EVOPEDIA

← all systems · see it in The System

Computer network

A group of communicating computers and peripherals (hosts) that send data to one another by communication protocols, through networking hardware (network interface controllers, repeaters, switches, routers, modems, firewalls) and transmission media (copper, optical fibre, radio). Walked as the thing network engineers describe: its media, links, nodes, addresses, protocols in layers, routing, congestion control and topology, from a single local area network up to the wide area networks it joins; and as the field that builds and studies it.

Properties published scienceour proposal

Sourced fields, and our own readings.

born at1940 (Stibitz's teletype at Dartmouth to the Complex Number Calculator in New York); packet-switched: NPL network 1968-69, ARPANET's first four nodes 1969
born precisionyear
boundaryInside: the hosts' network interfaces and protocol software, the links, and the nodes that forward between them (repeaters, hubs, bridges and switches, routers), under the addressing and protocols of one administration or one technology. Outside: the people and programs that use it, the operating systems the protocol software runs in (shared with the host), the physical medium's wider field, and the other networks it is joined to through routers. 'The cloud ... distinguishes between the nodes on the inside that implement the network ... and the nodes on the outside of the cloud that use the network' (src:pd-requirements). The boundary is recursive: 'a network can be constructed from a nesting of networks' (src:pd-requirements).
nested levelBuilt from hosts' network interfaces, transmission media and forwarding nodes; embedded in an organisation's administration and joined by routers into internetworks, the largest being the Internet, where a whole network is one 'directly connected network' seen from IP.
scaletechnological

Measures (23) published science

characteristic clock: transcontinental round-trip time100 ms
characteristic clock: Ethernet backoff slot51.2 μs (waits of k slots, k random in 0..2^n - 1 after n collisions; up to 16 tries)
characteristic clock: distance-vector advertisement period (RIP)30 s
AQM target setpoint (CoDel)5-10 % of the round-trip time (1-20 ms in tests)
signal speed in optical fibre2.0e8 m/s (copper 2.3e8, vacuum 3.0e8)
ARPANET circuit rate, 196950 kbit/s (first four nodes)
Ethernet top standard rate800 Gbit/s (IEEE, 2024)
minimum Ethernet frame512 bits (64 bytes); data 46 to 1500 bytes
first commercial computer modem rate (Bell 101, 1958)110 bit/s
classical Ethernet reach2500 m (no more than four repeaters between hosts)
twisted-pair segment length before a repeater100 m
payload share of a full Ethernet packet (TCP/IPv4)1460 of 1500 bytes (per-hand-off efficiency 0.973 at the IP layer, derived from the two sourced figures)
error check overhead on Ethernet32 per 12,000 bits (CRC-32 per full frame)
optical loss of the clearest glass, 1966200 dB/km (Kao and Hockham foresaw 20 dB/km; Corning reached 17 dB/km in 1970)
congestion collapse, October 198632 Kbps to 40 bps throughput LBL to UC Berkeley (a factor of about a thousand)
redundancy level for survivability (Baran)about 3 links per node (1.5 enough without attack)
reorganisation time after half the network destroyed (simulation)< 1 s of simulated time
telephone tandem link limit (AT&T, c. 1960)5 switched tandem links
one-way delay that feels transparent< 150 ms (400 ms the planning limit)
Mirai peak infections600,000 devices (attack on Krebs above 600 Gbps)
node failures tolerated with unchanged diameter (scale-free model)5 % of nodes
MAC address size6 octets (first three name the maker)
distance-vector 'infinity'16 hops

Open contentions (6) open contention

Where sources or readings disagree. Both stay on the map.

