DEV Community

Aisalkyn Aidarova
Aisalkyn Aidarova

Posted on • Edited on

High availability and scalability ELB and ASG

1. Scalability

Scalability means an application can handle increased load by adapting resources.

Types of Scalability

  • Vertical Scalability (Scale Up / Down):

    • Increase instance size (more CPU, RAM, etc.).
    • Example: Upgrading an EC2 instance from t2.micro → t2.large.
    • Common for non-distributed systems like databases (RDS, ElastiCache).
    • Limited by hardware capacity.

  • Horizontal Scalability (Scale Out / In):

    • Add more instances to share the load.
    • Example: Add more EC2 instances behind a Load Balancer.
    • Common for distributed systems and web applications.
    • In AWS: use Auto Scaling Groups.

2. High Availability (HA)

High availability means keeping the system operational even if one part fails.

  • Achieved by running applications in multiple Availability Zones (AZs).
  • Ensures resilience to data center or zone failure.

  • Example:

    • EC2 Auto Scaling Group across 2+ AZs.
    • RDS Multi-AZ (primary + standby replica).

  • Can be:

    • Active-Passive: standby server takes over on failure.
    • Active-Active: all instances handle traffic simultaneously.

3. Scalability vs. High Availability

Concept Purpose AWS Example
Scalability Handle increased load Auto Scaling (scale in/out)
High Availability Survive component failure Multi-AZ Deployment

4. Call Center Analogy

Concept Example
Vertical scaling One operator becomes faster (junior → senior)
Horizontal scaling Hire more operators
High availability Operators distributed in multiple cities

🧭 What Is Load Balancing?

Definition:
Load balancing is the process of distributing incoming network traffic across multiple backend servers (EC2 instances) to ensure no single instance is overloaded.

In AWS, this is handled by Elastic Load Balancing (ELB) — a fully managed service that automatically distributes traffic across targets such as EC2 instances, containers, IP addresses, and Lambda functions.

⚙️ How It Works

  • Users connect to a single endpoint — the load balancer’s DNS name.
  • The load balancer forwards each request to one of several healthy backend instances.
  • Traffic is distributed based on a routing algorithm (round-robin, least connections, etc.).
  • Health checks continuously monitor backend instances to ensure only healthy ones receive traffic.

🧩 Why Use a Load Balancer

Benefit Description
Single Access Point Users access your app through one endpoint (DNS).
Fault Tolerance Automatically removes unhealthy instances.
Scalability Easily add or remove instances without downtime.
SSL Termination Manage HTTPS certificates at the load balancer level.
Sticky Sessions (Cookies) Keep users on the same instance if needed.
Multi-AZ High Availability Distribute traffic across Availability Zones.
Security Segregation Public-facing LB with private backend instances.

🩺 Health Checks

Health checks ensure traffic goes only to healthy targets.

Example configuration:

Protocol: HTTP
Port: 4567
Path: /health
Enter fullscreen mode Exit fullscreen mode

If the endpoint returns an HTTP 200 OK, the instance is healthy.
If not, it’s marked unhealthy, and ELB stops routing traffic to it.

🧱 Types of Elastic Load Balancers

Type Year Protocols Supported Layer Use Case
Classic Load Balancer (CLB) 2009 HTTP, HTTPS, TCP, SSL L4 + L7 Legacy apps (deprecated)
Application Load Balancer (ALB) 2016 HTTP, HTTPS, WebSocket L7 Modern web apps, microservices
Network Load Balancer (NLB) 2017 TCP, TLS, UDP L4 High performance, low latency
Gateway Load Balancer (GWLB) 2020 IP L3 Firewalls, security appliances

✅ Use ALB or NLB for most modern AWS applications.

🔐 Security Groups Setup

1. Load Balancer Security Group

Inbound:
  Port: 80, 443
  Source: 0.0.0.0/0 (public access)
Enter fullscreen mode Exit fullscreen mode

2. EC2 Instance Security Group

Inbound:
  Port: 80
  Source: Load Balancer Security Group (not an IP range)
Enter fullscreen mode Exit fullscreen mode

This ensures only the load balancer can reach the backend instances.

🔗 Integrations

Load balancers work seamlessly with:

  • EC2 Auto Scaling Groups
  • ECS / EKS containers
  • AWS Certificate Manager (ACM)
  • CloudWatch (monitoring)
  • Route 53 (DNS)
  • AWS WAF (security)
  • Global Accelerator (performance)

🧠 Quick Recap

  • Load balancers distribute traffic → scalability
  • Run across multiple AZs → high availability
  • Health checks + security groups → resilience and safety
  • Choose ALB or NLB depending on your app layer (HTTP vs TCP/UDP).

🎯 Application Load Balancer (ALB)

Layer 7 Load Balancer (HTTP/HTTPS)
→ Operates at the Application Layer (OSI Layer 7).
→ Understands HTTP headers, URLs, and cookies, enabling smart routing decisions.

🔍 Key Features

Feature Description
Layer 7 (HTTP/HTTPS) Routes based on HTTP methods, headers, hostnames, paths, query strings.
Multiple Applications, One ALB One ALB can handle multiple apps or microservices (e.g., /user, /search).
Target Groups Logical grouping of backend targets (EC2, ECS, Lambda, or IPs).
Advanced Routing - Path-based routing (/user, /search)
- Host-based routing (api.example.com, admin.example.com)
- Query string or header-based routing.
Redirects Redirect HTTP → HTTPS automatically.
Protocol Support HTTP/1.1, HTTP/2, WebSockets.
Port Mapping (ECS) Dynamically route to container ports (used with ECS).
Fixed DNS Name AWS assigns a static hostname for each ALB.
Health Checks Done per target group.

🧩 Target Groups

A Target Group defines where the load balancer sends traffic.

Targets can be:

  • EC2 Instances (managed manually or via Auto Scaling)
  • ECS Tasks (containers)
  • Lambda Functions
  • Private IP Addresses (for hybrid or on-premises servers)

💡 Health checks are configured per target group.
If targets fail health checks, ALB automatically stops sending them traffic.

🧠 Routing Example (Microservices) 

            ┌──────────────────────────┐
            │   Application Load Balancer│
            └────────────┬─────────────┘
                         │
         ┌───────────────┼────────────────┐
         │                                │
  /user  │                                │  /search
(Target Group 1)                   (Target Group 2)
   EC2: User Service               EC2: Search Service
Enter fullscreen mode Exit fullscreen mode
  • /user → routed to Target Group 1
  • /search → routed to Target Group 2
  • Both microservices share the same ALB endpoint

🧠 Routing by Query or Headers

Example rule:

If query string = Platform=Mobile → Target Group 1
If query string = Platform=Desktop → Target Group 2
Enter fullscreen mode Exit fullscreen mode

This allows smart routing based on request parameters or headers.

🧾 X-Forwarded Headers

When traffic passes through an ALB,
the backend does not see the original client IP directly.

AWS adds extra HTTP headers:

Header Description
X-Forwarded-For Original client IP address
X-Forwarded-Port Port used by client
X-Forwarded-Proto Protocol (HTTP or HTTPS)

Backend servers use these headers to log or identify the true client source.

🔒 Security Overview

  • ALB is public-facing (port 80/443).
  • Backend EC2 instances allow only traffic from the ALB’s security group.
  • This ensures end users can’t directly access your backend servers.

🧠 ALB vs. Classic Load Balancer (CLB)

Feature Classic LB Application LB
Layer 4 & 7 7 only
Routing Basic Path/Host/Header-based
Multi-App Support ❌ ✅
WebSockets/HTTP/2 ❌ ✅
Container Support ❌ ✅
Cost Efficiency Multiple LBs One ALB for many apps

🧮 Use Cases

✅ Microservices architecture
✅ Container workloads (ECS / EKS)
✅ Web applications requiring intelligent routing
✅ Hybrid apps (some on AWS, some on-premises)
✅ HTTPS redirection and centralized SSL management

⚙️ Lab: Launching an Application Load Balancer

Objective

Create two EC2 instances serving simple web pages and distribute traffic between them using an Application Load Balancer (ALB).

🪜 Step 1: Launch Two EC2 Instances

  1. Go to EC2 → Launch Instances

  2. Name:

  • Instance 1 → My First Instance
  • Instance 2 → My Second Instance

  1. AMI: Ubuntu

  2. Instance Type: t2.micro (Free-tier)

  3. Key Pair: Proceed without key pair (we’ll use EC2 Instance Connect)

  4. Network Settings:

  • Select existing security group launch-wizard-1

  • Ensure inbound rules allow:

    • HTTP (port 80) from anywhere
    • SSH (port 22) from your IP (optional)
  1. User Data Script: Paste the following script to auto-start a simple web server: 
# Update the package index
sudo apt update -y

# Install Nginx (package name must be lowercase)
sudo apt install -y nginx

# Start and enable Nginx service
sudo systemctl start nginx
sudo systemctl enable nginx

# Create a simple web page
echo "<h1>Hello World from $(hostname -f)</h1>" | sudo tee /var/www/html/index.html
Enter fullscreen mode Exit fullscreen mode
  1. Launch 2 instances and wait for both to reach running and Status check: 2/2 passed.

Test:
Copy each instance’s IPv4 public address and open it in a browser —
you should see:
Hello World from ip-xxx-xxx-xxx-xxx.ec2.internal

🌐 Step 2: Create an Application Load Balancer

  1. Go to EC2 → Load Balancers → Create Load Balancer
  2. Choose Application Load Balancer (ALB).
  3. Name: DemoALB
  4. Scheme: Internet-facing Address Type: IPv4
  5. Network Mapping:
  • Select your VPC.

  • Enable all available Availability Zones.

