The Physical Layer: What the Internet Is Made Of
The internet runs on physical infrastructure. Long-distance traffic travels on fiber optic cables — thin glass strands that carry data as pulses of light at speeds close to the speed of light in glass, roughly 200,000 km/s. Undersea fiber cables connect continents. Land-based fiber runs through conduits under roads and alongside railways. Copper wire (Ethernet, coaxial, telephone) handles shorter distances, typically the last mile to homes and buildings. Wireless links — Wi-Fi, 4G LTE, 5G — bridge the final gap to mobile devices. The physical medium determines the fundamental speed, latency, and reliability of each segment.
ISP Hierarchy: Access, Regional, and Tier 1
Internet service providers are organized into a rough hierarchy based on how much of the internet they can reach independently. Access ISPs connect end users — homes, businesses, schools — to the broader network. Regional ISPs aggregate traffic from access ISPs across a geographic area and connect to larger networks. Tier 1 ISPs operate global backbone networks and can reach every other network on the internet entirely through settlement-free peering agreements — they do not pay anyone for transit. Examples include AT&T, Lumen, NTT, and Telia. Every packet from a home user travels up through this hierarchy to reach its destination.
Peering and Transit: How ISPs Connect
ISPs connect to each other through two mechanisms. Transit is a paid arrangement where a smaller network pays a larger network to carry traffic to destinations the smaller network cannot reach directly. Peering is a mutual exchange arrangement where two networks of roughly equal standing agree to carry each other's traffic for free, because both benefit from the direct route. Internet Exchange Points (IXPs) are neutral physical facilities where many ISPs, CDNs, and cloud providers connect to each other in one location, dramatically reducing the number of hops and the cost of moving traffic between networks. Major IXPs such as DE-CIX in Frankfurt and AMS-IX in Amsterdam exchange hundreds of terabits per second.
Routing: BGP Across Autonomous Systems
Each independently operated network on the internet is an Autonomous System (AS), identified by an ASN. BGP (Border Gateway Protocol) is the routing protocol that connects ASes. BGP routers announce which IP address prefixes they can reach and via which paths. Other BGP routers use these announcements to build routing tables and select the best path based on policy, AS path length, and operator preferences. BGP does not optimize for latency or bandwidth — it follows policy. This is why your traffic sometimes takes a longer path than geography would suggest, and why BGP misconfigurations can occasionally redirect large swaths of internet traffic.
DNS Resolution: Root, TLD, Authoritative
Before any connection, a domain name must be resolved to an IP address. DNS resolution follows a hierarchy. Your device queries a recursive resolver (usually provided by your ISP or a public resolver like 8.8.8.8). If the answer is not cached, the resolver asks a root nameserver which TLD server handles .com. It then asks the .com TLD server which authoritative nameserver handles example.com. Finally it queries that authoritative server for the A or AAAA record. The answer is returned to your device and cached according to the TTL. This entire chain typically completes in 20–100 ms on a good connection. DNS latency adds to every new connection your browser opens.
The Packet Journey: From Browser to Server
When you type a URL and press Enter, the sequence of events involves every layer of the stack:
- The browser resolves the hostname via DNS.
- The OS opens a TCP connection (SYN → SYN-ACK → ACK) to the server's IP on port 443.
- TLS negotiation establishes an encrypted session (1–2 round trips for TLS 1.3).
- The browser sends an HTTP GET request inside the encrypted TLS stream.
- The server processes the request and sends the response in TCP segments.
- Each IP packet crosses your LAN, through your router, into the ISP access network, through regional and backbone links, possibly through an IXP, to a CDN edge or origin server.
- Responses traverse back along a potentially different path.
- The browser reassembles the response, decrypts it, parses HTML, and begins loading sub-resources — each triggering additional DNS lookups and connections.
Content Delivery: What "The Cloud" Actually Is
CDNs (Content Delivery Networks) place cached copies of content at edge nodes distributed around the world — often co-located at IXPs or in ISP data centers close to end users. When you load a major website, you are typically not connecting to a single origin server in one data center. You are connecting to the nearest CDN edge node, which may be within 5–20 ms of you. The CDN serves cached content directly and only contacts the origin for cache misses or dynamic content. "The cloud" is physically thousands of data centers and networking facilities operated by companies like AWS, Google, Microsoft, Akamai, Cloudflare, and Fastly. Traffic routing, redundancy, and load balancing across these facilities are managed by BGP and proprietary overlay networks.
Where Latency Comes From
Latency accumulates at every stage of the journey. Speed-of-light propagation delay accounts for roughly 1 ms per 200 km in fiber. Processing delay at each router hop adds microseconds. Queuing delay at congested links can add milliseconds or more. DNS resolution adds the lookup chain latency. TCP connection setup adds one round trip. TLS negotiation adds one or two more. Server processing time — database queries, application logic — adds its own delay. A web page that feels slow is usually not bottlenecked by raw download speed but by accumulated round trips, DNS latency, server response time, and the number of sequential requests the page structure requires.
The Main Pieces
| Piece | What it does | Visible symptom when broken |
|---|---|---|
| Physical layer | Carries bits as light, electricity, or radio | Complete outage or severe degradation |
| ISP access link | Connects your home to the provider | Slow or offline for every device |
| DNS | Turns names into IP addresses | Sites fail by name but work by IP |
| BGP routing | Moves packets between autonomous systems | Only some destinations are slow or unreachable |
| Transport (TCP/UDP/QUIC) | Delivers application data reliably or quickly | Loss, retransmits, or connection failures |
| CDN / application | Serves content from nearby edge servers | One site slow while others are fine |
Frequently Asked Questions
What is the internet in simple terms?
The internet is a global network of networks. It lets devices exchange packets using shared protocols, especially IP, TCP, UDP, DNS, TLS, and HTTP.
Who owns the internet?
No single company owns the internet. It is made of many independently operated networks, including ISPs, cloud providers, CDNs, governments, universities, and private companies that interconnect through standards and agreements.
Why does internet speed vary by website?
Different websites use different servers, CDNs, routing paths, cache behavior, compression, and application code. Your access speed matters, but the remote service and the path to it matter too.