Foundations of Connectivity
Before a host can talk to anything it must be reachable physically and configured logically. This lesson defines what connectivity really means at both levels, then explains how data rates, delay, and message types shape every conversation on a network.
What connectivity means
Connectivity is the ability of one host to exchange data with another. It is not a single thing you can point at — it is the result of two independent layers working together:
- Physical connectivity — is there a working path (cable, fibre, or radio) that can carry a signal between the devices?
- Logical connectivity — does the host know where to send the data, and is it addressed consistently with everything around it?
Both must be correct. A PC with a cable plugged in but no IP address is physically connected yet logically isolated: its signal leaves the NIC, but nothing knows where to deliver it. Conversely, a perfectly configured PC with a cut cable cannot reach anything. Troubleshooting always starts by asking which kind of connectivity is broken.
Physical connectivity
Physical connectivity is built from three parts: the NIC, the media, and the connectors.
The network interface card (NIC) is the hardware that joins a host to the medium. It is usually built into the motherboard but can be an expansion card. The NIC holds the device's MAC address and converts data between the digital form used by the host and the signals used on the wire or in the air.
The media is the physical path the signals travel along. The three families are copper, fibre, and wireless:
| Medium | Signal | Typical connector | Notes |
|---|---|---|---|
| Copper twisted pair (UTP/STP) | Electrical | RJ-45 (8P8C) | Low cost; Ethernet over twisted pair is limited to 100 m; susceptible to interference. |
| Fibre optic | Light | LC, SC, ST | Long distance, high bandwidth; immune to electromagnetic interference (EMI). |
| Wireless | Radio | Antennas (no cable) | Mobility; affected by interference, distance, and obstacles. |
Copper cable uses twisted pairs to reduce interference and is terminated with the familiar RJ-45 plug. Common categories are Cat 5e, Cat 6, and Cat 6a, which differ in the speeds they support. Fibre optic carries data as pulses of light: single-mode fibre supports the longest distances, while multimode fibre is used for shorter runs such as within a building or data centre. Wireless media use radio waves (the IEEE 802.11 family, known as Wi-Fi).
Logical connectivity
Once the physical path exists, the host must be configured so traffic can actually be delivered. Four settings give a host everything it needs to communicate both locally and across the Internet:
- IP address — the host's logical, routable address on the network. It identifies the host to the rest of the internetwork.
- Subnet mask — tells the host which part of its IP address is the network portion and which is the host portion. It is how the host decides whether a destination is local or must be reached through a router.
- Default gateway — the IP address of the router on the local network that the host sends traffic to when the destination is outside its own subnet.
- DNS server — the server that translates human-readable names such as
example.cominto IP addresses. The Domain Name System listens on UDP/TCP port53.
A host learns these settings either statically (typed in by an administrator) or automatically from a DHCP server. DHCP can supply the address, the mask, the gateway, and the DNS server in one exchange.
Addressing and transmission
Communication depends on addresses being unique at every layer, because each layer answers a different question. If two devices in the same scope share an address, delivery breaks.
| Layer | Address | Where it must be unique | Failure if duplicated |
|---|---|---|---|
| Layer 2 | MAC address | On the local link (LAN) | Frames are delivered to the wrong device. |
| Layer 3 | IP address | Across the internetwork (globally, or within the private scope) | An IP conflict: traffic is misdelivered or dropped. |
| Layer 4 | Port number | Per host, per protocol, per connection | Data is handed to the wrong application. |
The MAC address is unique on the local network and is rewritten at every router hop. The IP address must be unique end to end. A socket — the combination of an IP address and a port number — identifies one specific conversation between two applications.
Unicast, multicast, and broadcast
A message can be addressed to one, some, or all recipients. These three modes appear constantly in exams and in protocol design.
| Mode | Destination | Typical example |
|---|---|---|
| Unicast | One sender to one receiver | A web request to a server; by far the most common traffic |
| Multicast | One sender to a group of receivers | Streaming video to subscribed viewers; IPv4 group range 224.0.0.0–239.255.255.255 |
| Broadcast | One sender to all hosts on the local network | ARP requests; IPv4 broadcast 255.255.255.255 (limited) or the subnet broadcast |
Recall from Lesson 2 that the switch floods broadcast and multicast frames out of every port in the VLAN except the ingress port. At Layer 2 the broadcast address is ff:ff:ff:ff:ff:ff. IPv4 defines broadcasts; IPv6 has no broadcast and instead uses multicast (for example ff02::1, the all-nodes address).