    1. Security Group:

  • Create new one named demo-sg-load-balancer

  • Allow Inbound:

    • Type: HTTP
    • Port: 80
    • Source: 0.0.0.0/0

  • Outbound: default (allow all)

    1. Listeners and Routing:

  • Protocol: HTTP

  • Port: 80

  • Create a new Target Group named demo-tg-alb

🎯 Step 3: Create Target Group

  1. Target Type: Instances
  2. Protocol: HTTP
  3. Port: 80
  4. Health Check Path: / (default OK)
  5. Click Next → Select both EC2 instances → Include as pending → Create target group

✅ Your target group now includes both instances.

🔄 Step 4: Attach Target Group to ALB

  • In ALB creation wizard → choose target group demo-tg-alb.
  • Review and click Create Load Balancer.
  • Wait for ALB State: active.

🔗 Step 5: Test the Load Balancer

  1. Copy the DNS name of your ALB (e.g., DemoALB-123456789.us-east-1.elb.amazonaws.com).
  2. Paste it into a browser.

You should see:
Hello World from ip-xx-xx-xx-xx.ec2.internal

🔁 Refresh repeatedly:
Each refresh alternates between your two EC2 instances — proof of load balancing!

🩺 Step 6: Verify Health Checks

  1. Go to Target Groups → demo-tg-alb → Targets tab You should see: 
   i-xxxxx  healthy
   i-yyyyy  healthy
Enter fullscreen mode Exit fullscreen mode
  1. Stop one instance → Wait ~30 seconds → Refresh the target group page. The stopped instance becomes: 
   i-xxxxx  unused (unhealthy)
Enter fullscreen mode Exit fullscreen mode

The ALB automatically stops sending traffic to it.

  1. Restart the instance → Wait for health status to become healthy again → both instances resume receiving traffic.

✅ Result

  • One ALB endpoint distributing traffic evenly across two EC2 instances.
  • Automatic detection of unhealthy instances.
  • Seamless recovery once instances come back online.

💡 Key Concepts Learned

Concept Meaning
Target Group Logical grouping of backend instances
Health Check Regular status probe (HTTP 200 OK) to detect healthy targets
Listener Defines protocol/port and routing rules (e.g., HTTP:80 → Target Group)
DNS Name Single entry point for users
Security Model ALB accepts public traffic; EC2 only from ALB SG

⚙️ Advanced Concepts – Application Load Balancer (ALB)

🧱 1. Network Security Enhancement

🧩 Current Setup

  • Load Balancer SG: demo-sg-load-balancer — allows inbound HTTP (80) from anywhere.
  • EC2 Instance SG: launch-wizard-1 — currently allows inbound HTTP (80) from anywhere too.

This means users can bypass the ALB and directly access EC2 instances — not ideal for production.

🔒 Goal

✅ Allow only the load balancer to access EC2 instances.
❌ Block direct public access to backend servers.

🪜 Steps to Tighten Security

  1. Go to EC2 → Security Groups → launch-wizard-1
  2. Click Edit inbound rules
  3. Delete the rule that allows: 
   Type: HTTP
   Source: 0.0.0.0/0
Enter fullscreen mode Exit fullscreen mode
  1. Add a new rule: 
   Type: HTTP
   Source: demo-sg-load-balancer
Enter fullscreen mode Exit fullscreen mode

(Select the security group of your ALB)

  1. Click Save rules

🔍 Result

  • Accessing EC2 instance directly via public IP → ❌ Connection timed out
  • Accessing via Load Balancer DNS → ✅ Works perfectly!

This ensures traffic must pass through the ALB, achieving layered security.

🧠 2. Listener Rules in ALB

Listeners define how incoming requests are routed.
Each listener listens on a port (e.g., 80/443) and processes rules top-to-bottom by priority.

🧩 Default Rule

“For every request → forward to target group demo-tg-alb.”

Now we’ll add custom rules to handle specific request paths or headers.

🪜 Steps to Add a Custom Rule

  1. Go to your ALB → Listeners → View/Edit Rules
  2. Add a New Rule Name: DemoRule
  3. Add Condition:
  • Type: Path

  • Value: /error

    1. Action:

  • Choose Fixed Response

  • Status Code: 404

  • Content Type: text/plain

  • Response Body: Not found – custom error

    1. Set Priority: 5 (lower numbers = higher priority)
    2. Save changes

Now your rules look like:

Priority Condition Action
5 Path = /error Fixed Response (404)
Default All requests Forward to demo-tg-alb

✅ Test Your Rule

  1. Copy the ALB DNS name, e.g.: 
   http://DemoALB-123456789.us-east-1.elb.amazonaws.com
Enter fullscreen mode Exit fullscreen mode
  1. Visit:
  • / → returns Hello World (from EC2 target group)
  • /error → returns Not found – custom error

💡 The /error path matched your new listener rule, so ALB responded directly with 404 instead of forwarding the request to instances.

🧩 3. Rule Priority System

  • ALB processes listener rules in order of priority (lowest = highest precedence).
  • Example:

| Priority | Rule |
| -------- | ----------------------------- |
| 1 | /admin → Admin Target Group |
| 5 | /error → Fixed Response |
| Default | → Main Target Group |

If multiple rules match, the rule with the lowest priority number wins.

💡 Key Takeaways

Concept Description
SG to SG reference Restricts EC2 access to only the ALB
Listener Rules Define conditional routing for requests
Fixed Response Send custom responses without touching backend
Priorities Determine which rule executes first
Path/Host/Header conditions Enable advanced routing for microservices

🏁 Summary

  • Secured backend instances (no direct internet access).
  • Implemented custom ALB listener rules (path-based routing).
  • Learned rule priority and conditional logic.
  • Demonstrated fixed response (useful for error pages or maintenance).

⚙️ Network Load Balancer (NLB)

Layer 4 Load Balancer — handles TCP and UDP traffic.
✅ Designed for ultra-high performance and low latency.

🧠 1. Core Concepts

Feature Description
Layer Layer 4 – Transport layer (TCP/UDP). Works below HTTP layer.
Protocols Supported TCP, UDP, TLS
Performance Millions of requests per second with microsecond latency.
Static IPs One static IP per Availability Zone (can assign Elastic IPs).
High Availability Multi-AZ support with failover.
Health Checks Supports TCP, HTTP, or HTTPS-based health checks.

🧩 2. When to Use NLB

Scenario Why NLB?
High performance needed Can handle millions of requests/sec.
Static IP requirement Use Elastic IPs for predictable endpoints.
Non-HTTP traffic Supports raw TCP/UDP connections (e.g., SMTP, RDP, DNS, gaming servers).
Hybrid network setup Can target on-prem servers via private IPs.
Load balance HTTP apps with fixed IPs Use NLB in front of ALB (NLB → ALB).

💡 Exam Tip:
If the question mentions:

  • “Static IPs required”
  • “Extreme performance”
  • “TCP or UDP traffic” → Answer: Network Load Balancer

🏗️ 3. Architecture Overview 

            ┌──────────────────────────────┐
            │      Network Load Balancer    │
            └─────────────┬────────────────┘
                          │
            ┌─────────────┴──────────────┐
            │                            │
     Target Group A                Target Group B
     (EC2 Instances)              (Private IPs / On-Prem)
Enter fullscreen mode Exit fullscreen mode
  • Frontend listener: TCP or UDP port (e.g., 80, 443, 25)
  • Backend targets: EC2 instances or private IP addresses
  • Health checks: TCP/HTTP/HTTPS
  • One static IP per AZ (Elastic IP optional)

🔗 4. Target Groups

Types of targets NLB can route to:

  • EC2 Instances (same VPC)
  • Private IPs (cross-network or on-prem)
  • Other Load Balancers (e.g., ALB for HTTP logic)

💡 Combo Pattern:
NLB (for static IPs) → ALB (for routing rules).
This combines Layer 4 and Layer 7 benefits.

🩺 5. Health Checks

NLB Target Groups support 3 protocols for health checks:

Protocol Use Case
TCP Simple connectivity check (fast, low overhead).
HTTP Checks for valid HTTP 200 OK response.
HTTPS Secure health checks for encrypted endpoints.

If a target fails its health check, NLB stops sending traffic until it’s healthy again.

🔒 6. Key Differences: ALB vs NLB

Feature Application Load Balancer (ALB) Network Load Balancer (NLB)
Layer Layer 7 (HTTP/HTTPS) Layer 4 (TCP/UDP/TLS)
Protocol Awareness Understands URLs, headers Works with raw packets
Static IPs ❌ No ✅ Yes
Performance High Extremely high
Use Case Web apps, microservices Gaming, databases, IoT, legacy TCP apps
Health Checks HTTP/HTTPS only TCP/HTTP/HTTPS
Integration ECS, Lambda Hybrid network, ALB chaining

🧾 Example Exam Question

“Your company must expose a TCP-based financial application that requires static IPs for firewall whitelisting and must handle millions of requests per second.”
✅ Answer: Network Load Balancer (NLB)

🧠 7. Summary

  • Layer 4 load balancer for TCP/UDP traffic.
  • Supports static IPs (Elastic IPs per AZ).
  • High performance + low latency.
  • Can sit in front of ALB for combined benefits.
  • Health checks: TCP, HTTP, HTTPS.

⚙️ Hands-On: Creating a Network Load Balancer (NLB)

🎯 Objective

Create a Network Load Balancer (Layer 4) to distribute TCP traffic between two EC2 instances and understand NLB-specific setup, routing, and security.

🪜 Step 1: Create the Network Load Balancer

  1. Go to EC2 → Load Balancers → Create Load Balancer
  2. Select Network Load Balancer
  3. Name: DemoNLB
  4. Scheme: Internet-facing Address type: IPv4
  5. Network Mapping:
  • Select your VPC
  • Enable all Availability Zones (AZs)
  • Each AZ will automatically get one static IPv4 address → You can replace these with Elastic IPs for fixed public IPs.

