EtherNet/IP was designed so that any device that can reach a controller on TCP or UDP port 44818 can ask it to identify itself, open a connection and, in many configurations, change its program or mode, with no cryptographic proof of who is asking. That trust model still describes most installed Logix and multi-vendor CIP networks. ODVA's CIP Security extension adds TLS, DTLS, device certificates and user authentication, but only on devices that support it and only when it is configured and deployed correctly. This guide explains how EtherNet/IP and CIP traffic works on the wire, what each CIP Security profile actually protects, how to deploy it with zones and conduits without breaking production, how to protect the large population of devices that will never support it, and what to monitor. It is written for OT engineers, control system integrators and security teams responsible for plants that run on EtherNet/IP. For general controller hardening, see the PLC Security Guide; this page goes deeper on the protocol and its security extension.
Key takeaways
- Classic EtherNet/IP has no authentication. Explicit messaging on port 44818 and cyclic I/O on UDP 2222 carry no proof of identity, so network reachability is the only access control unless CIP Security or controller-side protections are in place.
- CIP Security is a set of profiles, not a switch. The EtherNet/IP Confidentiality Profile adds TLS and DTLS with X.509 or pre-shared key device authentication, the CIP User Authentication Profile adds user identity via OpenID Connect, and a Resource-Constrained profile serves small devices.
- Integrity matters more than encryption. Message integrity and device authentication stop spoofed commands and man-in-the-middle tampering; confidentiality is optional and adds load, so choose it where data or credentials justify it.
- Brownfield is the real problem. Most installed devices will never support CIP Security, so zones, conduits, industrial firewalls with CIP inspection, controller mode switches and proxies carry the protection for years.
- Device certificates are non-human identities. The certificate authority that issues them, the policy server that deploys them and the engineering clients that hold trust are now critical identity infrastructure and need lifecycle management.
How EtherNet/IP and CIP work on the wire
EtherNet/IP carries the Common Industrial Protocol (CIP) over standard TCP/IP and UDP. CIP is the object-oriented application layer shared with DeviceNet and ControlNet: devices expose objects (Identity, Message Router, Connection Manager, Assembly and vendor objects), and clients invoke services on them. Two traffic types matter for security.
- Explicit messaging: request and response traffic over TCP 44818 (unconnected messages through the UCMM, or connected Class 3 messaging). Engineering software uses it to upload and download programs, change controller mode, read and write tags and flash firmware. HMIs and historians use it to read and write tags.
- Implicit (I/O) messaging: cyclic Class 0 and Class 1 connections, usually UDP 2222, often multicast, carrying I/O data between controllers, adapters and drives at fixed requested packet intervals. It is time-sensitive and unacknowledged.
- Encapsulation layer: every EtherNet/IP packet carries an encapsulation command. ListIdentity (0x63) answers with vendor, product, revision and serial number to anyone who asks, including over UDP broadcast; RegisterSession (0x65) opens a session; SendRRData (0x6F) and SendUnitData (0x70) carry unconnected and connected CIP traffic.
- Connection establishment: the CIP Forward Open service sets up connections; Reset on the Identity object restarts a device. Both are ordinary services with no built-in authorisation in classic CIP.
| Port | Transport | IANA service | Use | Security note |
|---|---|---|---|---|
| 44818 | TCP and UDP | EtherNet/IP messaging | Explicit messaging, ListIdentity discovery, engineering and HMI traffic | Unauthenticated in classic CIP; the main attack surface |
| 2222 | UDP (TCP registered) | EtherNet/IP I/O | Class 0/1 implicit I/O, often multicast | Spoofable cyclic data; sensitive to latency from inspection devices |
| 2221 | TCP (TLS) and UDP (DTLS) | ethernet-ip-s | EtherNet/IP over TLS and DTLS with CIP Security | Only present where CIP Security is enabled; disabling it on some products disables automatic policy deployment |
| 80 / 443 | TCP | HTTP / HTTPS | Embedded web servers on adapters, communication modules and drives | Often enabled by default; CIP Security policy tools can disable HTTP per zone |
The consequence is simple: on a classic EtherNet/IP network, anything that can reach port 44818 on a controller can enumerate it, and anything that can open a session can attempt the same services the engineering software uses. MITRE ATT&CK for ICS describes the resulting behaviours as Remote System Information Discovery (T0888), Program Download (T0843), Program Upload (T0845), Change Operating Mode (T0858), Device Restart/Shutdown (T0816) and Adversary-in-the-Middle (T0830).