Partners and parts published science

Built from / builds (11)

Sits inside / holds (6)

Trades with (valence) (26)

expressed by (29)

floor filled by (36)

Kinds (14)

Notes published science

Layers

The ways it takes in its world, where they meet, and how it acts back, each from L1 (where it meets the world) up.

the boardroom (shared floors)

  1. L3: The network layer in every router and switch: control plane (routing table built from routing protocols, configuration and learned addresses) and forwarding plane (each packet matched against the forwarding table), with an output queue per link
  2. L4: Transport in the end hosts: segmentation, retransmission, flow control and congestion control; only the hosts run it
  3. L5: Sessions between applications: setup, checkpoint, restart, teardown
  4. L6: Formats, encryption and compression agreed between peers
  5. L7: Applications and services (web, mail, file sharing, DNS, DHCP) and management across all layers (CMIP, network management, SDN controller applications), where the owners' purposes for the network are set

the computer network as network engineers draw it: signals on the medium up to the applications people use (the OSI seven layers; reader's view)

  1. L1: Physical layer: signals on copper, optical fibre or radio; network interface circuits, repeaters and hubs
  2. L2: Data link layer: frames between directly connected nodes; medium access (CSMA/CD, CSMA/CA, token), MAC addresses, error detection (CRC-32); bridges and switches forward by MAC address
  3. L3: Network layer: packets addressed and routed between networks by routers; routing tables built by routing protocols (distance vector, link state) or by hand; queues at each output link
  4. L4: Transport layer, in the hosts: segments, end-to-end reliability, flow control and congestion control
  5. L5: Session layer: setting up, managing and tearing down sessions; checkpointing and restart (in the Internet, folded into applications and transport)
  6. L6: Presentation layer: data formats, translation, encryption and compression
  7. L7: Application layer: the services people use (web, email, file sharing, printing, DNS, DHCP) and the management functions that configure and monitor the network

the network as a graph: nodes, links and how they are joined (topology and robustness; secondary stack)

  1. L1: Nodes and links: hosts, switches and routers joined by point-to-point or multiple-access links
  2. L2: Topology: bus, star, ring, mesh, tree, fully connected; logical topology may differ from physical layout
  3. L3: Redundancy: extra links so that surviving nodes stay connected (Baran's redundancy levels)
  4. L4: Degree structure and its robustness to random failure and to targeted attack (contested for the router-level Internet)
  5. L5: Value against size: Metcalfe's n squared and its rivals (n log n, n, 2^n, n cubed)

way in: Acknowledgments returning from receivers

  1. L1: Acknowledgment signals on the return path, spaced by the bottleneck
  2. L2: Ack frames on each link of the return path
  3. L3: Ack packets routed back to the sending host

way in: Address resolution and configuration requests (ARP, DHCP)

  1. L1: A request broadcast to every station on the local network
  2. L2: The ARP module reads the request; the owner of the address answers; the sender's pair is cached in the translation table

way in: Hostile traffic (scans, worms, floods)

  1. L1: Traffic from hundreds of thousands of compromised devices, or probes and worms
  2. L2: Frames arriving at the network edge; port-based access control (802.1X) asks who is attaching

way in: Link and node failures, damage and cuts

  1. L1: A link or node stops: damage, fire, power failure, flooding, a cut
  2. L2: The link is found down; learned entries age out; switches reconfigure the spanning tree

way in: Load on the queues (arrivals faster than a link drains)

  1. L1: Several input streams arriving for one outgoing link
  2. L2: Frames buffered in the output queue; queue occupancy and each packet's sojourn time measured

way in: Noise, interference and attenuation (bit errors)

  1. L1: Noise, crosstalk, interference and attenuation corrupting bits; worse on radio
  2. L2: The frame check: CRC-32 over the frame; a frame that fails is discarded (or corrected where codes allow)

way in: Operators' configuration and policy

  1. L1: Administrators' commands and policies (static routes, access rules, VLANs)
  2. L2: Each device's configuration interface, or one controller program over many devices (SDN)

way in: Routing and topology messages from neighbouring switches and routers

  1. L1: Distance-vector advertisements every 30 s, link-state packets flooded to all, spanning-tree configuration messages between switches
  2. L2: Spanning-tree configuration messages read at the link: claimed root, distance to it, sender

way in: The shared medium's state: idle, busy or colliding

  1. L1: Energy on the shared cable or radio channel: another station transmitting, or two at once
  2. L2: The adaptor's medium access logic: carrier sense, collision detection (wired) or acknowledgment (radio), random exponential backoff

way in: Traffic from attached hosts and neighbouring nodes (frames and packets)