💡 Tip: Each AZ = one static IP → excellent for firewall whitelisting or static endpoint access.

🧱 Step 2: Create a Security Group for the NLB

  1. Click Create security group
  • Name: demo-sg-nlb
  • Description: Allow HTTP into NLB
    1. Add Inbound Rule: 
   Type: HTTP
   Port: 80
   Source: 0.0.0.0/0
Enter fullscreen mode Exit fullscreen mode
  1. Keep default outbound rule (allow all).
  2. Save and attach demo-sg-nlb to the load balancer.

✅ Purpose: Controls inbound traffic into the NLB (acts as a front gate).

🧩 Step 3: Configure Listener & Target Group

  1. Under Listeners and Routing:
  • Protocol: TCP

  • Port: 80

    1. Create Target Group:

  • Name: demo-tg-nlb

  • Target type: Instances

  • Protocol: TCP

  • Port: 80

  • VPC: Select the same one used above

  • Health check protocol: HTTP
    (because your EC2 web servers respond to HTTP)

  • Advanced health check settings:

     Healthy threshold: 2
 Timeout: 2 seconds
 Interval: 5 seconds
  1. Register targets: Select both EC2 instances → Include as pending → Create Target Group.

✅ Target group created → ready for backend connections.

🌐 Step 4: Finalize NLB Setup

  • Back on NLB creation screen → Refresh Target Group list.
  • Choose demo-tg-nlb as backend.
  • Review configuration → Create Load Balancer.
  • Wait until State: active.

🧠 Step 5: Troubleshooting (Unhealthy Targets)

If targets show as Unhealthy, check:

  1. Go to Target Group → Targets tab → status: unhealthy
  2. Verify EC2 security group inbound rules:
  • It currently allows HTTP only from demo-sg-alb (Application LB)
  • You must also allow traffic from NLB SG

✅ Fix:

  1. Go to EC2 → Security Groups → launch-wizard-1
  2. Edit inbound rules:

  • Add new rule:

     Type: HTTP
 Source: demo-sg-nlb

  • (Keep the old “allow from ALB” rule if both are in use)

    1. Save rules.

🔁 Step 6: Verify Load Balancing

  1. Wait ~30 seconds for health checks to pass → status becomes Healthy.
  2. Copy the DNS name of your NLB (e.g., DemoNLB-123456.us-east-1.elb.amazonaws.com)
  3. Paste in browser → you should see: 
   Hello World from ip-xxx-xxx-xxx-xxx
Enter fullscreen mode Exit fullscreen mode
  1. Refresh multiple times — each refresh switches between instance responses.

✅ Result:
Traffic alternates between your two EC2 instances — proving TCP load balancing works.

🧹 Step 7: Clean Up

To avoid charges:

  1. Delete DemoNLB
  2. Optionally delete:
  • Target group demo-tg-nlb
  • Security group demo-sg-nlb

🧩 Key Takeaways

Concept Explanation
Layer 4 (TCP/UDP) NLB operates below HTTP, ideal for raw network traffic
Static IPs / Elastic IPs One per AZ, ideal for firewall rules
Health Checks Can use TCP, HTTP, or HTTPS
Security Must explicitly allow NLB SG in backend EC2 SG
Performance Millions of requests per second, microsecond latency
Troubleshooting Tip “Unhealthy” targets usually mean blocked inbound rules

💡 Exam + Real-World Insight

  • If a system requires static IPs → Use NLB.
  • If traffic uses TCP/UDP → Use NLB.
  • For HTTP-based routing → Use ALB.
  • For hybrid setups → Chain them: NLB → ALB → EC2.

⚙️ Gateway Load Balancer (GWLB)

Layer 3 Load Balancer (Network Layer – IP)

🧠 1. Concept Overview

Purpose:
The Gateway Load Balancer (GWLB) allows you to deploy, scale, and manage third-party network appliances such as:

  • Firewalls
  • Intrusion Detection/Prevention Systems (IDS/IPS)
  • Deep Packet Inspection (DPI) tools
  • Network traffic analyzers
  • Custom packet filters or payload modifiers

✅ It makes all traffic in your VPC pass through these appliances transparently before reaching applications.

🧩 2. Key Idea: Transparent Traffic Inspection

Without GWLB:
User traffic → Application Load Balancer → EC2 app directly.

With GWLB:
User traffic → Gateway Load Balancer → Virtual Appliances (firewalls, IDS) → GWLB → ALB/Application.

🌐 3. How It Works (Flow Diagram) 

Users 
  ↓
Gateway Load Balancer (GWLB)
  ↓
Target Group (Virtual Appliances – Firewalls, IDS)
  ↓
GWLB (return path)
  ↓
Application Load Balancer / EC2 Application
Enter fullscreen mode Exit fullscreen mode
  • GWLB intercepts all traffic.
  • Routes traffic to your security appliances for inspection.
  • Appliances accept (forward) or drop (block) packets.
  • If accepted, traffic returns through GWLB to your app.
  • For applications, this process is completely transparent.

🧱 4. Technical Characteristics

Feature Description
Layer Layer 3 (Network Layer – IP)
Protocol Uses GENEVE protocol (port 6081) for encapsulation
Gateway Function Acts as a transparent network gateway — single ingress & egress point
Load Balancing Function Distributes traffic across appliance targets
Target Group Types EC2 instances or private IP addresses
Supported Targets - AWS-hosted EC2 appliances
- On-premise devices (via private IPs)
Routing Integration Automatically updates VPC route tables to redirect traffic through GWLB

🧰 5. Use Cases

Use Case Example
Firewalling All VPC inbound/outbound traffic passes through a firewall fleet
Intrusion Detection (IDS/IPS) Detect and block malicious packets before app layer
Deep Packet Inspection Analyze packet contents for compliance or malware
Hybrid Security Inspect traffic between AWS and on-prem environments

🧠 6. Exam Tips (AWS Certified Solutions Architect / SysOps)

Look for these keywords to identify a Gateway Load Balancer question:

Keyword Points To
“GENEVE protocol” or “port 6081” Gateway Load Balancer
“Third-party firewall/IDS/DPI appliances” Gateway Load Balancer
“All traffic must be inspected before reaching apps” Gateway Load Balancer
“Layer 3 load balancing” Gateway Load Balancer
“Transparent routing or VPC route modification” Gateway Load Balancer

🧩 7. Summary

Layer Protocols Purpose Typical Targets Example Services
Layer 3 – Gateway Load Balancer (GWLB) IP (GENEVE port 6081) Inspect or filter all network traffic Firewalls, IDS/IPS, DPI Palo Alto, Fortinet, Check Point
Layer 4 – Network Load Balancer (NLB) TCP/UDP/TLS Raw network traffic distribution EC2, private IPs Databases, games, IoT
Layer 7 – Application Load Balancer (ALB) HTTP/HTTPS Web traffic routing and microservices EC2, ECS, Lambda Web apps, APIs

🧩 8. Key Takeaway

Gateway Load Balancer = Gateway + Load Balancer

  • Gateway: Single ingress/egress for traffic in VPC.
  • Load Balancer: Distributes traffic among virtual appliances.
  • Use GENEVE protocol (port 6081).
  • Main use: Routing all traffic through inspection or firewall systems.

🧠 Remember for the Exam

If you see:

“Traffic must go through a firewall or IDS before reaching application servers,”
“GENEVE protocol,”
“Transparent inspection,”

✅ The answer is Gateway Load Balancer (GWLB).

🍪 Sticky Sessions (Session Affinity) in AWS Load Balancers

🧠 1. Concept Overview

Definition:
Sticky sessions (also called session affinity) ensure that a user is consistently routed to the same backend instance for all requests during a session.

Without stickiness:
Requests are distributed evenly across all targets (round-robin).

With stickiness:
Each client “sticks” to one backend target — useful for preserving session data stored in memory.

🧩 2. Why Use Sticky Sessions

✅ Use Cases

  • Applications that store session state locally on EC2 (not in Redis or DynamoDB).
  • Web apps requiring user login persistence.
  • Shopping carts, dashboards, chat sessions.

❌ Avoid when

  • You want perfect load distribution (stickiness can cause uneven load).
  • The app already uses stateless design or shared storage.

⚙️ 3. Supported Load Balancers

Load Balancer Type Supports Stickiness? Cookie Name
Classic Load Balancer (CLB) ✅ AWSELB
Application Load Balancer (ALB) ✅ AWSALB or AWSALBAPP
Network Load Balancer (NLB) ✅ (via source IP) No cookie (IP-based)

🧾 4. How Sticky Sessions Work

Example:

You have ALB → 2 EC2 instances

  • Client 1 → Instance A
  • Client 2 → Instance B

When stickiness is ON:

  • Client 1’s next request always goes to Instance A.
  • Client 2’s next request always goes to Instance B.

Mechanism:

  1. The load balancer sends a cookie to the client.
  2. The client includes the cookie in each subsequent request.
  3. The load balancer uses that cookie to direct traffic to the same target.

If the cookie expires, a new instance may be chosen.

🍪 5. Cookie Types

There are two main stickiness cookie types:

Type Who Generates It Description Example Name
Application-Based Cookie The application (your backend) Custom cookie defined by your app; you control its name and duration. MYCUSTOMCOOKIEAPP
Duration-Based Cookie The Load Balancer Automatically created by ELB with a set expiry time. ALB: AWSALB / CLB: AWSELB

Application-Based Cookie (Custom)

  • Created by your app logic.

