- Atlas
- Computer Science
- Network Security
Network Security
Network security is the practice of protecting the links between computers: the traffic that crosses cables, radio, and the public internet, and the devices that decide what may cross.
10 frameworks, from 1980 to today · Cybersecurity · Computer Science
- Intrusion Detection and Prevention 1980-Present
- Incident Response 1988-Present
- Public Key Infrastructure 1988-Present
- Perimeter Security Model 1989-2010
- Protocol Vulnerability Analysis 1994-Present
- IPsec 1995-Present
- Transport-Layer Security 1995-Present
- DDoS mitigation 2000-Present
- Network Segmentation 2000-Present
- Zero Trust Architecture 2010-Present
Overview
Network security is the practice of protecting the links between computers: the traffic that crosses cables, radio, and the public internet, and the devices that decide what may cross1. Its central problem is that a network is shared ground. Once two machines are connected, anything one sends can be read, altered, replayed, or aimed at the other, and the defender usually cannot see the attacker. What distinguishes the field from systems or software security is that its object is the connection itself—its protocols, its addressing, its routing, and the choke points where traffic can be inspected or stopped. The history below follows how practitioners built those choke points, learned that the traffic passing through them could be attacked in its own right, and gradually stopped assuming that anything inside the network could be trusted.
Firewalls and the Perimeter Model
Through the 1980s, networks were mostly closed communities—university and corporate machines on local-area networks, joined to the outside world through a single router. The first widely used defence was the packet filter, which examines each packet's header fields (source and destination addresses and ports) against a rule list and drops what the rules forbid. The idea is usually credited to work at Bell Labs and to Jeff Mogul's 1989 paper on gateways, and by the early 1990s packet-filtering routers were standard equipment at the boundary between a trusted internal network and the untrusted internet.
Packet filters had a weakness: they judged each packet in isolation, so they could not tell whether a connection was legitimate. The circuit-level gateway, and then the application-level gateway or proxy, answered this by terminating connections and re-originating them on the far side, inspecting the application protocol in between. The landmark commercial product was the DEC SEAL, built at Digital Equipment Corporation's Western Research Laboratory around 1990–1991 by Marcus Ranum and colleagues; it was a proxy firewall for the internet gateway. In 1994 Ranum, at Trusted Information Systems, released the TIS Firewall Toolkit, a freely available set of proxy programs that made the approach widely reproducible.
The synthesis came with stateful inspection, commercialized by Check Point Software's FireWall-1 in the mid-1990s: the firewall keeps a table of active connections and admits return traffic belonging to a connection it has already approved, rather than writing a rule for every direction. This made firewalls practical at scale and cemented the perimeter security model—a hardened boundary with a defended interior, often drawn as a "demilitarized zone" of public-facing servers between two firewalls. The model's assumption was explicit: inside the perimeter, hosts could largely trust one another.
Intrusion Detection and the Limits of the Boundary
A firewall enforces policy; it does not report when policy is being probed or when an attacker is already inside. That gap produced intrusion detection. In 1980 James Anderson's report for the U.S. Air Force, Computer Security Threat Monitoring and Surveillance, framed the problem as detecting misuse through audit records. In 1987 Dorothy Denning published a model, developed with Peter Neumann at SRI International, that defined intrusion detection as comparing a system's observed behaviour against profiles of normal use—an anomaly-detection approach—alongside rule-based detection of known bad patterns.
Two operational systems followed at SRI: the Intrusion Detection Expert System (IDES), a real-time host monitor, and the Network Security Monitor (NSM), built by L. Todd Heberlein and colleagues and reported in 1990, which watched traffic on a network segment rather than a single machine. NSM established the network intrusion detection system as a distinct instrument. The 1990s brought signature-based systems that matched traffic against libraries of known attack patterns, and in 1998 the open-source Snort, begun by Martin Roesch, made network intrusion detection widely accessible. Intrusion prevention—dropping the offending traffic rather than only alerting—followed as inline deployment became common in the 2000s. The persistent difficulty, still unresolved, is the false-positive problem: anomaly detectors flag unusual but benign behaviour, and signature detectors miss anything new.
