Network Performance: Bandwidth, Latency, and Packet Loss

The global internet is not an abstract cloud; it is a vast, physical mesh of subsea optical cables, landing stations, Wide Area Networks (WANs), Metropolitan Area Networks (MANs), Internet Exchange Points (IXPs), and Internet Service Providers (ISPs). Understanding this physical supply chain of data is critical for DevOps and system design engineers who must diagnose latencies, design redundant architectures, and optimize application throughput.

1. The Global Physical Backbone: Subsea Optical Cables & Landing Stations

At its base, global connectivity relies on physical fiber-optic cables buried deep beneath the ocean floor. These submarine cables span thousands of kilometers across ocean beds, linking continents and countries together.

Subsea Cable System (Undersea Ocean Bed)
   |
   | Transmits optical signals across continents
   v
Coastal Landing Stations (e.g., Mumbai, Kochi, Trivandrum, Tuticorin)
   |
   | Converts optical signals for terrestrial transport
   v
National Fiber Network

Key Infrastructure Concepts

• Submarine Cable Systems: Cables such as Falcon, Raman, SMW (SEA-ME-WE), and EIG (Europe India Gateway) transport data across global waters using high-density light signals.
• Coastal Landing Stations: Specific geographical entry points where undersea cables make landfall. In India, key landing hubs include coastal cities like Mumbai, Kochi, Trivandrum, Tuticorin, and Digha.
• High Operational Expenditure (OpEx & CapEx): Laying and maintaining undersea cables requires immense capital. Specialized repair ships, continuous electrical power feeding, deep-sea cooling consideration, and mechanical shielding against anchor cuts run into millions of dollars in continuous maintenance.

2. Wide Area Networks (WAN) & Infrastructure Ownership

Once undersea cables reach coastal landing stations, data must travel across national territories. This long-distance terrestrial infrastructure is known as a Wide Area Network (WAN).

Landing Station (Coastal Entry)
   |
   | Feeds data into long-distance fiber channels
   v
National WAN Backbone (Owned by Telecom Conglomerates)
   |
   | Routes data across inland regions and major cities
   v
Metropolitan Area Network (MAN)

Because laying thousands of kilometers of terrestrial dark fiber requires massive capital investment, regulatory licenses, power grid access, and physical maintenance teams, national WAN backbones are typically owned and operated by major infrastructure and telecommunication conglomerates (e.g., Reliance Jio, Bharti Airtel, Tata Communications, Adani Group).

Note: The entity that owns the physical long-haul fiber controls the routing pathways through which national data flows. Small internet providers rarely own long-distance subsea or national fiber; instead, they lease capacity from these WAN infrastructure owners.

3. Metropolitan Area Networks (MAN) & Urban Traffic Routing

When long-haul data on a national WAN reaches a target urban center (e.g., Mumbai, Pune, Bengaluru), it transitions into a Metropolitan Area Network (MAN).

• Definition: A Metropolitan Area Network spans a city or town, acting as the intermediate distribution layer between national long-distance backbones (WAN) and local neighborhoods (LAN).
• Role in System Performance: The MAN handles intra-city routing and connects high-capacity enterprise networks, localized data centers, and consumer distribution hubs.

4. Internet Exchange Points (IXP) vs. Internet Service Providers (ISP)

Within a Metropolitan Area Network (MAN), two critical architectural components facilitate data routing: Internet Exchange Points (IXPs) and Internet Service Providers (ISPs).

Key Differences & Roles

Real-World Analogy

Think of an IXP as a central wholesale mail-sorting facility where different international cargo trains deposit mail. An ISP is the local neighborhood mail delivery driver who picks up mail from that central sorting facility and drives it directly to your doorstep.

5. The Financial & Operational Flow of Internet Access

Internet connectivity relies on a cascading economic supply chain. Consumer subscription fees trickle up the infrastructure stack to fund global cable laying, power, and maintenance.

The Upstream Economic Cascade

1. End User: Pays a monthly bill to a retail ISP for home Wi-Fi to stream content (e.g., Netflix, YouTube).
2. Retail ISP: Pays IXP facilities for data center space, port access, and peering traffic.
3. IXP / Regional Data Center: Leases long-distance bandwidth capacity from primary WAN infrastructure owners.
4. WAN Infrastructure Owners: Reinvest revenue into capital expenditure (CapEx) for subsea optical cable deployment, power infrastructure, landing station cooling, and continuous ocean maintenance.

6. Complete End-to-End Traffic Walkthrough

Below is the vertical execution flow tracing how a user request (e.g., clicking a video stream) moves from a local machine out through physical global infrastructure and back:

User Device (Laptop / Smart TV)
   |
   | Sends HTTP/DNS Request over Home Wi-Fi (2.4GHz / 5GHz)
   v
Local Area Network (LAN) / Home Gateway Router
   |
   | Transmits signals through Fiber-to-the-Home (FTTH) cable
   v
Retail Internet Service Provider (ISP) - e.g., ACT Fibernet / JioFiber
   |
   | Routes packet into the local city network
   v
Metropolitan Area Network (MAN)
   |
   | Hands over packet at physical data center switching fabric
   v
Internet Exchange Point (IXP)
   |
   | Routes long-distance traffic onto national backbone
   v
Wide Area Network (WAN) Backbone (Telecom Infrastructure Owners)
   |
   | Converts electronic/optical signals into subsea optical fibers
   v
Subsea Fiber Optic Cable Landing Station
   |
   | Light signals travel at ~200,000 km/s through subsea glass fibers
   v
Remote Edge Data Center / Origin Application Server

7. Why DevOps & System Engineers Must Understand Physical Infrastructure

A common pitfall for software and DevOps engineers is treating the network as an abstract, infinite, zero-latency pipe. When debugging production issues, software-level abstractions (like Docker containers or Kubernetes clusters) are only half the picture.

Production Troubleshooting Realities

1. Subsea Cable Cuts & Packet Drops: If an undersea cable breaks due to seismic activity or ship anchors, traffic is dynamically rerouted across secondary fallback cables. This longer physical path instantly spikes round-trip time (Latency) and causes intermediate Packet Loss.
2. IXP Congestion & ISP Peering Bottlenecks: High latency or request timeouts during peak traffic hours are often caused by saturated peering ports at an IXP, rather than application code bottlenecks or microservice CPU limits.
3. Edge Deployment Decisions: Knowing where subsea landing stations and IXPs reside helps engineers choose optimal cloud deployment regions (e.g., AWS ap-south-1 in Mumbai) to minimize physical fiber transit distances for end users.

8. Summary and Key Takeaways

• Physical Reality of the Internet: The global internet is fundamentally a physical network composed of undersea fiber-optic cables, coastal landing stations, and terrestrial fiber.
• Landing Stations: Primary coastal nodes (e.g., Mumbai, Kochi, Trivandrum) where subsea cables connect to national land networks.
• WAN (Wide Area Network): Long-distance backbone infrastructure spanning nations, operated by telecom conglomerates due to massive capital requirements.
• MAN (Metropolitan Area Network): City-scale networks bridging national WAN backbones with localized delivery hubs.
• IXP vs. ISP: IXPs are neutral physical data centers where networks exchange bulk traffic, whereas ISPs deliver last-mile internet access to homes and offices.
• Economic Pipeline: Consumer ISP subscription fees flow upstream to fund IXPs, WAN owners, and subsea cable deployment.
• DevOps Mindset: Production performance issues like elevated latency, buffer bloat, and dropped connections often trace back to physical fiber routes, IXP peering constraints, and underlying infrastructure degradation—not just microservice logic.