  • You can specify:

    • Cookie name
    • Duration
    • Attributes

  • Must not use reserved AWS cookie names:

  AWSALB, AWSALBAPP, AWSALBTG
Enter fullscreen mode Exit fullscreen mode

Duration-Based Cookie (Managed)

  • Created by the Load Balancer.
  • Default lifetime: 1 day, but configurable from 1 second → 7 days.
  • Cookie expires automatically; user might be rebalanced after expiration.

🔧 6. Enabling Sticky Sessions (Hands-On)

Steps:

  1. Go to EC2 → Target Groups
  2. Select your Target Group (e.g., demo-tg-alb)
  3. Click Actions → Edit Attributes
  4. Scroll to Stickiness
  5. Turn it ON
  6. Choose cookie type:
  • Load Balancer–generated (default)
  • Application–based (custom cookie)
    1. Set duration (default = 1 day)
    2. Save changes.

🔍 7. Verifying Stickiness

  1. Open your app in the browser.
  2. Enable Web Developer Tools → Network → Cookies tab.
  3. Make several requests (refresh multiple times).

✅ You’ll notice:

  • Responses always come from the same EC2 instance.

  • Under Cookies, a new cookie like AWSALB appears with:

    • Expires: tomorrow (default 1 day)
    • Path: /
    • Value: session token.

🧪 When the browser includes that cookie in the next request,
the load balancer routes back to the same backend.

⚖️ 8. Disabling Stickiness

To revert to normal round-robin load balancing:

  1. Go back to the Target Group attributes.
  2. Turn Stickiness → Off.
  3. Save changes.

Now each request can hit any backend instance again.

🧠 9. Key Points to Remember

Concept Description
Sticky Sessions Keep the same client bound to the same target
Mechanism Load balancer cookie or app cookie
Default Duration 1 day (configurable 1s–7d)
Main Benefit Maintains user session consistency
Main Risk Uneven load or instance “hot spots”
Where Configured Target group attributes (per ALB or CLB)

🧩 10. Exam & Real-World Tip

Exam trigger words:

“User sessions must remain on the same EC2 instance.”
✅ Answer: Enable sticky sessions (session affinity) on your Load Balancer.

Real-world tip:
If you need session persistence but want balanced load, use shared session storage (e.g., Redis, ElastiCache, DynamoDB) instead of stickiness.

⚖️ Cross-Zone Load Balancing in AWS

🧠 1. What Is Cross-Zone Load Balancing?

Definition:
Cross-zone load balancing ensures that each load balancer node in every Availability Zone (AZ) distributes traffic evenly across all registered targets (EC2 instances) in all AZs.

Without it, each load balancer node only sends traffic to targets within its own AZ.

📊 2. Example Scenario

💡 Setup

  • 2 Availability Zones (AZs)
  • AZ1: 2 EC2 instances
  • AZ2: 8 EC2 instances
  • Both zones are under the same load balancer.

✅ With Cross-Zone Load Balancing (Enabled)

Each load balancer node sends traffic evenly across all 10 instances, regardless of zone.

Traffic flow example:

  • Client traffic → distributed 50/50 between AZ1 and AZ2 load balancer nodes.
  • Each node → evenly distributes requests across all 10 targets.

➡️ Result:
Every EC2 instance gets 10% of total traffic, ensuring perfect balance.

❌ Without Cross-Zone Load Balancing (Disabled)

Each load balancer node only sends traffic to targets in its own AZ.

Traffic flow example:

  • Client traffic → split 50/50 between AZ1 and AZ2 load balancer nodes.
  • AZ1’s node → distributes traffic to its 2 instances only → each gets 25%.
  • AZ2’s node → distributes traffic to its 8 instances only → each gets 6.25%.

➡️ Result:
AZ1’s instances are overloaded (25% each), while AZ2’s are underutilized.

⚙️ 3. How It Works (Visual Concept) 

       ┌────────────────────┐
       │     Clients        │
       └────────┬───────────┘
                │
   ┌────────────┼───────────────┐
   │        Load Balancer        │
   │ (multiple AZ nodes inside)  │
   └────────────┬───────────────┘
                │
 ┌──────────────┼────────────────────────┐
 │              │                        │
 │         AZ1 (2 instances)         AZ2 (8 instances)
 │      (25% each w/o CZLB)       (6.25% each w/o CZLB)
 │      (10% each w/ CZLB)        (10% each w/ CZLB)
 └────────────────────────────────────────┘
Enter fullscreen mode Exit fullscreen mode

🧩 4. AWS Load Balancer Default Behaviors

Load Balancer Type Default State Can Be Changed? Inter-AZ Data Charges
Application Load Balancer (ALB) ✅ Enabled Yes (at target group level) ❌ No charge
Network Load Balancer (NLB) ❌ Disabled Yes 💲 Charged
Gateway Load Balancer (GWLB) ❌ Disabled Yes 💲 Charged
Classic Load Balancer (CLB) ❌ Disabled Yes ❌ No charge

🧱 5. Configuration Locations

Application Load Balancer (ALB):

  • Cross-Zone LB → Enabled by default.

  • Can be overridden per Target Group:

    • Inherit from ALB (default)
    • Force ON or OFF

Network Load Balancer (NLB):

  • Disabled by default.

  • To enable:

    • Go to Attributes → Edit
    • Toggle Cross-Zone Load Balancing → ON
    • Warning: “May incur inter-AZ data charges.”

Gateway Load Balancer (GWLB):

  • Same behavior as NLB (off by default, charges apply when enabled).

💰 6. Pricing Implications

LB Type Cross-Zone Behavior Inter-AZ Cost Impact
ALB Enabled by default Free (no inter-AZ charge)
NLB Optional Charged per GB of cross-AZ data
GWLB Optional Charged per GB of cross-AZ data
CLB Optional Free (legacy, no charge)

🧠 7. When to Enable / Disable

✅ Enable Cross-Zone Load Balancing when:

  • AZs have unequal numbers of EC2 instances
  • You want even traffic distribution
  • Cost impact is minimal or acceptable (ALB or low data volume)

❌ Disable Cross-Zone Load Balancing when:

  • You have balanced infrastructure across AZs
  • Want to avoid inter-AZ data charges (NLB or GWLB)
  • Need strict AZ isolation for compliance or latency

🔧 8. Hands-On Practice

  1. Go to your Load Balancer → Attributes
  2. Locate Cross-Zone Load Balancing
  3. Edit:
  • For ALB: verify it’s enabled (default)
  • For NLB/GWLB: toggle ON to enable (charges warning)
    1. At Target Group level, override ALB cross-zone setting if desired.

🧾 9. Key Takeaways

Concept Summary
Cross-Zone Load Balancing Distributes traffic across all registered targets in all AZs
Without It Each LB node only routes to its own AZ targets
ALB Enabled by default, free
NLB & GWLB Disabled by default, extra cost when enabled
Use Case Uneven AZ target distribution or global load spreading

🧠 10. Exam Tip

If a question says:

“You have more EC2 instances in one AZ than another, and want equal traffic distribution.”
✅ Answer: Enable Cross-Zone Load Balancing on your Load Balancer.

If it says:

“You need static IPs and low latency TCP traffic; avoid cross-AZ data costs.”
✅ Answer: Use NLB and disable cross-zone balancing.

🔒 SSL / TLS Certificates in AWS Load Balancing

🧠 1. What Is SSL / TLS?

Term Meaning Status
SSL (Secure Sockets Layer) Original protocol for encrypting connections Outdated
TLS (Transport Layer Security) Modern replacement for SSL 🔹 Current standard

✅ Both ensure data encryption in transit between client ↔ server.
This is called “in-flight encryption.”

🧩 2. Purpose of SSL / TLS in Load Balancing

Goal:
Encrypt all traffic between the client and the load balancer over HTTPS.

Flow Example:

Client → HTTPS (encrypted) → Load Balancer
       → HTTP (plain, private VPC) → EC2 instances
Enter fullscreen mode Exit fullscreen mode
  • The Load Balancer terminates SSL/TLS — decrypts incoming traffic before forwarding to backend targets.
  • This process is called SSL Termination.

💡 Backend communication (within the VPC) can be HTTP for performance,
or HTTPS for end-to-end encryption (optional).

🧾 3. Certificates and Management

  • SSL/TLS certificates are digital credentials that:

    • Verify the identity of your domain.
    • Enable encryption between client and server.

  • Certificates are issued by Certificate Authorities (CAs) such as:

    • Comodo, DigiCert, GlobalSign, GoDaddy, Let’s Encrypt, etc.

  • Certificates expire periodically (e.g., 1 year) and must be renewed.

🧰 4. Managing Certificates in AWS

AWS provides AWS Certificate Manager (ACM) to:

  • Automatically provision, store, and renew public certificates.
  • Let you upload your own private or third-party certificates.

💡 When configuring an HTTPS listener on a Load Balancer:

  • You must attach at least one default certificate.
  • You can attach multiple certificates (for different domains) if SNI is supported.

🔐 5. What Is SNI (Server Name Indication)?

Problem (before SNI):

  • Only one SSL certificate per IP address.
  • Hosting multiple secure websites on one server required multiple IPs.

Solution:

  • SNI (Server Name Indication) allows clients to tell the server which hostname they’re trying to reach during the SSL handshake.
  • The server (load balancer) then chooses the correct certificate dynamically.

✅ Result: One ALB/NLB can host multiple domains with different SSL certificates.

