Edge Delivery: CDN Edge Locations, Regions, and Caching

This chapter explores the physical and architectural reality of modern cloud data centers, global network routing, and edge delivery mechanisms. Through practical hands-on demonstrations using AWS and Linux terminal utilities, we examine how physical infrastructure, subsea fiber optic cables, key-based security, and diagnostic tools like traceroute interact to deliver data across the globe.

1. The Physical Reality of Data Centers & Global Infrastructure

When we access web applications or stream media, the data originates from physical computers housed inside industrial facilities known as Data Centers. Data centers represent the physical backbone—or the "heart"—of the internet.

Extreme Cooling & Infrastructure Engineering

Servers operating 24/7 produce immense heat. Traditional ground-level data centers dedicate vast amounts of space and energy exclusively to industrial air conditioning (AC) and cooling infrastructure to prevent thermal throttling and hardware failure.

Real-Life Analogy: Underwater Data Centers

To solve the massive energy and cooling challenges of traditional facilities, tech giants explored radical environments:

• Microsoft Project Natick (2018): Microsoft deployed a fully functional, sealed data center off the coast of Scotland, submerged deep underwater.
• Why Submerge Data Centers?: The surrounding ocean water acts as a natural, continuous heat sink, drastically reducing cooling costs and operational energy consumption.

2. How Data Travels: Subsea Optics to Local Wi-Fi

Every bit of information requested by an end-user must physically travel from a remote data center to the client's screen.

The Physical Connectivity Highway

Data does not simply float through the air; it relies on an intricate network of undersea optical fiber cables spanning thousands of miles across ocean floors.

Remote Data Center (e.g., Oregon, US)
   |
   | Transits through high-speed internal network
   v
Undersea Optical Fiber Cables (Trans-Oceanic Highway)
   |
   | Reaches regional landing station
   v
Tier-1/Tier-2 Internet Backbone Providers
   |
   | Routes through regional infrastructure
   v
Local Internet Service Provider (ISP)
   |
   | Transmits via home router / Wi-Fi
   v
Client Device (Laptop / Mobile Browser)

Note: The fees charged by cloud providers and ISPs cover this massive chain of physical infrastructure—from subsea cable maintenance and regional transit to electricity and cooling.

3. Cloud Provisioning: Renting Regional Infrastructure

Instead of building physical facilities, developers rent compute resources on-demand from public cloud providers like Amazon Web Services (AWS).

Regional Isolation and Cost Models

When launching a cloud server (such as an AWS EC2 instance), you select a target geographical region (e.g., us-west-2 in Oregon, USA).

• On-Demand Utility Model: Cloud providers charge per-second or per-minute micro-rates for server utilization.
• Cost Drivers: These costs encompass compute allocation, hardware wear, active cooling, physical security, and high-speed global transit bandwidth.

4. Hands-on Workflow: Launching a Remote Cloud Instance

To examine global routing firsthand, we provision a lightweight cloud virtual server in a remote region.

AWS EC2 Instance Launch Specification

1. Name: Networking-Practice-Server
2. OS Image (AMI): Ubuntu Server (LTS)
3. Instance Type: t3.micro (Free Tier eligible)
4. Key Pair: networking-practice-key.pem
5. Network Settings: Enable SSH traffic (Port 22) from custom/anywhere IP ranges
6. Storage Allocation: Standard root volume

5. Remote Access via SSH & Cryptographic Key Management

To establish a secure terminal session with a freshly provisioned cloud instance across the ocean, we use SSH (Secure Shell).

Secure File Permission Assignment

Open SSH clients enforce strict security checks. If a private key file (.pem) has overly permissive file rights (allowing other local users to read or modify it), the SSH client will reject the connection attempt.

# Navigate to the directory containing the downloaded key pair
cd ~/Downloads

# Restrict permissions: Read-only for the file owner (User: Read, Group: None, Others: None)
chmod 400 networking-practice-key.pem

Security Mandate: chmod 400 sets key permissions to read-only for the owner (r--------), satisfying SSH security constraints.