Bandwidth vs throughput
These two terms are related but not the same, and exams test the difference.
- Bandwidth — the maximum capacity of a link: how many bits it can carry per second. It is fixed by the hardware and the standard (for example 1 Gb/s).
- Throughput — the actual rate of successful data delivery, which is always less than or equal to the bandwidth. Overhead, latency, congestion, and errors reduce throughput.
b/s, Kb/s (1,000 b/s), Mb/s (1,000,000 b/s), Gb/s (1,000,000,000 b/s), and Tb/s.b/s; file sizes use B. Confusing them is one of the most common mistakes in networking.| Standard Ethernet speed | Written as | Bits per second |
|---|---|---|
| Ethernet | 10 Mb/s | 10,000,000 |
| Fast Ethernet | 100 Mb/s | 100,000,000 |
| Gigabit Ethernet | 1 Gb/s | 1,000,000,000 |
| 10 Gigabit Ethernet | 10 Gb/s | 10,000,000,000 |
Latency, jitter, duplex and the data flow
Latency (also called delay) is the time it takes data to travel from source to destination. It is the sum of several parts:
- Propagation delay — the time for the signal to physically travel the medium.
- Transmission (serialisation) delay — the time to place all the bits onto the link, which depends on frame size and link speed.
- Processing delay — the time devices spend examining headers and making decisions.
- Queuing delay — the time a frame waits in a buffer when a link is congested.
Adding the time there and back gives the round-trip time (RTT), the figure the ping command reports. Jitter is the variation in latency between successive packets. A steady delay is tolerable, but variable delay causes choppy voice and video, so real-time traffic uses a buffer to smooth jitter out before playback. A network optimised for real-time media needs both low latency and low jitter.
Duplex describes the direction(s) of data flow on a link. Half-duplex links carry one direction at a time and use CSMA/CD to recover from collisions. Full-duplex links carry both directions at once on a point-to-point connection, so collisions cannot occur. Modern switch-to-host and switch-to-switch links run full-duplex; a duplex mismatch degrades performance even though the link stays up.
The client-server data flow
With connectivity in place, a typical exchange between a client and a server follows a predictable path. Using the settings above, the client browses example.com:
- The client uses DNS (the configured DNS server) to translate
example.cominto an IP address. - Because the server's address is outside the host's subnet (
192.168.1.0/24), the host sends the traffic to its default gateway,192.168.1.1. - The frame travels to the switch, which forwards it within the LAN at Layer 2.
- The router accepts the frame, keeps the IP addresses, and builds a new frame for the next hop toward the Internet.
- The request crosses the Internet to the server, which returns a response along the reverse path.
Review
Check what you have learned with these review questions.
Sources & references
- Postel, J. RFC 791: Internet Protocol (1981) — 32-bit IPv4 addresses and the datagram header. rfc-editor.org/rfc/rfc791.
- Braden, R. RFC 1122: Requirements for Internet Hosts — Communication Layers (1989) — host addressing and the requirements for communication. rfc-editor.org/rfc/rfc1122.
- Droms, R. RFC 2131: Dynamic Host Configuration Protocol (1997) — automatic host configuration. rfc-editor.org/rfc/rfc2131.
- Alexander, S. & Droms, R. RFC 2132: DHCP Options and BOOTP Vendor Extensions (1997) — subnet mask (option 1), router/default gateway (option 3), and DNS (option 6). rfc-editor.org/rfc/rfc2132.
- IANA. IPv4 Address Space Registry and Service Name and Transport Protocol Port Number Registry. iana.org/assignments/ipv4-address-space.
- IEEE. IEEE 802.11™ Wireless LAN (wireless media) and IEEE 802.3™ Ethernet (copper and twisted-pair links). standards.ieee.org.
- Cisco. CCNA (200-301) Exam Topics, Network Fundamentals. cisco.com.
Target exams: CCNA 200-301 v1.1 · CyberOps Associate 200-201. Last content verification: 2026-10-11.