🧭 6. How SNI Works 

1️⃣ Client → “I want to connect to www.mycorp.com”
2️⃣ Load Balancer → Looks up www.mycorp.com
3️⃣ Load Balancer → Loads correct certificate
4️⃣ Connection established (TLS encrypted)
Enter fullscreen mode Exit fullscreen mode

AWS Implementation Example:

Component Description
ALB Has 2 certificates → www.mycorp.com and domain1.example.com
Client 1 Requests www.mycorp.com → ALB uses certificate for that domain
Client 2 Requests domain1.example.com → ALB uses the other certificate
Routing ALB forwards each to correct target group

🌍 7. SNI Support by AWS Load Balancers

Load Balancer Type SNI Support Certificates Supported
Classic Load Balancer (CLB) ❌ No Only one SSL certificate
Application Load Balancer (ALB) ✅ Yes Multiple certificates per listener
Network Load Balancer (NLB) ✅ Yes Multiple certificates per listener (TLS listeners only)
Gateway Load Balancer (GWLB) ❌ N/A Works at Layer 3 (no SSL support)

🧩 8. Security Policies

When you create an HTTPS listener, you can choose a Security Policy, which defines:

  • Supported TLS versions (e.g., TLS 1.2, TLS 1.3)
  • Supported cipher suites
  • Legacy client compatibility (e.g., old browsers or systems)

💡 Modern practice: Use only TLS 1.2+ for strong encryption.

💡 9. Real-World Example

Architecture: 

Client
  ↓ HTTPS (TLS)
Application Load Balancer (SSL termination)
  ↓ HTTP (private)
Target Group (EC2 instances)
Enter fullscreen mode Exit fullscreen mode

Benefits:

  • Users get the green padlock (secure site).
  • ALB handles certificate management via ACM.
  • EC2 servers can stay simple — no local certificates needed.

🧠 10. AWS Exam and Interview Tips

Scenario Correct Answer
“Multiple secure websites on one ALB.” Use SNI with multiple certificates.
“Need automatic certificate renewal.” Use AWS Certificate Manager (ACM).
“Encrypt traffic between ALB and targets too.” Use HTTPS target group or end-to-end encryption.
“Legacy system supports only old SSL versions.” Choose a custom security policy on the listener.
“One IP per SSL domain required.” (Old case) → Classic Load Balancer (no SNI).

🧾 11. Key Takeaways

Concept Summary
SSL/TLS Encrypts traffic between client and LB
SSL Termination Decryption happens at the Load Balancer
ACM AWS Certificate Manager handles certificates
SNI Enables multiple SSL certs on one LB
ALB/NLB Support SNI
CLB One certificate only
Security Policies Define TLS versions and cipher suites

🔐 Enabling SSL/TLS on Application and Network Load Balancers

🧠 1. Why Enable SSL/TLS

  • To ensure encrypted traffic between clients and your Load Balancer (HTTPS).
  • Protects data in transit and gives users a secure connection (padlock icon).
  • Implemented using port 443 (HTTPS) and valid SSL/TLS certificates from a trusted authority.

🧩 2. SSL/TLS on the Application Load Balancer (ALB)

Steps to Enable HTTPS Listener:

  1. Go to your ALB → Listeners tab
  2. Click Add listener
  3. Protocol: HTTPS Port: 443 (default)
  4. Default Action: → Forward traffic to a chosen Target Group (e.g., demo-tg-alb)

5. Configure Security Settings

(a) SSL Security Policy

  • Defines which TLS versions and cipher suites are supported.

  • Use defaults unless you need to support legacy clients.
    Example:

    • ELBSecurityPolicy-TLS13-1-2-2021-06 (modern)
    • ELBSecurityPolicy-2016-08 (legacy-compatible)

(b) SSL/TLS Certificate Source
You have three options:

Source Description
ACM (AWS Certificate Manager) Recommended. Automatically renews certificates.
IAM Can store certs but not ideal for external domains.
Import manually Paste private key, certificate body, and chain directly.

💡 Best Practice: Use ACM-issued certificates for your domain (e.g., via Route 53 DNS validation).

6. Result

Once the listener is active:

  • All client requests via https:// are encrypted.
  • ALB terminates the SSL connection at port 443.
  • Backend targets can still receive HTTP (port 80) traffic inside the VPC.

🌐 3. SSL/TLS on the Network Load Balancer (NLB)

Steps to Enable TLS Listener:

  1. Open your NLB → Listeners tab
  2. Click Add listener
  3. Protocol: TLS Port: 443 (default)
  4. Default Action: → Forward to a Target Group (e.g., demo-tg-nlb)

5. Configure Security Settings

(a) Security Policy

  • Defines TLS versions and cipher suites for negotiation.
    You can choose policies like:

    • ELBSecurityPolicy-TLS13-1-2-2021-06
    • ELBSecurityPolicy-FS-1-2-Res-2020-10

(b) Certificate Source
Same three options:

  • ACM (recommended)
  • IAM (for internal)
  • Import manually (if needed)

(c) (Optional) ALPN — Application-Layer Protocol Negotiation

  • Used for modern TLS-based protocols (e.g., HTTP/2, gRPC).
  • Usually left at default unless you specifically need it.

🧱 4. Behind the Scenes

  • ALB TLS Termination: Decrypts HTTPS → Forwards plaintext HTTP to backend.
  • NLB TLS Termination: Decrypts TCP-level encryption → Forwards decrypted traffic downstream.
  • You can also configure end-to-end encryption (client → ALB/NLB → EC2 HTTPS).

🧾 5. Key Differences Between ALB and NLB SSL Configuration

Feature ALB NLB
Layer Layer 7 (Application) Layer 4 (Network)
Listener Protocol HTTPS TLS
Certificate Management ACM / IAM / Import ACM / IAM / Import
Policy Control SSL/TLS cipher & version policies TLS cipher & version policies
Advanced Option Redirect HTTP→HTTPS ALPN (App-Layer Protocol Negotiation)

🧠 6. AWS Exam / Real-World Tips

Scenario Best Practice
“We need to secure our public web app.” Use ALB + ACM certificate on port 443.
“Our backend runs TCP-based apps (e.g., database proxy).” Use NLB with TLS listener.
“We host multiple domains.” Use SNI with multiple certs on ALB/NLB.
“We want automatic certificate renewals.” Use AWS Certificate Manager (ACM).
“Need strong encryption only.” Choose a TLS 1.2/1.3 security policy.

🧩 7. Quick Visual Summary 

          ┌─────────────────────────────┐
          │       Clients (HTTPS)       │
          └────────────┬────────────────┘
                       │ Port 443 (TLS)
             ┌─────────▼────────────┐
             │ Application LB (ALB) │  ← ACM Certificate + TLS Policy
             └─────────┬────────────┘
                       │
              ┌────────▼────────┐
              │ EC2 / ECS Tasks │  ← HTTP or HTTPS Target Group
              └─────────────────┘
Enter fullscreen mode Exit fullscreen mode 
             ┌───────────────────────┐
             │ Network LB (NLB)      │  ← TLS Listener + ACM Certificate
             └────────┬──────────────┘
                      │
               ┌──────▼──────┐
               │ TCP Targets │
               └─────────────┘
Enter fullscreen mode Exit fullscreen mode

🔁 Connection Draining / Deregistration Delay

🧠 1. Concept Overview

Goal:
To gracefully remove EC2 instances (targets) from a Load Balancer without breaking active connections.

When an instance is:

  • Manually deregistered,
  • Marked unhealthy, or
  • Being terminated / replaced (like in Auto Scaling),

…the load balancer needs to stop sending new requests to that instance
but still allow ongoing requests to finish.

This is exactly what Connection Draining (CLB) or Deregistration Delay (ALB/NLB) does.

🧩 2. Two Different Names, Same Purpose

Load Balancer Type Feature Name Default Timeout Setting Location
Classic Load Balancer (CLB) Connection Draining 300 sec (5 min) ELB attributes
Application Load Balancer (ALB) Deregistration Delay 300 sec (5 min) Target Group attributes
Network Load Balancer (NLB) Deregistration Delay 300 sec (5 min) Target Group attributes

🧾 3. How It Works

Let’s imagine 3 EC2 instances behind a Load Balancer:

       ┌──────────────────────┐
       │     Load Balancer    │
       └────────┬─────────────┘
                │
 ┌──────────────┼──────────────┐
 │     EC2-1 (draining)       │
 │     EC2-2 (active)         │
 │     EC2-3 (active)         │
 └────────────────────────────┘
Enter fullscreen mode Exit fullscreen mode

When EC2-1 enters draining state:

  1. Existing connections (in-flight requests) → continue until they finish.
  2. New connections → redirected to EC2-2 and EC2-3.
  3. After the draining timeout, any remaining open connections are terminated.

⚙️ 4. Configuration Settings

Parameter Description
Deregistration Delay Timeout Time (in seconds) that Load Balancer waits for in-flight requests to finish.
Range 1 – 3600 seconds
Default 300 seconds (5 minutes)
Disable Set to 0 seconds (no delay).

🧮 5. Example Behavior

Timeout Value Use Case Behavior
30 seconds Short API calls or static website Instance drains quickly
300 seconds (default) Balanced workload Allows moderate in-flight request completion
600–3600 seconds Long-running uploads, video streams Gives users time to finish connections

🚦 6. Real-World Example

  • Your Auto Scaling Group triggers a scale-in event.
  • One EC2 instance is selected for termination.
  • Before stopping, the Load Balancer marks it as “draining.”
  • All new traffic goes to other instances.
  • Current users finish their requests → instance shuts down gracefully.