Initiating the SSH Connection

# Connect to the remote instance using identity file (-i)
ssh -i "networking-practice-key.pem" ubuntu@<PUBLIC_IP_OR_DNS>

When prompted:

The authenticity of host 'ec2-xx-xx-xx-xx.us-west-2.compute.amazonaws.com' can't be established.
Are you sure you want to continue connecting (yes/no/[fingerprint])? yes

Upon confirmation, your local shell context transitions to the remote server shell running physically in Oregon.

6. Linux Package Management: update vs. upgrade

When setting up network diagnostic utilities on a fresh Linux server, software packages must be managed using the apt (Advanced Package Tool) package manager.

Key Distinction: Index Updating vs. System Upgrading

# 1. Update package index (Fetches latest package list metadata)
sudo apt update

# 2. Install specific target package
sudo apt install traceroute

Why apt update was chosen over apt upgrade

• sudo apt update: Downloads the latest package lists from repository indexes. It does not install or modify existing software binaries. Execution duration: ~3–5 seconds.
• sudo apt upgrade: Downloads and upgrades all installed system software packages to their latest versions. Execution duration: 10+ minutes.

Efficiency Principle: When you only need to install a single diagnostic utility like traceroute, run sudo apt update followed immediately by sudo apt install <package>. Avoid running sudo apt upgrade during active diagnostic tasks to eliminate unnecessary installation overhead.

7. Tracing Network Hops with traceroute

Once logged into the remote server, we trace the route network packets take to reach a target domain.

Understanding Network Packet Pathways

Data does not travel instantaneously in a single leap. Instead, packets hop through multiple intermediate routers (gateways) before reaching the destination host.

# Execute traceroute to track packet hops to a target server
traceroute trainwithshubham.com

Conceptual Output Analysis

traceroute to trainwithshubham.com (104.21.78.12), 30 hops max, 60 byte packets
 1  ec2-router-gateway.us-west-2.internal (192.168.1.1)  0.412 ms  0.380 ms
 2  aws-backbone-border.amazon.com (54.240.225.12)       1.120 ms  1.050 ms
 3  transit-provider-node.net (203.0.113.45)            12.340 ms 12.110 ms
 ...
12  104.21.78.12 (trainwithshubham.com)                  45.610 ms 45.480 ms

Packet Routing Execution Model

Client Machine / EC2 Instance
   |
   | Hop 1: Internal VPC Gateway Router
   v
Regional Cloud Edge Router
   |
   | Hop 2–5: Autonomous System (AS) Backbone Transit
   v
Tier-1 Internet Exchange / Transit Providers
   |
   | Hop 6–11: CDN / Edge Proxy Layer (Cloudflare/AWS CloudFront)
   v
Hop 12: Target Origin Web Server

Each row returned by traceroute represents an intermediate router (hop), along with the round-trip time (RTT) recorded across three separate test packets.

Summary and Key Takeaways

• Data Centers are the Heart of the Internet: Every cloud service runs on physical servers housed inside specialized facilities requiring massive cooling infrastructure.
• Subsea Global Highways: Intercontinental internet traffic travels via physical undersea optical fiber cables across ocean beds, not through pure wireless satellite transmission.
• Radical Infrastructure Design: Microsoft Project Natick demonstrated that placing sealed data centers underwater utilizes ocean temperatures for efficient heat dissipation.
• Regional Cloud Allocation: Renting cloud servers (e.g., AWS EC2 in Oregon) incurs usage costs driven by compute, electricity, global network transit, and cooling requirements.
• Key Permissions (chmod 400): SSH mandates strict file permissions (400) on private key files to ensure non-authorized local system users cannot access sensitive keys.
• Package Management Efficiency: Use sudo apt update to quickly index repository lists without incurring the heavy time penalty of running sudo apt upgrade.
• Network Inspection via traceroute: The traceroute utility maps the hop-by-hop path taken by IP packets from source to destination across global network boundaries.