NIS2 for utilities: a practical connectivity checklist
NIS2 raises the bar on how you protect and prove control over connected assets. Rather than wade through legal text, use this as a field-ready to-do list. It focuses on what you can enforce in the connectivity layer and what you should align with elsewhere in the organisation.

Identity (people and devices)
What to do with connectivity
-
Bind each device to a SIM/eUICC-based identity and your asset inventory.
-
Require named user identities with MFA for engineers and contractors.
-
Grant least-privilege: one person, one device, one job, time-boxed.
Also line up
-
Your IAM/JML process (joiners/movers/leavers) so accounts appear, change and disappear on time.
Evidence to keep
-
Device↔SIM/eUICC map, decommission records, access reviews.
Segmentation and access control
What to do with connectivity
-
Replace flat site-to-site VPNs with thin, per-session paths.
-
Enforce allow-lists per device type:
-
IEC-104/DNP3 → named SCADA masters (TCP 2404/20000)
-
DLMS/COSEM → MDM FQDNs (no wildcards)
-
MQTT → broker FQDN (TLS/mTLS)
-
-
Keep OT traffic off the public internet (private APN or private routes to SCADA/MDM/cloud).
Evidence to keep
-
Current policy set, port/protocol matrix, change log.
Logging, monitoring and audit
What to do with connectivity
-
Capture session-level records: who, what, when, which policy.
-
Alert on policy breaches (unexpected protocol, host, volume).
-
Use time-boxed change windows that auto-revert.
Also line up
-
SIEM/SOC run-books so someone owns the alert and the response.
Evidence to keep
-
Logs streamed to SIEM, retention policy, MTTR metrics.
Supply chain and provisioning
What to do with connectivity
-
Factory provision identity: bind SIM/eUICC to the device before it ships; apply a baseline policy.
-
Use eUICC + multi-IMSI for resilience and local profile requirements; swap profiles with guardrails and rollback.
-
Ask vendors to support protocol/FQDN allow-lists, signed firmware, and per-session access for maintenance.
Also line up
-
Supplier security clauses and a simple acceptance checklist at goods-in.
Evidence to keep
-
Provisioning logs, profile library, vendor attestations.
Incident response and continuity
What to do with connectivity
-
Contain by default: default-deny limits lateral movement; be able to revoke device or user identity fast.
-
Health-based failover: steer to healthy networks (latency/loss/attach) and trigger profile swaps during outages.
-
Keep forensics-ready logs for the legal retention window.
Also line up
-
Your wider IR plan (roles, comms, legal hold) and tabletop exercises.
Evidence to keep
-
Revoke tests, failover drill results, incident timelines.