💡 Without connection draining, active users could face:

  • Timeout errors
  • File upload failures
  • Interrupted web sessions

🧰 7. How to Configure in AWS Console

For ALB or NLB:

  1. Go to EC2 → Target Groups
  2. Select your Target Group
  3. Choose Attributes → Edit
  4. Set Deregistration delay timeout (seconds)
  5. Save changes

For Classic Load Balancer:

  1. Go to EC2 → Load Balancers
  2. Select Classic Load Balancer
  3. Go to Description → Edit Attributes
  4. Enable Connection Draining
  5. Set timeout (1–3600 sec)

💡 8. CLI Configuration Examples

ALB / NLB (Deregistration Delay): 

aws elbv2 modify-target-group-attributes \
  --target-group-arn arn:aws:elasticloadbalancing:us-east-1:123456789012:targetgroup/demo-tg/abc123 \
  --attributes Key=deregistration_delay.timeout_seconds,Value=120
Enter fullscreen mode Exit fullscreen mode

CLB (Connection Draining): 

aws elb modify-load-balancer-attributes \
  --load-balancer-name demo-clb \
  --load-balancer-attributes \
  "{\"ConnectionDraining\":{\"Enabled\":true,\"Timeout\":300}}"
Enter fullscreen mode Exit fullscreen mode

🧠 9. Exam and Interview Tips

Question Answer
What is Connection Draining used for? Allows in-flight requests to complete before instance removal.
What is it called in ALB/NLB? Deregistration Delay.
Default value? 300 seconds (5 minutes).
How to disable it? Set timeout = 0.
Where to configure in ALB? In Target Group attributes.
Scenario: “Long file uploads fail during instance termination.” Increase Deregistration Delay value.

🧾 10. Key Takeaways

Concept Description
Purpose Gracefully remove instances without breaking connections
CLB Term Connection Draining
ALB/NLB Term Deregistration Delay
Default Timeout 300 seconds
Configurable Range 1–3600 seconds
Effect Allows existing requests to complete before shutdown
Disable Set to 0 seconds

⚙️ Amazon EC2 Auto Scaling Groups (ASG)

🧠 1. What Is an Auto Scaling Group?

An Auto Scaling Group (ASG) automatically adds (scales out) or removes (scales in) EC2 instances to match the demand on your application.

✅ Main goals:

  • Maintain the right number of instances.
  • Automatically replace unhealthy instances.
  • Optimize costs by removing idle capacity.
  • Ensure high availability by balancing load across AZs.

💡 Key Idea:
ASG uses CloudWatch metrics + scaling policies to decide when to launch or terminate instances.

🧩 2. ASG Core Concepts

Term Meaning
Scale Out Add EC2 instances when load increases
Scale In Remove EC2 instances when load decreases
Minimum Size The smallest number of instances ASG maintains
Desired Capacity The number of instances ASG wants to have (starts with this)
Maximum Size The largest number of instances ASG can launch

📊 3. Example Configuration

Setting Value Meaning
Min Size 2 Always keep at least 2 instances running
Desired Capacity 4 Start with 4 instances
Max Size 7 Never exceed 7 instances

If load increases:
→ ASG automatically launches up to 7 instances (scale out).

If load decreases:
→ ASG automatically terminates instances (scale in).

🔁 4. How ASG Works with a Load Balancer

Architecture Flow: 

         ┌───────────────────────────┐
         │      Load Balancer        │
         └────────────┬──────────────┘
                      │
      ┌───────────────┼────────────────┐
      │       Auto Scaling Group       │
      └───────────────┬────────────────┘
          ┌───────────┴────────────┐
          │    EC2 Instances       │
          └────────────────────────┘
Enter fullscreen mode Exit fullscreen mode

Behavior:

  • New EC2 instances automatically register with the Load Balancer.
  • ELB performs health checks and sends the status to the ASG.
  • If an instance is unhealthy, ASG terminates and replaces it.
  • Traffic is evenly load balanced across all healthy instances.

💡 Works perfectly with:

  • Application Load Balancer (ALB)
  • Network Load Balancer (NLB)

🧱 5. Launch Template (or Launch Configuration)

ASG uses a Launch Template that defines how new EC2 instances are created.

Launch Template includes:

  • Amazon Machine Image (AMI)
  • Instance Type (e.g., t3.micro)
  • User Data (startup script)
  • EBS volumes
  • Security Groups
  • Key Pair (SSH access)
  • IAM Role for EC2
  • Network Subnets / VPC
  • Optional Load Balancer attachment

💡 Launch Configurations are older — use Launch Templates for all new setups.

📈 6. Integration with CloudWatch

ASGs rely on CloudWatch Alarms to scale dynamically.

Example Flow:

CloudWatch Metric → Alarm Triggered → ASG Scaling Policy → Adjust Instance Count
Enter fullscreen mode Exit fullscreen mode

Common metrics:

  • Average CPU Utilization
  • Network In/Out
  • Request Count per Target
  • Custom Application Metrics

Scale-Out Policy Example:

  • If CPU > 70% for 5 minutes → Add 2 instances

Scale-In Policy Example:

  • If CPU < 20% for 10 minutes → Remove 1 instance

🧮 7. Example Scenario

Time Average CPU ASG Action Instance Count
12:00 PM 20% Scale In 2
1:00 PM 75% Scale Out 4
2:00 PM 90% Scale Out 6
4:00 PM 15% Scale In 3

ASG keeps scaling dynamically to maintain optimal performance and cost.

💰 8. Cost Model

Auto Scaling Groups are free —
you only pay for:

  • EC2 instances it launches
  • EBS volumes
  • Load balancer usage

🧠 9. Exam & Real-World Tips

Concept What to Remember
Purpose of ASG Automatically scale EC2 instances based on demand
Scale Out Add instances
Scale In Remove instances
Min/Max/Desired Controls group size
Launch Template Defines instance configuration
CloudWatch Triggers scaling policies
Health Checks Replace unhealthy instances automatically
Cost ASG is free — pay for underlying EC2 only

🧩 10. Quick Recap Diagram 

┌────────────────────────────┐
│      CloudWatch Alarm      │
└──────────────┬─────────────┘
               │
               ▼
┌────────────────────────────┐
│  Auto Scaling Group (ASG)  │
│   ├─ Min: 2                │
│   ├─ Desired: 4            │
│   └─ Max: 7                │
└──────────────┬─────────────┘
               │
               ▼
       ┌─────────────┐
       │ EC2 Fleet   │  ← Instances automatically added/removed
       └─────────────┘
               │
               ▼
     ┌─────────────────┐
     │ Load Balancer   │  ← Routes traffic to healthy EC2s
     └─────────────────┘
Enter fullscreen mode Exit fullscreen mode

⚙️ Auto Scaling Group (ASG) – Hands-On Practice

🧩 1. Pre-Setup: Clean Environment

  • Go to EC2 → Instances
  • Terminate all existing EC2 instances → You should have 0 running before starting.

🏗️ 2. Step 1 – Create an Auto Scaling Group

Go to:

EC2 → Auto Scaling Groups → Create Auto Scaling Group
Enter fullscreen mode Exit fullscreen mode

Name: Demo-ASG

Since an ASG needs instructions on how to launch instances,
we must first create a Launch Template.

🧰 3. Step 2 – Create a Launch Template

  1. Name: my-demo-template
  2. Description: template for ASG practice
  3. AMI (Amazon Machine Image):
  • Choose Amazon Linux 2 (x86, Free Tier Eligible)
    1. Instance Type: t2.micro
    2. Key Pair: EC2-tutorial (or your existing key)
    3. Security Group: launch-wizard-1 (HTTP + SSH allowed)
    4. Storage: Default 8 GB gp2 volume
    5. User Data: Paste the web-server startup script you’ve been using: 
   #!/bin/bash
   yum update -y
   yum install -y httpd
   systemctl start httpd
   systemctl enable httpd
   echo "<h1>Hello world from $(hostname -f)</h1>" > /var/www/html/index.html
Enter fullscreen mode Exit fullscreen mode
  1. Create Template → Done

✅ You’ve defined how EC2 instances will be launched by the ASG.

🌐 4. Step 3 – Configure the Auto Scaling Group

  1. Select Launch Template: my-demo-template
  2. VPC: Choose your default or custom VPC
  3. Availability Zones: Select at least two or three subnets for multi-AZ distribution.
  4. Load Balancer Integration:
  • Choose Attach to an existing target group

  • Pick your ALB target group (e.g., demo-tg-alb)

    1. Health Checks:

  • Enable both EC2 and ELB health checks
    → ASG will replace any failed instance.

    1. Capacity Settings: 
   Min capacity: 1
   Desired capacity: 1
   Max capacity: 1
Enter fullscreen mode Exit fullscreen mode
  1. Scaling Policies: Skip for now (manual scaling only)
  2. Notifications / Tags: Optional
  3. Create Auto Scaling Group

🚀 5. Step 4 – Observe Instance Launch

Go to:

Auto Scaling Group → Demo-ASG → Activity History
Enter fullscreen mode Exit fullscreen mode
  • You’ll see a new activity: “Launching a new EC2 instance”
  • ASG automatically provisions the instance defined by the Launch Template.

Then check under:

EC2 → Instances
Enter fullscreen mode Exit fullscreen mode

✅ One instance is running.

⚙️ 6. Step 5 – Load Balancer & Health Check Integration

  • Go to: 
  EC2 → Target Groups → demo-tg-alb → Targets
Enter fullscreen mode Exit fullscreen mode
  • You’ll see your new instance in initializing state.
  • After the User Data completes and health checks pass → Healthy.

💡 The instance is now automatically registered to your Application Load Balancer (ALB).

🌍 7. Step 6 – Test the Web Application

  • Copy the ALB DNS name and open it in a browser.
  • You should see: 
  Hello world from ip-xx-xx-xx-xx.ec2.internal
Enter fullscreen mode Exit fullscreen mode

✅ Your instance created by the ASG is fully serving traffic through the ALB.

📈 8. Step 7 – Manual Scaling (Scale Out)

To test scaling out manually:

  1. Go to: 
   Auto Scaling Group → Demo-ASG → Edit
Enter fullscreen mode Exit fullscreen mode
  1. Change: 
   Desired capacity: 2
   Max capacity: 2
Enter fullscreen mode Exit fullscreen mode
  1. Save changes.

