Understanding Your IP Address & Network Basics

A public IP is the address the internet uses to route traffic back to your connection. Every site you visit necessarily sees it, because without it there is nowhere to send the response.

From it, a site can look up your internet provider and an approximate location. The accuracy of that location is routinely overstated: city-level is often roughly right, and street-level is not. Geolocation databases map address ranges to places, and those mappings are frequently stale, sometimes assigned to a provider's regional hub rather than the customer, and occasionally badly wrong.

Worth knowing: your address is likely not permanent. Most residential connections get a dynamic address that changes when the router reconnects or the provider's lease expires. And a growing number of connections — mobile especially — sit behind carrier-grade NAT, where hundreds of subscribers share one public address, which is why an IP-based block can catch people who did nothing.

Check your computer's network settings and you will usually see something like 192.168.1.42, which is not the address the internet sees. These are private ranges, reserved for use inside networks and not routable across the internet.

Your router performs Network Address Translation: it rewrites outgoing packets to use its single public address, remembers which internal device each conversation belongs to, and rewrites the replies on the way back. Every device on your network shares one public identity.

This has a side effect people rely on without realising. Because incoming connections have no internal address to go to unless one is configured, NAT incidentally blocks unsolicited inbound traffic — which is why devices on a home network are not directly reachable from the internet, and why hosting a game server needs port forwarding set up deliberately.

A subnet written as 192.168.1.0/24 defines a range of addresses. The number after the slash is the prefix length — how many leading bits are fixed as the network portion — and everything left over identifies individual hosts.

The pattern is that each additional bit of prefix halves the range. A /24 is the familiar home network of 254 usable addresses; a /30 provides exactly two, which is the standard allocation for a point-to-point link between two routers.

Two addresses in every subnet are reserved and cannot be given to a device: the first identifies the network itself, and the last is the broadcast address. This is why the usable count is always two below the total, and why a /31 is a special case that only works under a specific convention for point-to-point links.

IPv4 has about 4.3 billion addresses, which seemed generous when the protocol was designed and ran out in practice years ago. NAT is the workaround that has kept it viable, at the cost of the internet's original design where every device was directly addressable.

IPv6 uses 128-bit addresses, giving a number of possible addresses large enough that exhaustion is not a consideration on any timescale worth discussing. Addresses are written as eight groups of four hexadecimal digits, with a run of zero groups abbreviated to :: exactly once per address.

Adoption has been slow but is now substantial, and most connections run both protocols simultaneously. The practical consequence for troubleshooting is that a device has more than one address, and a service reachable over one protocol may not be reachable over the other — which produces failures that look intermittent until you notice which protocol each attempt used.

A MAC address identifies a network interface on the local network segment. Unlike an IP, it is assigned by the manufacturer and does not change as the device moves between networks — and unlike an IP, it does not travel beyond the local network, so websites cannot see it.

The first three bytes are the Organisationally Unique Identifier, registered to a manufacturer with the IEEE. Looking it up tells you who made the interface, which is genuinely useful when auditing what is connected to a network: an unrecognised device on a home network resolves to a make, and that is usually enough to identify the smart plug or console you forgot about.

These get conflated constantly, and they cause different problems. Bandwidth is how much data can flow per second. Latency is how long a single packet takes to make the round trip. Upgrading the first does nothing for the second.

• →Latency governs anything interactive — video calls, gaming, remote desktop, and how responsive a website feels while loading. Below about 50 ms is comfortable; above 150 ms, conversation starts to overlap awkwardly.
• →Bandwidth governs throughput — how fast a large download completes, how many simultaneous video streams a household supports.
• →Jitter is the variation in latency, and for real-time audio it matters more than the average. Consistent 80 ms is fine; an average of 40 ms that spikes to 300 ms produces the choppy call.
• →Packet loss forces retransmissions. Even a few percent degrades a video call sharply, while barely affecting a file download that can simply request the missing pieces again.

There is a floor set by physics. Light in fibre covers roughly 200 km per millisecond, so a round trip between London and Sydney cannot beat about 160 ms no matter what you pay. When a service feels slow from far away, the fix is a server closer to the user, not a faster connection.

URLs are the addresses of the web, and being able to read one in parts makes debugging redirects, tracking parameters and API calls considerably easier.

https://shop.example.co.uk:443/products/shoes?colour=blue&size=9#reviews
└─┬─┘   └──────┬────────┘└┬┘└─────┬─────┘└────────┬────────┘└──┬──┘
scheme      host       port    path            query        fragment

• →The host is read right to left: .uk is the top-level domain, example.co.uk the registered domain, and shop a subdomain. Phishing exploits this — example.com.attacker.net is a page on attacker.net, and the leading familiar name is there to be misread.
• →The query string carries key-value pairs after the ?, separated by &. Tracking parameters such as utm_source live here and can usually be removed without changing what page you get.
• →The fragment after # is never sent to the server. It is handled entirely by the browser, which is why it does not appear in server logs and cannot be read by the site's backend.
• →Characters outside a restricted ASCII set must be percent-encoded, which is why a space becomes %20 and why non-English text in a URL appears as long sequences of percent codes.