What CIP Security adds: profiles and properties
ODVA defines CIP Security as a device-level extension that supports a defence-in-depth architecture with three goals: integrity, authenticity and authorisation. It is delivered through profiles that a device declares and conformance testing verifies.
| Profile | What it provides | Mechanism | Typical use |
|---|---|---|---|
| EtherNet/IP Confidentiality Profile | Device authentication, message integrity, optional encryption between endpoints | TLS (RFC 5246) for TCP traffic including encapsulation, UCMM and Class 3; DTLS (RFC 6347) for UDP Class 0/1 I/O; X.509 certificates or pre-shared keys | Controllers, communication modules, engineering and HMI clients |
| CIP User Authentication Profile | User-level authentication and basic role-based authorisation for CIP sessions | OpenID Connect with an external identity provider, or on-device user management for small systems | Restricting who, not just which machine, can change a device |
| Resource-Constrained CIP Security Profile | A lighter version of the protections for devices with limited CPU and memory | Reduced protocol footprint | Simple I/O devices, sensors, small drives |
| Pull Model Profile | Automated, device-initiated security configuration | Device retrieves its configuration from a policy source | Large fleets where pushing configuration per device does not scale |
- Integrity: TLS and DTLS record protection (HMAC) means a tampered or injected packet is rejected. This is the property that defeats spoofed Forward Open, write and mode-change requests from an unauthenticated host.
- Authenticity: each endpoint proves its identity with a certificate chained to a trusted authority, or with a shared key. Devices can enrol their own certificates to simplify commissioning.
- Confidentiality: optional encryption using the cipher negotiated in the handshake. Use it for engineering sessions (program logic and credentials cross the wire) and for data with commercial or safety sensitivity; integrity-only is often sufficient for cyclic I/O.
- Trust domains: ODVA describes trust that can be broad (a group of devices) or narrow (per user and role). In practice this maps onto IEC 62443 zones: devices inside a zone trust each other, and explicit conduits define trust between zones.
CIP Security does not fix weaknesses inside the device, such as a parser vulnerability in the CIP stack reachable before authentication, nor does it stop an authenticated but compromised engineering workstation from downloading malicious logic. It narrows who can talk; it does not judge what they say.
Deploying CIP Security with zones and conduits
Vendor tooling implements CIP Security policy as a security model of zones, conduits and devices. Rockwell Automation's application technique documents a typical implementation: FactoryTalk Policy Manager designs and deploys the model, FactoryTalk System Services runs the policy deployment and a Certificate Service that acts as the certificate authority for device certificates, and FactoryTalk Linx acts as the secure EtherNet/IP client for engineering and HMI software (RSLinx Classic cannot be used for CIP Security).
Policy choices
| Setting | Options | Guidance |
|---|---|---|
| Device authentication | X.509 certificate or pre-shared key (PSK) | Prefer certificates: stronger, per-device and revocable. PSK has lower connection set-up cost, a device can hold only one, and conduits from a PSK zone to other zones must use Trusted IP |
| I/O data security | None, Integrity Only, Integrity and Confidentiality | Integrity Only for most cyclic I/O; test RPI headroom before adding encryption |
| Messaging security | Integrity Only, Integrity and Confidentiality | Integrity and Confidentiality for engineering and configuration traffic |
| Conduit trust | Certificate based or Trusted IP (allow list) | Trusted IP only for legacy devices and clients that cannot run a secure client; treat it as a compensating control, not authentication |
| Zone option | Disable HTTP ports | Disable embedded HTTP where not used for operations |
Phased rollout
- Inventory and classify: list every EtherNet/IP device with vendor, catalogue number and firmware (passive monitoring of ListIdentity responses gives most of this), and mark which support CIP Security and which profile. Identify every client that talks CIP: engineering workstations, HMIs, historians, OPC servers, asset tools.