  1. L1: Bits arriving as voltage, light or radio on a port
  2. L2: The frame read at the port: preamble to synchronise, 48-bit addresses, type field naming the protocol above, data up to 1500 bytes, CRC checked

way out: Advertise routes and topology

  1. L1: Flooded or sent on every directly connected link
  2. L3: Routes or link states sent to neighbours, periodically and on change

way out: Deliver to applications and people

  1. L1: On the host's display or speaker, where people meet it
  2. L7: The receiving application gets the data (a page, a message, a file)

way out: Drop or mark (the congestion signal)

  1. L1: The packet never leaves; the gap travels on as a missing acknowledgment
  2. L3: The queue manager drops early, or marks, when the standing queue exceeds target

way out: Forward and transmit

  1. L1: Bits leave as voltage, light or radio
  2. L2: Framed and sent when the medium allows (carrier sense, backoff)
  3. L3: The forwarding table picks the outgoing interface

way out: Refuse (firewall, access control)

  1. L1: The refused traffic goes no further onto the inner links
  2. L3: Rules reject unrecognised sources and allow recognised ones

way out: Slow down (window reduction, backoff)

  1. L1: Fewer packets on the links
  2. L4: The sender's window shrinks on loss and grows on acks

Who fills each layer

Sources (45) published science

Computer network (Wikipedia, via Wikimedia Enterprise snapshot 2026-11-01)
RFC 1122, Requirements for Internet Hosts: Communication Layers (1989) (behind its links)
Intrusion detection system (Wikipedia, via Wikimedia Enterprise snapshot 2026-11-01) (behind its links)
Firewall (computing) (Wikipedia, via Wikimedia Enterprise snapshot 2026-11-01) (behind its links)
Wireless network (Wikipedia, via Wikimedia Enterprise snapshot 2026-11-01) (behind its links)
Packet switching (Wikipedia, rev 1376334725, 2026-09-23, via Wikimedia Enterprise; read by section) (behind its links)
Baran, P. (2002) The beginnings of packet switching: some underlying concepts. IEEE Communications Magazine 40(7):42-48 (doi:10.1109/MCOM.2002.1018006) (behind its links)
RFC 7567: IETF Recommendations Regarding Active Queue Management (F. Baker, G. Fairhurst, eds., July 2015) (behind its links)
Prabhakar, B., Dektar, K. N. & Gordon, D. M. (2012) The regulation of ant colony foraging activity without spatial information. PLoS Computational Biology 8(8):e1002670 (doi:10.1371/journal.pcbi.1002670) (behind its links)
Graham, D. J. (2014) Routing in the brain. Frontiers in Computational Neuroscience 8:44 (doi:10.3389/fncom.2014.00044) (behind its links)
Peterson & Davie, Computer Networks: A Systems Approach (CC BY 4.0), §2.6 Multi-Access Networks (Ethernet) (behind its links)
Peterson & Davie, Computer Networks: A Systems Approach (CC BY 4.0), §1.5 Performance (behind its links)
Peterson & Davie, Computer Networks: A Systems Approach (CC BY 4.0), §3.4 Routing (behind its links)
RFC 8289: Controlled Delay Active Queue Management (K. Nichols, V. Jacobson, A. McGregor, J. Iyengar, January 2018) (behind its links)
OSI model (Wikipedia, via Wikimedia Enterprise snapshot 2026-11-01) (behind its links)
Peterson & Davie, Computer Networks: A Systems Approach (CC BY 4.0), §2.4 Error Detection (behind its links)
Kao, C. K. (2009) Sand from centuries past: send future voices fast. Nobel Lecture, 8 December 2009 (behind its links)