⏳ ASG detects that current count = 1, desired = 2 → launches a new instance.

  • Watch Activity History:

    • “Launching a new EC2 instance”

  • Check Target Group → Targets:

    Both instances become Healthy after bootstrapping.

✅ ALB now balances traffic between two EC2 instances.

You can verify by refreshing the ALB DNS —
you’ll see different hostnames/IPs alternating.

📉 9. Step 8 – Manual Scaling (Scale In)

Now test scaling in:

  1. Edit ASG again: 
   Desired capacity: 1
Enter fullscreen mode Exit fullscreen mode
  1. Save changes.

ASG will:

  • Pick one instance to terminate
  • Deregister it from the Target Group
  • Retain one healthy EC2 instance

Check Activity History:

Terminating EC2 instance to match desired capacity
Enter fullscreen mode Exit fullscreen mode

✅ Traffic now routes only to the remaining instance.

🧠 10. Key Takeaways

Concept Description
Launch Template Defines how ASG creates EC2 instances
Desired Capacity Number of instances to maintain
Scaling Out Add instances (handle more load)
Scaling In Remove instances (reduce cost)
Load Balancer Integration ALB automatically distributes traffic & checks health
Health Check ASG replaces failed instances automatically
Cost ASG is free; pay for EC2 + Load Balancer only

🧾 11. Quick Visual Summary 

┌───────────────────────────────┐
│      Application Load Balancer│
└──────────────┬────────────────┘
               │
   ┌───────────▼───────────┐
   │  Auto Scaling Group   │
   │  ├ Min = 1            │
   │  ├ Desired = 2         │
   │  └ Max = 2            │
   └───────────┬───────────┘
               │
   ┌───────────▼───────────┐
   │   EC2 Instances       │
   │   (Web servers w/ UserData) │
   └───────────────────────┘
Enter fullscreen mode Exit fullscreen mode

⚙️ Auto Scaling Group (ASG) – Scaling Policies Explained

🧠 1. Purpose of Scaling Policies

Scaling policies tell the Auto Scaling Group when and how to adjust the number of EC2 instances based on metrics, schedules, or predictions.

Goal:
Keep your application stable, cost-efficient, and responsive to changes in load.

🚀 2. Types of Scaling Policies

A. Dynamic Scaling

Automatically adjusts capacity in response to changing demand.

1️⃣ Target Tracking Scaling (most common & simplest)

  • You define a metric and a target value.
  • ASG automatically scales in/out to maintain that target.

Example:
Target metric = Average CPU utilization
Target value = 40%

→ ASG keeps average CPU ≈ 40% by adding/removing instances as needed.

Analogy: Like a thermostat — you set a temperature (target metric), and it adjusts automatically.

Typical metrics:

  • AWS/EC2:CPUUtilization
  • AWS/ApplicationELB:RequestCountPerTarget
  • AWS/EC2:NetworkIn / NetworkOut
  • Custom CloudWatch metrics

2️⃣ Simple / Step Scaling

  • You create CloudWatch alarms that trigger scaling actions.
  • When an alarm threshold is reached, ASG adds or removes instances.

Example:

If average CPU > 70% → Add 2 instances
If average CPU < 30% → Remove 1 instance
Enter fullscreen mode Exit fullscreen mode

Step Scaling enhances this by defining multiple thresholds:

Condition Action
CPU ≥ 70% Add 1 instance
CPU ≥ 85% Add 2 instances
CPU ≤ 20% Remove 1 instance

💡 Useful for gradual adjustments instead of big jumps.

B. Scheduled Scaling

Used when you know in advance that load will increase or decrease.

Example:

  • Every Friday at 5 PM → increase min capacity to 10
  • Every Sunday at 2 AM → decrease min capacity to 2

🕓 Useful for predictable workloads, like office hours or batch jobs.

C. Predictive Scaling

Uses machine learning to analyze historical trends and automatically forecast demand.

  • AWS forecasts future traffic based on patterns.
  • It pre-provisions instances ahead of time to prevent performance drops.

💡 Ideal for cyclical traffic patterns, such as:

  • E-commerce peaks (weekends, holidays)
  • Daily business workloads (9 AM – 5 PM)

📈 3. Common Scaling Metrics

Metric Description When to Use
CPUUtilization Average CPU load across ASG For compute-intensive apps
RequestCountPerTarget Requests handled per target (via ALB) For web apps / APIs
NetworkIn / NetworkOut Data transferred (in bytes) For upload/download-heavy apps
Custom Metric Any app-specific metric pushed to CloudWatch For special workloads (e.g., queue depth, active users)

🧮 4. Cooldown Period (Stabilization Delay)

After a scaling activity (adding/removing instances), the ASG enters a cooldown period — a “quiet time” before the next scaling action.

Setting Description
Default Cooldown 300 seconds (5 minutes)
Purpose Prevents rapid up/down scaling before metrics stabilize
Disable / Adjust You can customize or shorten it if using pre-baked AMIs

🧠 How it works:

  • If a scaling action occurs during cooldown → ignored.
  • If no cooldown in progress → action executed.

Best practice:
Use ready-to-use AMIs (with pre-installed dependencies)
→ new instances become healthy faster → shorter cooldowns → quicker scaling response.

🧩 5. Example Scenario

Let’s say your web app has 3 instances and CPU jumps to 80%.

Metric Rule Action Result
CPU > 70% Add 2 instances Scale Out 5 total
CPU < 30% Remove 1 instance Scale In 4 total
CPU stabilizes at 40% No change Stable 4 total

🔍 6. Recommended Setup Flow

  1. Enable Detailed Monitoring (1-minute metric intervals)
  2. Define metric and target value
  3. Configure Target Tracking Policy
  4. Adjust Cooldown Period
  5. Test with manual scaling first
  6. Gradually enable predictive/scheduled scaling

💡 7. Exam & Interview Tips

Question Answer
What’s the easiest scaling policy to use? Target Tracking Scaling
Which scaling type uses CloudWatch alarms? Simple / Step Scaling
Which one is based on future load forecasts? Predictive Scaling
Default cooldown period? 300 seconds
How to reduce cooldown? Use pre-baked AMIs & shorter warm-up time
What’s the most common scaling metric? Average CPUUtilization
How to prepare for predictable spikes? Scheduled Scaling

🧾 8. Visual Summary 

 ┌──────────────────────────────────────┐
 │           CloudWatch Metrics         │
 │  (CPU, Requests, Network, Custom)    │
 └──────────────────┬───────────────────┘
                    │
                    ▼
      ┌────────────────────────────┐
      │ Auto Scaling Group (ASG)   │
      │ ├ Target Tracking Policy   │
      │ ├ Step/Simple Policy       │
      │ ├ Scheduled Policy         │
      │ └ Predictive Policy        │
      └────────────┬───────────────┘
                   │
                   ▼
       ┌─────────────────────────┐
       │ EC2 Instances Adjusted  │
       └─────────────────────────┘
Enter fullscreen mode Exit fullscreen mode

⚙️ Auto Scaling Group (ASG) – Automatic Scaling Hands-On

🧠 Goal

Set up automatic scaling so your ASG can add or remove EC2 instances automatically based on CPU utilization (or any other metric).

We’ll practice:

  • Scheduled scaling
  • Predictive scaling
  • Dynamic scaling (with Target Tracking)

🗓️ 1. Scheduled Scaling

Schedule capacity changes at specific times.

Example use cases:

  • You expect heavy traffic during weekends or marketing events
  • You want to scale down during off-hours to save cost

Steps:

  1. Go to your ASG → Automatic Scaling → Scheduled actions → Create
  2. Define:
  • Start time / end time

  • Desired, Min, or Max capacity

    1. Example:

  • At Saturday 8 AM, increase min capacity to 10

  • At Sunday 11 PM, decrease min capacity to 2

✅ Used when load is predictable.

🧮 2. Predictive Scaling

Uses machine learning to automatically forecast load and scale ahead of time.

How it works:

  • AWS analyzes your app’s historical metrics (like CPU or request counts)
  • Builds a forecast for upcoming load
  • Schedules scaling before the spike

Setup:

  • Choose metric (CPU, ALB RequestCountPerTarget, etc.)
  • Define a target value (e.g., 50% CPU)
  • AWS automatically handles the rest

💡 Works best after at least 24 hours of steady metrics — typically used for apps with repeating patterns (e.g., business hours, daily peaks).

⚡ 3. Dynamic Scaling (Hands-On)

Automatically adjusts instance count based on real-time metrics.

There are three types:

Type Description Example
Simple Scaling Uses one CloudWatch alarm “If CPU > 70%, add 1 instance”
Step Scaling Uses multiple alarm thresholds “If CPU > 85%, add 2; if > 70%, add 1”
Target Tracking (Recommended) Automatically maintains a target metric “Keep average CPU = 40%”

🧰 Step-by-Step: Target Tracking Policy

  1. Go to your ASG → Automatic Scaling → Add Policy
  2. Choose Dynamic Scaling Policy
  3. Select Target Tracking Scaling Policy
  4. Name it CPU-TargetTracking
  5. Metric: Average CPU Utilization
  6. Target value: 40%
  7. Leave cooldown at default (300s)
  8. Save

✅ This automatically creates two CloudWatch alarms:

  • AlarmHigh → Scale Out (when CPU > 40%)
  • AlarmLow → Scale In (when CPU < 28%)

🔬 4. Stress Test to Trigger Scaling

Now we’ll simulate load to make scaling happen.