Protocol Vulnerabilities and the Discovery of the Network as an Attack Surface
As the internet opened to commerce in the 1990s, researchers began attacking the protocols themselves. In 1994, Kevin Mitnick's attack on Tsutomu Shimomura's machines exploited a weakness in TCP's sequence-number handling—predictable initial sequence numbers let an attacker forge packets that a host would accept as part of an established connection. In 1996, Laurent Joncheray published the technique of TCP session hijacking, showing that an on-path attacker could take over a connection. The same era produced the SYN flood, in which a flood of half-open connection requests exhausts a server's connection table; it was used in a notorious 1996 attack on Panix, a New York internet provider, and became the archetype of a denial-of-service attack.
The lesson practitioners drew was that a protocol designed for a cooperative network of peers had no defence against a hostile one. Defences were retrofitted: SYN cookies, proposed by Daniel Bernstein in 1996, let a server answer connection requests without storing state until the handshake completes. The broader response was a discipline of protocol vulnerability analysis—formal and informal study of what an adversary can do to a protocol's messages, ordering, and state—which fed into later design principles that assume the network is hostile.
Cryptography Moves into the Network Stack
The decisive shift was to stop trusting the path and protect the payload end to end. Public-key cryptography, published by Whitfield Diffie and Martin Hellman in 1976 and made practical for encryption and signatures by RSA in 1977, gave the field its tools. But a public key is only useful if you know whose it is, and that required infrastructure.
Public key infrastructure (PKI) is the system of certificate authorities that sign certificates binding a public key to an identity, so that a party can verify a key it has never seen. Its foundational standards were X.509, first issued by the ITU in 1988, and the PKIX profile work of the IETF in the late 1990s. PKI's weakness is trust concentration: compromise a certificate authority and every certificate it signed becomes suspect, a problem that surfaced repeatedly in the 2010s when authorities were found to have issued fraudulent certificates.
At the network layer, IPsec was standardized by the IETF in the mid-1990s (the architecture in RFC 2401, 1998) to authenticate and encrypt IP packets themselves, making it possible to build virtual private networks over untrusted links. At the transport layer, Netscape introduced the Secure Sockets Layer in the mid-1990s for web traffic; its successor, Transport Layer Security, was standardized by the IETF in 1999 (TLS 1.0, RFC 2246). TLS's history is a long retreat from weak choices: version 1.1 and 1.2 removed broken constructions, and TLS 1.3 (2018) removed entire categories of legacy cipher suites and cut the handshake to one round trip. The 2014 Heartbleed bug in OpenSSL's TLS implementation—a missing bounds check that leaked server memory—showed that even a sound protocol can be undone by its code.
Denial of Service and the Turn to Resilience
Denial of service grew from single-source floods into distributed attacks launched from networks of compromised machines. The high-water marks were the 2000 attacks that briefly took down major commercial sites, and the 2016 Mirai botnet, which conscripted poorly secured internet-connected cameras and routers and produced record-breaking traffic volumes. Defence evolved from filtering at the target to absorbing and scrubbing traffic upstream: content delivery networks and dedicated scrubbing centres spread attack traffic across capacity far larger than any single site, and anycast routing disperses it geographically. The field's framing shifted from prevention to resilience—designing so that an attack degrades service gracefully rather than stopping it—and to the economics of attack, since a defender must absorb the largest flood an adversary can rent.
Segmentation and the Erosion of the Trusted Interior
Even with encrypted links and firewalls, the perimeter model kept failing in the same way: an attacker who got one foothold inside could move laterally almost freely. The answer was network segmentation—dividing the interior into zones with enforced boundaries between them, so that compromise of one zone does not grant access to the next. Virtual local-area networks, internal firewalls, and later microsegmentation in virtualized data centres all implement the idea. Segmentation is the practical bridge to the present: it treats the interior as untrusted in pieces, which is the premise of what came next.
Zero Trust
Zero Trust Architecture states the premise outright: no user or device is trusted because of its network location, and every request must be authenticated and authorized as if it came from an untrusted network. The term was coined by John Kindervag at Forrester Research around 2010, building on earlier "de-perimeterization" arguments. Its implementation rests on strong identity, per-request policy enforcement, and continuous verification rather than a one-time login at the boundary. In 2020 the U.S. National Institute of Standards and Technology published SP 800-207, giving the approach a formal architecture and pushing it into government and enterprise planning. Zero trust is best understood as the culmination of the history above: each earlier failure—the trusted interior, the static rule, the single boundary—removed one more assumption that location confers safety.