- Design zones and conduits: group devices by function and consequence (for example one zone per production cell, safety controllers separate), and define conduits only for the flows the inventory shows are needed. See the OT Network Segmentation, Purdue Model and IEC 62443 Zones and Conduits Guide.
- Stand up the trust infrastructure: install the policy and certificate services on a hardened, backed-up server inside the OT network, restrict administrative access to it, and document certificate validity periods and renewal.
- Update firmware and clients: bring CIP Security capable devices to firmware that supports the chosen profile, and move engineering and HMI clients to the secure communication driver.
- Pilot on a test cell or lab rig: deploy integrity-only first, measure connection set-up times, I/O jitter and controller communication load, then add confidentiality where planned.
- Deploy in a maintenance window: deployment resets connections. Rockwell recommends deploying only changed device communication ports and resetting existing connections; the "maintain existing connections" option can leave a system partly secured and cause outages later.
- Close the old paths: once secure communication is verified, disable unsecured explicit messaging where the product allows, disable unused HTTP, and tighten firewall rules so 44818 is reachable only from conduits you intended.
- Plan rollback: keep the pre-deployment policy export and controller project backups. Note product-specific traps: on some devices disabling CIP Security (port 2221) also disables automatic policy deployment and re-enabling requires a factory reset, which can clear a controller's application program. Changing a device IP address requires redeploying the model.
Protecting devices that will never support CIP Security
Controllers and adapters stay in service for 15 to 25 years, so most plants will run mixed networks for a long time. Layer these controls around legacy devices.
- Security proxies: a CIP Security proxy (for example Rockwell's 1783-CSP) terminates secure connections on behalf of non-capable devices placed behind it, so the device joins the security model even though it cannot do TLS itself.
- CIP-aware firewalls: industrial firewalls with deep packet inspection can allow read services from HMIs and historians while blocking write, program download, mode change and reset services from anything except the engineering conduit. A port-only rule on 44818 cannot make that distinction.
- Controller-side protections: keep the physical mode switch in RUN where the platform has one, so remote program changes are refused until someone is at the cabinet; use vendor features such as source protection, trusted slot or communication module restrictions, and disable firmware updates over the network where possible.
- Discovery suppression: block ListIdentity broadcasts and UDP 44818 at zone boundaries; there is rarely a legitimate reason for a device in one cell to enumerate another cell.
- No internet exposure: EtherNet/IP devices reachable from the internet are routinely indexed by search engines. Remote engineering belongs behind a brokered remote access service, never a port forward. See the OT Remote Access Guide.
The importance of these layers shows in CVE-2021-22681: Studio 5000 Logix Designer and RSLogix 5000 used a key to verify that Logix controllers were communicating with genuine software, and an unauthenticated attacker could bypass that verification and authenticate to ControlLogix, CompactLogix, GuardLogix and other controllers (CVSS 9.8). CISA added it to the Known Exploited Vulnerabilities Catalog in March 2026. Where the controller-side check fails, the network, the mode switch and CIP Security are what remain.
Where identity fits
CIP Security turns every participating device and engineering client into a holder of a cryptographic identity, which brings identity lifecycle problems into the control network.
- Device certificates: issued by the policy server's certificate authority. Record issuer, validity and renewal process; an expired device certificate is an availability incident. Revoke and reissue when a device is replaced or returned for repair.
- The certificate authority and policy server: whoever controls them can enrol a rogue client into a trusted zone. Treat them as Tier 0 for OT: dedicated, hardened, backed up offline and administered only from privileged access workstations. See the Machine Identity, PKI and Certificate Lifecycle Guide.