Step 1 – Connect to EC2 Instance

Use EC2 Instance Connect or your SSH key:

ssh -i mykey.pem ec2-user@<public-ip>
Enter fullscreen mode Exit fullscreen mode

Step 2 – Install stress tool 

sudo yum install -y stress
Enter fullscreen mode Exit fullscreen mode

Step 3 – Generate CPU load 

stress -c 4
Enter fullscreen mode Exit fullscreen mode

→ Forces CPU utilization close to 100%.

Step 4 – Observe scaling

Go to:

EC2 → Auto Scaling Groups → Demo-ASG → Activity history
Enter fullscreen mode Exit fullscreen mode

After a few minutes you’ll see:

Launching a new EC2 instance due to target tracking policy
Enter fullscreen mode Exit fullscreen mode

Check under:

Instance management → Instances
Enter fullscreen mode Exit fullscreen mode

✅ New EC2 instances appear (scale-out event)

📊 5. CloudWatch Alarms in Action

Go to:

CloudWatch → Alarms
Enter fullscreen mode Exit fullscreen mode

You’ll see:

  • AlarmHigh → State: In ALARM (CPU > 40%)
  • AlarmLow → State: OK or INSUFFICIENT_DATA

Details:

  • AlarmHigh: triggers scale-out after CPU > 40% for 3 data points (3 min)
  • AlarmLow: triggers scale-in after CPU < 28% for 15 data points (15 min)

🔁 6. Observe Scale-In

Once you stop stressing the CPU:

  • Press Ctrl + C to stop stress, or reboot instance: 
  sudo reboot
Enter fullscreen mode Exit fullscreen mode
  • CPU utilization drops → alarmLow activates
  • ASG automatically terminates extra instances

Check Activity History:

Terminating EC2 instance to match desired capacity
Enter fullscreen mode Exit fullscreen mode

✅ ASG gradually scales back down to the minimum instance count.

📈 7. Verification

Step Action Expected Result
Stress CPU Run stress -c 4 CPU → 100%, triggers scale-out
Stop stress Ctrl+C / reboot CPU → drops, triggers scale-in
Check CloudWatch Two alarms visible AlarmHigh and AlarmLow
Check ASG activity History updated Launch/Terminate events visible

⚙️ 8. Cleanup

When finished:

  1. Stop stress test (if still running)
  2. In ASG → Automatic Scaling, delete your target tracking policy
  3. Optionally, delete the ASG and launch template to stop all EC2 billing

🧠 9. Key Takeaways

Concept Summary
Target Tracking Simplest, recommended scaling method
Automatic Scaling Adds/removes instances based on metrics
Scaling Alarms Created automatically by target tracking
Predictive Scaling Uses ML to forecast and pre-scale
Scheduled Scaling Manual scheduling of capacity changes
Cooldown Period Default 300s; prevents rapid oscillation
Stress Tool Useful for testing scaling logic
Detailed Monitoring Provides 1-minute metric granularity

🧾 10. Visual Summary 

CloudWatch Metric (CPU) ───► AlarmHigh / AlarmLow
        │
        ▼
  ┌─────────────────────────────┐
  │ Auto Scaling Group (Demo-ASG) │
  │  • Min: 1                    │
  │  • Max: 3                    │
  │  • Target: 40% CPU           │
  └───────────┬─────────────────┘
              │
   ┌──────────▼─────────┐
   │ EC2 Instances Auto │
   │ Added / Removed    │
   └────────────────────┘
Enter fullscreen mode Exit fullscreen mode

📊 Lecture: Visualizing ASG Metrics in CloudWatch Dashboard

🎯 Goal

By the end of this lab, students will:

  • Create a CloudWatch Dashboard
  • Add ASG metrics (CPU, instance count, network)
  • Observe scale in/out events in real time
  • Understand how to interpret these graphs for troubleshooting and tuning scaling policies

🧩 1. What Is a CloudWatch Dashboard?

A CloudWatch Dashboard is a customizable view that displays your AWS metrics in one screen.

You can:

  • Monitor ASG, EC2, ELB, and CloudWatch Alarms
  • View CPU spikes, instance count changes, and traffic load
  • Compare multiple metrics side by side

✅ This is essential for real-world DevOps monitoring and alerting setups.

⚙️ 2. Step-by-Step: Create the Dashboard

  1. Go to CloudWatch Console 
   AWS Console → CloudWatch → Dashboards
Enter fullscreen mode Exit fullscreen mode
  1. Click Create Dashboard
  2. Name it: ASG-Monitoring-Dashboard
  3. Choose Line widget (for graphs)
  4. Click Configure

📈 3. Add Key Metrics to the Dashboard

Let’s add 4 key widgets that every DevOps engineer should monitor for ASGs.

🧠 Widget 1: Average CPU Utilization

  • Namespace: AWS/EC2
  • Metric: CPUUtilization
  • Statistic: Average
  • Period: 1 minute (requires Detailed Monitoring)
  • Filter: Select your Auto Scaling Group instances
  • Label: “Average CPU Utilization”

✅ This shows how busy your EC2 instances are.

🧩 Widget 2: ASG Group Desired, InService, and Pending Instances

  • Namespace: AWS/AutoScaling

  • Metrics:

    • GroupDesiredCapacity
    • GroupInServiceInstances
    • GroupPendingInstances

  • Dimensions: Choose your ASG (Demo-ASG)

  • Statistic: Average

  • Label: “ASG Capacity Overview”

✅ Shows when scaling events occur (launch/terminate).

🌐 Widget 3: NetworkIn / NetworkOut

  • Namespace: AWS/EC2

  • Metrics:

    • NetworkIn
    • NetworkOut

  • Statistic: Sum

  • Period: 1 minute

  • Label: “Network Traffic (bytes)”

✅ Useful for data-heavy applications or detecting high traffic peaks.

🧱 Widget 4: Application Load Balancer Request Count

If your ASG is behind an ALB:

  • Namespace: AWS/ApplicationELB
  • Metric: RequestCountPerTarget
  • Statistic: Sum
  • Dimension: Choose your Target Group (e.g., demo-tg-alb)
  • Label: “Requests Per Target”

✅ Helps correlate incoming traffic with scaling actions.

💡 Optional Widget: CloudWatch Alarms State

  • Widget Type: “Alarm Status”

  • Select alarms created by your Target Tracking Policy:

    • AlarmHigh
    • AlarmLow

  • Label: “Scaling Alarms”

✅ Visually shows which alarm triggered (red = active).

🎨 4. Arrange and Save Dashboard

  • Drag widgets to organize logically: 
  ┌──────────────────────────────────────────────┐
  │ Average CPU Utilization (%)                  │
  ├──────────────────────────────────────────────┤
  │ ASG Capacity Overview (Desired vs InService) │
  ├──────────────────────────────────────────────┤
  │ Network Traffic (In/Out Bytes)               │
  ├──────────────────────────────────────────────┤
  │ Requests Per Target (ALB)                    │
  └──────────────────────────────────────────────┘
Enter fullscreen mode Exit fullscreen mode
  • Click Save Dashboard

🔍 5. Test in Real Time

To see the dashboard in action:

  1. Run the stress test again on one instance: 
   stress -c 4
Enter fullscreen mode Exit fullscreen mode
  1. Wait 2–3 minutes.
  2. Watch:
  • CPUUtilization spike upward
  • AlarmHigh trigger

  • ASG Desired / InService Instances increase

    1. Stop stress test with Ctrl + C
    2. Observe:

  • CPUUtilization drops

  • AlarmLow triggers

  • Instances terminate (scale-in)

✅ You’ll see the entire auto-scaling process visually.

🧮 6. Understanding the Dashboard

Metric What It Shows What To Look For
CPU Utilization Resource stress level Sudden spikes → scaling triggers
Desired / InService Instance count Confirms scaling in/out actions
NetworkIn/Out Load intensity High traffic → possible cause of scale-out
Requests per Target User load via ALB Useful for performance tuning
Alarm Status Which threshold triggered Red (scale out), Green (scale in)

🧠 7. Best Practices

Area Recommendation
Granularity Enable 1-minute detailed monitoring on ASG
Metrics Track both ASG-level and EC2-level metrics
Naming Use consistent names (e.g., demo-asg, demo-tg-alb)
Alarms Always have one for scale-out and one for scale-in
Retention Keep dashboard open during load testing
Automation Optionally export to JSON for IaC (Terraform or CloudFormation)

🧾 8. Example Dashboard JSON (optional)

If you want to create via CLI or CloudFormation:

{
  "widgets": [
    {
      "type": "metric",
      "x": 0,
      "y": 0,
      "width": 24,
      "height": 6,
      "properties": {
        "metrics": [["AWS/EC2", "CPUUtilization", "AutoScalingGroupName", "Demo-ASG"]],
        "title": "Average CPU Utilization",
        "period": 60,
        "stat": "Average"
      }
    },
    {
      "type": "metric",
      "x": 0,
      "y": 6,
      "width": 24,
      "height": 6,
      "properties": {
        "metrics": [
          ["AWS/AutoScaling", "GroupDesiredCapacity", "AutoScalingGroupName", "Demo-ASG"],
          ["AWS/AutoScaling", "GroupInServiceInstances", "AutoScalingGroupName", "Demo-ASG"]
        ],
        "title": "ASG Capacity Overview"
      }
    }
  ]
}
Enter fullscreen mode Exit fullscreen mode

🎓 9. Key Takeaways

Concept Summary
CloudWatch Dashboard Custom visualization of ASG performance
Dynamic scaling visibility Instantly shows scale events & CPU trends
Metrics correlation Links alarms, traffic, and scaling patterns
Practical DevOps skill Used in every production-grade AWS setup
Next Step Integrate with SNS or Slack alerts for scaling notifications

Top comments (1)

Collapse
 
juzonb1r profile image
JIM •

Wow