Vendor support should use named identities, strong authentication, an approved access broker, explicit destination and time limits, least privilege, monitored sessions, controlled file transfer, emergency revocation, and a tested local fallback. A permanent shared VPN credential cannot provide accountable access.
The short answer
Terminate remote access at a managed boundary instead of exposing controllers or SCADA directly. Verify a named person with MFA, require a ticket and time-bounded approval, permit only the needed jump host and protocol, separate observation from change authority, inspect file transfer, capture session and change evidence, and revoke access automatically. Test how the plant remains safe and supportable when the remote service is unavailable.
- Use individual identities; never share a permanent vendor account or password.
- Make access purpose-, asset-, protocol-, privilege- and time-specific.
- Keep a human plant owner responsible for approval, supervision and return-to-service.
- Design logging, revocation, incident response and local maintenance before enabling connectivity.
Decision guide
| Control point | Required decision | Acceptance evidence |
|---|---|---|
| Identity and approval | Who may connect, for which task, asset and window? | Named account, MFA, sponsor, ticket, approval and expiry |
| Path and privilege | Where does the session terminate and what can it reach or change? | Zone/conduit rule, jump host, allowlist and least privilege |
| Session and files | How are commands, uploads, downloads and changes reviewed? | Logs, recording where lawful, malware/file checks and revision link |
| Failure and response | How is access revoked and the plant supported during outage or compromise? | Kill procedure, alerts, fallback, exercise and incident evidence |
A practical workflow
- 01
Map support use cases, assets, zones, data sensitivity, vendor responsibilities and forbidden paths.
- 02
Select a controlled architecture with managed identity, MFA, approval, jump access, segmentation and monitoring.
- 03
Define request, authorization, supervised work, file exchange, change control, session close and automatic expiry.
- 04
Test allowed and denied paths, privilege, timeout, revocation, evidence, service outage and incident escalation.
- 05
Review active accounts, vendors, sessions, exceptions and controls periodically and after every significant change.
Evidence and acceptance
Retain architecture, data flows, account owner, vendor agreement, request and approval, MFA and access-policy configuration, destination allowlist, session metadata, transferred-file checksums, related revisions, actions, alerts, exceptions, revocation tests, incident exercises and periodic access reviews.
Common mistakes
- Exposing RDP, VNC, engineering ports or a PLC directly to the internet.
- Using one shared always-on vendor credential without MFA or expiry.
- Treating VPN connectivity as authorization to every reachable asset.
- Allowing unreviewed file transfer or software changes outside revision control.
- Collecting logs without ownership, alerting, retention or tested revocation.
Engineering and security boundary
Primary sources
Technical claims and product boundaries were checked against these primary sources. Fact-checked
Frequently asked questions
Is a VPN sufficient for vendor access?
No. A VPN provides transport; identity, authorization, segmentation, privilege, supervision, evidence and revocation remain necessary.
Should vendors have permanent accounts?
Prefer disabled-by-default or just-in-time named access with automatic expiry and recurring review.
Must every remote session be recorded?
Choose proportionate logging or recording based on risk, law and policy, and always retain accountable access and change evidence.
Does AutoNexit provide remote PLC access?
No. It records the approved work and evidence around a separately selected secure access system.