CTERA Content Services provides a secure data pipelines for GenAI workloads. In addition to the built-in security of the CTERA Portal, security is achieved by defining guardrails and firewall restrictions.
To add a guardrail:
After signing in, the Dashboard page is displayed.
- Select Security in the navigation pane.
- Click New Guardrails.
- Enter the guardrail details:
Name – A name to identify the guardrail.
Description – Text describingthe guardrail.
Type – The type of guardrail to be implemented.
Path – The path to the guardrail.
Guardrail Status – The current status of the rule, Enabled or Disabled.
Initialization Parameters – Add initialization parameter key and value pairs, clicking
to add each pair.
LLM Model – The LLM model the guardrail applies to.
On Fail Action – The action to take when the guardrail is triggered.
User Local Models – Slide on to apply the guardrail to local models.
To add a firewall rule:
Firewall rules can only be defined after defining one or more guardrails.
After signing in, the Dashboard page is displayed.
-
Select Security in the navigation pane.
-
Click the Firewall tab.
-
Click New Rule.
-
Enter the firewall rule details:
Name – A name to identify the rule.
Stage – When the rule is triggered:- Ingestion
- Retrieval
- Chat Pre
- Chat Post
Action – The action to take when the rule is triggered.
- Block
- Log
- Filter
Rule Status – The current status of the rule, Enabled or Disabled.
Guards – The guardrail to use from the drop-down list.
Datasets – The datasets that the rule applies to. Select each dataset from the list, one at a time.
Experts – The experts that the rule applies to. Select each expert from the list, one at a time.
Users – The users that the rule applies to. Select each user from the list, one at a time.
To edit security:
- Click Security in the navigation pane.
- Click the guardrail to edit or cllick the Firewall tab and then the firewall entry to edit.
Edit Guardrail Edit Firewall Rule 
- Make the changes as required.
- Click Save Changes.