure Load Balancer

O(n)

Understanding Time Complexity

We want to understand how the time to set up and manage an AKS cluster with an Azure Load Balancer changes as the number of services grows.

Specifically, how does adding more services affect the number of API calls and operations?

Scenario Under Consideration

Analyze the time complexity of creating multiple services in AKS that use Azure Load Balancer.

az aks create --resource-group myResourceGroup --name myAKSCluster --node-count 3 --enable-addons monitoring --generate-ssh-keys

for i in range(1, n+1):
    az aks service create --name service{i} --cluster-name myAKSCluster --load-balancer

This sequence creates an AKS cluster and then creates n services, each with its own Azure Load Balancer configuration.

Identify Repeating Operations

Identify the API calls, resource provisioning, data transfers that repeat.

• Primary operation: Creating a service with a load balancer involves an API call to provision the load balancer and configure it.
• How many times: This happens once per service, so n times.

How Execution Grows With Input

Each new service adds one load balancer provisioning operation, so the total operations grow directly with the number of services.

Input Size (n)	Approx. Api Calls/Operations
10	10 load balancer provisioning calls
100	100 load balancer provisioning calls
1000	1000 load balancer provisioning calls

Pattern observation: The number of operations grows linearly as the number of services increases.

Final Time Complexity

Time Complexity: O(n)

This means the time to create and configure load balancers grows directly in proportion to the number of services.

Common Mistake

[X] Wrong: "Adding more services does not increase load balancer provisioning time because Azure handles it automatically in the background."

[OK] Correct: Each service requires its own load balancer setup, which involves separate API calls and resource provisioning, so time grows with the number of services.

Interview Connect

Understanding how resource provisioning scales helps you design efficient cloud architectures and explain your reasoning clearly in interviews.

Self-Check

What if we changed to using a single shared load balancer for all services? How would the time complexity change?

Practice

1. What is the primary purpose of using an Azure Load Balancer with AKS (Azure Kubernetes Service)?

easy

A. To store data persistently for containers

B. To distribute incoming network traffic evenly across multiple pods

C. To build container images automatically

D. To monitor container resource usage

Solution

1. Step 1: Understand AKS and Load Balancer roles

   AKS runs containerized apps, and Azure Load Balancer distributes traffic to these apps.

2. Step 2: Identify the main function of Load Balancer

   It balances incoming requests across pods to improve availability and scalability.

3. Final Answer:

   To distribute incoming network traffic evenly across multiple pods -> Option B

4. Quick Check:

   Load Balancer = traffic distribution [OK]

Hint: Load Balancer = spreading traffic evenly [OK]

Common Mistakes:

• Confusing Load Balancer with storage or monitoring
• Thinking Load Balancer builds container images
• Assuming Load Balancer manages pod resources

2. Which Kubernetes service type should you specify in your AKS deployment YAML to create an Azure Load Balancer automatically?

easy

A. LoadBalancer

B. NodePort

C. ClusterIP

D. ExternalName

Solution

1. Step 1: Review Kubernetes service types

   ClusterIP exposes service internally, NodePort exposes on node port, LoadBalancer creates cloud LB, ExternalName maps to external DNS.

2. Step 2: Identify service type for Azure Load Balancer

   Using type: LoadBalancer triggers Azure to provision a Load Balancer automatically.

3. Final Answer:

   LoadBalancer -> Option A

4. Quick Check:

   Service type LoadBalancer = Azure LB creation [OK]

Hint: Use type LoadBalancer to get Azure LB automatically [OK]

Common Mistakes:

• Choosing ClusterIP which is internal only
• Confusing NodePort with automatic LB creation
• Using ExternalName which is DNS mapping only

3. Given this Kubernetes service YAML snippet in AKS:

apiVersion: v1
kind: Service
metadata:
  name: myapp-service
spec:
  type: LoadBalancer
  selector:
    app: myapp
  ports:
  - protocol: TCP
    port: 80
    targetPort: 8080

What happens when this service is applied?

medium

A. An Azure Load Balancer is created and routes port 80 traffic to pods on port 8080

B. Pods are exposed only inside the cluster on port 8080

C. Traffic on port 8080 is blocked by default

D. A NodePort service is created exposing port 80 on all nodes

Solution

1. Step 1: Analyze service type and ports

   Service type is LoadBalancer, so Azure LB is created. It listens on port 80 externally and forwards to targetPort 8080 on pods.

2. Step 2: Understand traffic flow

   External traffic on port 80 hits Azure LB, which routes it to pods' port 8080 matching selector app: myapp.

3. Final Answer:

   An Azure Load Balancer is created and routes port 80 traffic to pods on port 8080 -> Option A

4. Quick Check:

   LoadBalancer + port mapping = external traffic routing [OK]

Hint: LoadBalancer routes external port to pod targetPort [OK]

Common Mistakes:

• Thinking pods are exposed only internally
• Confusing NodePort with LoadBalancer
• Assuming traffic is blocked without explicit rules

4. You deployed an AKS service with type: LoadBalancer, but the external IP remains <pending> for a long time. What is the most likely cause?

medium

A. The service selector labels do not match any pods

B. The Kubernetes cluster is not running

C. The pods are not listening on the targetPort

D. The Azure Load Balancer quota is exceeded in the subscription

Solution

1. Step 1: Understand LoadBalancer IP allocation

   Azure assigns an external IP when provisioning the Load Balancer. If quota is exceeded, IP remains pending.

2. Step 2: Differentiate causes

   Selector mismatch or pod ports cause traffic issues but do not block IP assignment. Cluster down would prevent service creation.

3. Final Answer:

   The Azure Load Balancer quota is exceeded in the subscription -> Option D

4. Quick Check:

   Pending IP often means quota limit reached [OK]

Hint: Pending IP usually means Azure LB quota exceeded [OK]

Common Mistakes:

• Blaming selector mismatch for IP assignment delay
• Assuming pods not listening blocks IP allocation
• Thinking cluster down still allows service creation

5. You want to design a highly available AKS application exposed via Azure Load Balancer that can handle sudden traffic spikes. Which combination of strategies is best?

hard

A. Use type: NodePort service and rely on Azure VM scale sets only

B. Use type: ClusterIP service with manual pod scaling and no health probes

C. Use type: LoadBalancer service, enable Horizontal Pod Autoscaler, and configure Azure Load Balancer health probes

D. Use type: ExternalName service pointing to an external DNS

Solution

1. Step 1: Choose correct service type for external exposure

   type: LoadBalancer creates Azure LB to distribute traffic externally.

2. Step 2: Enable autoscaling and health checks

   Horizontal Pod Autoscaler adjusts pod count for traffic spikes; health probes ensure LB routes only to healthy pods.

3. Step 3: Evaluate other options

   ClusterIP is internal only; NodePort exposes ports but lacks automatic LB; ExternalName is DNS mapping, not load balancing.

4. Final Answer:

   Use type: LoadBalancer service, enable Horizontal Pod Autoscaler, and configure Azure Load Balancer health probes -> Option C

5. Quick Check:

   LoadBalancer + autoscale + health probes = high availability [OK]

Hint: Combine LoadBalancer, autoscaling, and health probes for HA [OK]

Common Mistakes:

• Using ClusterIP or ExternalName for external traffic
• Ignoring autoscaling for traffic spikes
• Not configuring health probes causing downtime