Incident Response and the Institutional Layer
Attacks that succeed require a response, and the field built institutions for it. The first computer emergency response team, CERT/CC, was established at Carnegie Mellon University in 1988 in the wake of the Morris worm2. The Forum of Incident Response and Security Teams (FIRST) was founded in 1990 to coordinate national teams, and the IETF's RFC 2350 (1998) standardized how a response team publishes its charter and contact information. Incident response developed its own discipline—triage, containment, eradication, recovery, and the preservation of evidence—and its own tension with law enforcement and disclosure norms. Coordinated vulnerability disclosure, in which a finder reports a flaw privately and a fix is released before public details, became the field's working norm through the 2000s, though its timing and obligations remain contested.
The Present
Today's network security is layered and assumes hostility at every hop. Firewalls persist but are increasingly identity-aware and distributed; intrusion detection and prevention are embedded in network sensors and endpoint agents; TLS and IPsec carry most traffic; segmentation and zero-trust policy engines govern the interior; and denial-of-service defence is largely outsourced to providers with global scrubbing capacity. The open questions are the ones this history kept exposing: how to verify identity without a single point of trust, how to detect novel attacks without drowning in false alarms, how to secure the protocols and devices of a network of billions of constrained machines, and whether a defender can ever do better than absorb an attack large enough to overwhelm any provisioned capacity.
References
- Network security, Wikipedia
- Computer emergency response team (History), Wikipedia
The frameworks of Network Security
Each framework is a way of working the field has used: when it held, and the question it answers.
- 1980-Present
Intrusion Detection and Prevention
How do we tell when a network is under attack?
- 1988-Present
- 1988-Present
- 1989-2010
- 1994-Present
- 1995-Present
- 1995-Present
Transport-Layer Security
How can web traffic be kept private and intact?
- 2000-Present
- 2000-Present
- 2010-Present
Zero Trust Architecture
How do we secure a network with no trusted inside?
How the ideas connect
Each arrow runs from an earlier idea to the later one it shaped. The list below says what passed between them.
IPsec influenced by Public Key Infrastructure
IPsec's authentication and key exchange depend on public key infrastructure and certificates to authenticate peers and establish session keys.
Carried by IETF IPsec working group; RFC 2401 (1998).
Transport-Layer Security influenced by Public Key Infrastructure
TLS uses PKI certificates to bind a server's public key to its identity, so the web's transport security rests on the certificate authority system.
Carried by Netscape; RFC 2246 (1999).
Zero Trust Architecture reacted against Perimeter Security Model
Zero trust architecture defined itself against the perimeter model's premise that hosts inside the boundary could be trusted, replacing boundary-based trust with per-request verification.
Carried by John Kindervag.
Incident Response grew out of Intrusion Detection and Prevention
Intrusion detection supplied the detection and alerting that incident response teams acted on; the 1988 Morris worm, which prompted the first CERT coordination center, made the need for organized response to detected intrusions concrete.
Carried by CERT Coordination Center; Morris worm (1988).
Network Segmentation influenced by Perimeter Security Model
Network segmentation arose because the perimeter model's assumption of a trusted interior failed once attackers gained a foothold; segmentation replaced the single hardened boundary with enforced internal zones.
DDoS mitigation influenced by Protocol Vulnerability Analysis
Protocol vulnerability analysis of the SYN flood and related state-exhaustion attacks established the denial-of-service threat model that DDoS mitigation later engineered against.
Carried by Panix attack (1996); Daniel Bernstein.
Zero Trust Architecture grew out of Network Segmentation
Microsegmentation in virtualized data centres was a direct precursor of zero trust, extending zone-based segmentation down to individual workloads and identities.
Protocol Vulnerability Analysis influenced by Intrusion Detection and Prevention
Network intrusion detection's practice of watching traffic for anomalous patterns fed the study of protocol-level attacks, since detectors had to encode the protocol weaknesses attackers exploited.
Zero Trust Architecture influenced by IPsec
IPsec's ability to authenticate and encrypt traffic between any two endpoints provided a technical basis for zero trust's encrypted, identity-verified connections regardless of network location.
Zero Trust Architecture influenced by Transport-Layer Security
TLS made it practical to protect application traffic end to end over untrusted networks, supporting zero trust's move away from relying on a trusted internal network path.
Beyond the field
Where Network Security's ideas came from in other fields, and what they led to there: the strongest 12 of 36 links.