- Pre-shared keys: shared secrets across a zone. Store them in a vault, limit who can read them and rotate them when staff or integrators leave.
- User authentication: the CIP User Authentication Profile uses OpenID Connect, so the identity provider and its accounts become part of control system access. Keep OT accounts separate from corporate accounts and avoid shared engineering logins. See the OT and ICS Identity and Access Guide.
- Engineering clients: a workstation with a trusted secure client is now an authenticated path to controllers, which makes its compromise more valuable, not less. See the Engineering Workstation Security Guide.
Detection and monitoring
EtherNet/IP and CIP are well understood by passive OT monitoring tools and by open-source analysers such as Zeek and Wireshark, which decode the encapsulation and CIP service layers. Monitor from a SPAN or TAP at zone boundaries and on the engineering conduit. See the OT Network Security Monitoring Guide.
| What to watch | Why it matters | ATT&CK for ICS |
|---|---|---|
| ListIdentity requests or broadcasts from hosts that are not asset tools | Reconnaissance of vendor, model and firmware | T0888 Remote System Information Discovery; T0846.002 Broadcast Discovery |
| New source opening sessions to a controller on 44818 | Unexpected client, possibly a compromised host or rogue laptop | T0886 Remote Services |
| Program download, online edit, upload or firmware update services outside change windows | Logic manipulation or theft | T0843 Program Download; T0845 Program Upload |
| Controller mode change and Identity object Reset requests | Stopping or restarting the process | T0858 Change Operating Mode; T0816 Device Restart/Shutdown |
| Write services to tags from hosts that normally only read | Unauthorised command or parameter change | T1692.001 Unauthorized Message: Command Message; T0836 Modify Parameter |
| Cleartext 44818 traffic between devices in a CIP Security zone | Policy not deployed, downgraded or bypassed | T0830 Adversary-in-the-Middle |
| TLS handshake failures and certificate errors on 2221 | Expiring certificates, misconfiguration or impersonation attempts | None (operational signal) |
Metrics that show progress
- Coverage: percentage of CIP Security capable devices with policy deployed, and percentage of all EtherNet/IP devices behind a proxy or CIP-aware firewall.
- Exposure: number of hosts able to reach 44818 on safety and critical controllers (target: only the defined engineering and HMI conduits).
- Change hygiene: program downloads and mode changes matched to an approved change record, target 100%.
- Certificate health: device certificates expiring in the next 60 days, and time to renew.
- Discovery noise: ListIdentity requests from unapproved sources per week, trending to zero.
Decision guidance
- Greenfield line on current hardware: specify CIP Security conformance in procurement, use certificates rather than PSK, integrity for all traffic and confidentiality for engineering traffic, and design zones before commissioning.
- Brownfield plant with mostly legacy devices: start with segmentation and a CIP-aware firewall on the engineering conduit, mode switches in RUN and passive monitoring; add proxies for the highest consequence legacy devices; enable CIP Security on capable devices as they are upgraded.
- Small site with one or two controllers: the policy infrastructure may be more than the site can maintain. A locked-down engineering workstation, a simple firewall allowing 44818 only from it and the HMI, mode switches in RUN and offline project backups give most of the benefit.
- Large multi-site estate: standardise the security model per site type, centralise certificate lifecycle reporting, and consider the Pull Model profile and a central identity provider for user authentication where products support them.
- Integrity only or encryption: choose encryption for engineering sessions, recipe or intellectual property data and anything crossing a conduit between zones of different trust; integrity only for high-rate I/O where timing margins are tight.