Jacobson and Karels, Congestion Avoidance and Control (November 1988; SIGCOMM '88, doi:10.1145/52324.52356) (behind its links)
ITU-T Recommendation G.114 (05/2003): One-way transmission time (behind its links)
Antonakakis, M. et al. (2017) Understanding the Mirai Botnet. 26th USENIX Security Symposium (behind its links)
Albert, Jeong and Barabasi, Error and attack tolerance of complex networks, Nature 406, 2000 (arXiv cond-mat/0008064) (behind its links)
Peterson & Davie, Computer Networks: A Systems Approach (CC BY 4.0), §3.1 Switching Basics and §3.2 Switched Ethernet (learning bridges, spanning tree) (behind its links)
Verma A, Pedrosa L, Korupolu M, Oppenheimer D, Tune E, Wilkes J (2015). Large-scale cluster management at Google with Borg. EuroSys 2015. doi:10.1145/2741948.2741964 (behind its links)
Peterson & Davie, Computer Networks: A Systems Approach (CC BY 4.0), §2.7 Wireless Networks (behind its links)
Peterson & Davie, Computer Networks: A Systems Approach (CC BY 4.0), §1.3 Architecture (behind its links)
RFC 826: An Ethernet Address Resolution Protocol (D. Plummer, November 1982) (behind its links)
Router (computing) (Wikipedia, via Wikimedia Enterprise snapshot 2026-11-01) (behind its links)
Peterson & Davie, Computer Networks: A Systems Approach (CC BY 4.0), §6.1-6.4 Congestion control and avoidance (RED, DECbit, ECN) (behind its links)
Network switch (Wikipedia, via Wikimedia Enterprise snapshot 2026-11-01) (behind its links)
Feamster, N., Rexford, J. & Zegura, E. (2014) The Road to SDN: An Intellectual History of Programmable Networks. ACM SIGCOMM Computer Communication Review 44(2):87-98 (doi:10.1145/2602204.2602219) (behind its links)
Wikipedia: Operating system (behind its links)
Yang, L., Chen, Z., Wang, C., Zhang, Z., Booma, S., Cao, P., Adam, C., Withers, A., Kalbarczyk, Z., Iyer, R. K. & Wang, G. (2024). True Attacks, Attack Attempts, or Benign Triggers? An Empirical Measurement of Network Alerts in a Security Operations Center. 33rd USENIX Security Symposium (open access) (behind its links)
Cichonski P, Millar T, Grance T, Scarfone K (2012) Computer Security Incident Handling Guide. NIST Special Publication 800-61 Rev. 2. doi:10.6028/NIST.SP.800-61r2 (US government work, not subject to copyright) (behind its links)
Botnet (Wikipedia, via Wikimedia Enterprise snapshot) (behind its links)
Kopp, Dietzel, Hohlfeld 2021, DDoS Never Dies? An IXP Perspective on DDoS Amplification Attacks, PAM 2021 (arXiv:2103.04443) (behind its links)
Information security (Wikipedia, via Wikimedia Enterprise snapshot) (behind its links)
Span of control (Wikipedia, via Wikimedia Enterprise snapshot) (behind its links)
Computer hardware (Wikipedia, via Wikimedia Enterprise snapshot) (behind its links)
IEEE 802.11 (Wikipedia, via Wikimedia Enterprise snapshot 2026-11-01) (behind its links)
DeCandia G et al. (2007) Dynamo: Amazon's Highly Available Key-value Store. SOSP 2007. doi:10.1145/1294261.1294281 (behind its links)
Cyber threat intelligence (Wikipedia, via Wikimedia Enterprise snapshot) (behind its links)
Communication (Wikipedia, via Wikimedia Enterprise snapshot 2026-11-01) (behind its links)
Telephone (Wikipedia, via Wikimedia Enterprise snapshot 2026-11-01) (behind its links)
Peterson & Davie, Computer Networks: A Systems Approach (open edition, CC BY 4.0), §1.2 Requirements (behind its links)
Wang, C., Wang, Y. & Jiang, C. (2023) Emergence of Metcalfe's Law: Mechanism and Model (arXiv:2312.11110v2) (behind its links)