Intrusion Detection and Prevention led to Host-based intrusion detection system in Systems Security
Host-based intrusion detection grew directly out of the intrusion detection framework, applying Denning's anomaly and misuse models to audit logs on a single machine. Early host-based systems such as IDES and later Tripwire implemented that model locally.
Carried by Dorothy E. Denning.
Public Key Infrastructure grew out of Public-Key Cryptography in Applied Cryptography
Public key infrastructure grew directly out of public-key cryptography: Diffie and Hellman's 1976 scheme and RSA in 1977 supplied the primitives, and PKI added certificates, authorities and revocation to make them usable at scale.
Carried by Whitfield Diffie; Martin Hellman; Rivest, Shamir and Adleman.
Perimeter Security Model provoked Zero Trust Architecture in Systems Security
Zero trust architecture arose as an explicit rejection of the perimeter security model, arguing that once the perimeter is breached, implicit trust inside the network is indefensible; Kindervag's 2010 work framed zero trust as its replacement.
Carried by John Kindervag; Forrester Research.
Protocol Vulnerability Analysis grew out of Cryptographic protocol in Applied Cryptography
Protocol vulnerability analysis grew out of cryptographic protocol research: formal analysis of authentication protocols by Needham, Schroeder, Burrows, Abadi and Needham exposed flaws that motivated systematic adversarial analysis of protocols.
Carried by Roger Needham; Michael Schroeder; Burrows, Abadi and Needham.
IPsec grew out of Internet Protocol in Routing And Internetworking
IPsec was designed as a security extension of the Internet Protocol, adding authentication and encryption headers to IP packets; it is a direct outgrowth of the IP architecture.
Intrusion Detection and Prevention influenced by Reference monitor and trusted computing base in Cybersecurity
Anderson's 1980 report on computer security threats, which coined the term intrusion detection, built directly on the reference monitor and trusted computing base concepts from the Anderson report of 1972, treating audit data as a way to check whether the reference monitor's policy was being violated.
Carried by James P. Anderson.
Intrusion Detection and Prevention influenced Zero trust architecture in Cybersecurity
Intrusion detection supplied zero trust with the practice of continuously monitoring traffic and behaviour rather than trusting a static boundary. Zero trust's assumption of breach and continuous verification draws on the monitoring and anomaly-detection methods of IDS.
Intrusion Detection and Prevention influenced Incident response in Cybersecurity
Intrusion detection systems, developed from the 1980s, provided the detection and alerting that incident response procedures took as their trigger and evidence source.
Carried by Dorothy E. Denning.
Incident Response influenced by Vulnerability Discovery and Exploit Research in Software Security
Vulnerability discovery and exploit research produced the incidents that incident response was built to handle, and its disclosure practices shaped the reporting and coordination procedures of response teams from the late 1980s.
Incident Response influenced Coordinated vulnerability disclosure in Cybersecurity
Incident response practice, built around CERT coordination from 1988, gave coordinated vulnerability disclosure its model of a central coordinator receiving reports and managing the timeline of notification and fix. The disclosure process borrowed the coordinator role and the multi-party handling of a reported flaw.
Carried by CERT Coordination Center.
Public Key Infrastructure influenced by Cryptographic protocol in Applied Cryptography
Cryptographic protocol design, including authentication and key exchange protocols such as those analyzed by Needham and Schroeder, provided the message flows that PKI certificates were introduced to support.
Carried by Roger Needham; Michael Schroeder.
Perimeter Security Model influenced by Reference Monitor and Trusted Computing Base in Security Theory
The reference monitor and trusted computing base concept was extended from a single system to the network boundary: the firewall became the reference monitor for traffic between zones.

Network Security is a planet in Nousfarer
Its city has a tower for each of these 10 frameworks. Land, walk in, and climb one concept at a time.
Land on Network SecurityRelated fields
Fields its ideas are linked to
- Applied Cryptography 7 links
- Systems Security 7 links
- Cybersecurity 6 links
- Network Protocols 5 links
- Security Theory 2 links
- Software Security 2 links
- Communication Systems 1 link
- Technology And Internet Law 1 link
- Congestion Control 1 link
- Networks 1 link
- Routing And Internetworking 1 link
- Software Testing 1 link
More in Cybersecurity
- Applied Cryptography
- Cybersecurity
- Digital Forensics
- Digital Privacy
- Malware Analysis
- Security Economics
- Security Theory
- Software Security
- Systems Security
See where Network Security sits on the Map of Human Knowledge