Framework mapping
| Control area | NIST SP 800-53 Rev. 5 | NIST CSF 2.0 | Other |
|---|---|---|---|
| Zones, conduits and CIP-aware filtering | SC-7 Boundary Protection; AC-3 Access Enforcement | PR.IR-01 | IEC 62443-3-2 zones and conduits; ISO/IEC 27001:2022 A.8.22 Segregation of networks; D3FEND D3-NI Network Isolation |
| TLS and DTLS integrity and confidentiality | SC-8 Transmission Confidentiality and Integrity; SC-23 Session Authenticity | PR.DS-02 | ISO/IEC 27001:2022 A.8.24 Use of cryptography; D3FEND D3-MAN Message Authentication |
| Device certificates and PSK | IA-3 Device Identification and Authentication; IA-5 Authenticator Management; SC-17 Public Key Infrastructure Certificates | PR.AA-01; PR.AA-03 | CIS Controls v8.1 Control 12 Network Infrastructure Management |
| Disabling unused services such as HTTP and discovery | CM-7 Least Functionality | PR.PS-01 | CIS Controls v8.1 Control 4 Secure Configuration |
| Monitoring CIP services | SI-4 System Monitoring; AU-2 Event Logging | DE.CM-01 | CIS Controls v8.1 Control 13 Network Monitoring and Defense |
| Asset inventory from ListIdentity | CM-8 System Component Inventory | ID.AM-01 | NIST SP 800-82 Rev. 3 asset management guidance |
Common failure patterns
- Securing the controller but not the path: CIP Security deployed on new controllers while engineering software still connects through an unsecured driver or a Trusted IP conduit open to a whole subnet. Check what the client actually uses.
- Trusted IP everywhere: allow lists used as the default conduit type because certificates felt harder. Any host that can take an allowed IP address inherits the trust; Rockwell's own documentation warns that legacy devices enable hijacking and IP spoofing.
- Partial deployment left in place: choosing to maintain existing connections during deployment, then never resetting them, so the plant runs half old policy and half new until an unplanned restart causes an outage.
- Unmanaged certificate lifecycle: no record of device certificate expiry, a policy server with no backup, and a certificate authority administered from a general-purpose workstation. The first symptom is a line that will not reconnect after a power cycle.
- Flat EtherNet/IP networks bridged to IT: a single layer 2 network where office laptops, vendor laptops and controllers share a broadcast domain, so ListIdentity enumerates the plant from anywhere. Segmentation failures of this kind are how IT incidents become operational ones, as in the Colonial Pipeline Ransomware Attack 2021, where ransomware confirmed only in IT led the operator to shut down the pipeline as a precaution.
- Mode switches left in REMOTE: convenient for engineers, but it removes the one control that refuses a remote program download regardless of network path.
Practitioner checklist
- Build an inventory of every EtherNet/IP device and CIP client, with firmware and CIP Security capability, from passive monitoring.
- Block 44818, 2222 and 2221 at the IT and OT boundary and confirm no EtherNet/IP device is reachable from the internet.
- Define zones and conduits and restrict 44818 to the engineering and HMI conduits that need it.
- Place a CIP-aware firewall on the engineering conduit and block write, download, mode change and reset services from other sources.
- Deploy CIP Security on capable devices with certificates, integrity for I/O and confidentiality for engineering traffic, piloted on a test cell first.
- Harden, back up and restrict administration of the policy server and certificate authority; track device certificate expiry.
- Keep controller mode switches in RUN and use source protection and firmware update restrictions where available.
- Monitor ListIdentity, new sessions, program downloads, mode changes and cleartext CIP inside secured zones, and match downloads to change records.
- Keep offline, tested backups of controller projects and the security model so rollback does not depend on the network.
Standards and references
- ODVA: CIP Security
- ODVA: EtherNet/IP
- Rockwell Automation: CIP Security with Rockwell Automation Products Application Technique
- IANA: Service Name and Transport Protocol Port Number Registry
- NIST: SP 800-82 Rev. 3 Guide to Operational Technology (OT) Security
- NIST NVD: CVE-2021-22681
- MITRE ATT&CK for ICS: Program Download
- CISA: Known Exploited Vulnerabilities Catalog
Related NHI Mgmt Group resources: PLC Security Guide · Rockwell ControlLogix and FactoryTalk Hardening Guide · IEC 62443 Guide · OT and ICS Identity and Access Guide · Modbus TCP and Modbus Security Guide