You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
* mark
* doc updates
* link out from guide
* rename
* add to sidebar
* Update content/en/llm_observability/monitoring/data_access_controls.md
Co-authored-by: Barry Eom <[email protected]>
* move data access controls to its own section
* tweak
* qf
* changes
---------
Co-authored-by: cecilia saixue watt <[email protected]>
Co-authored-by: Barry Eom <[email protected]>
Copy file name to clipboardExpand all lines: content/en/account_management/rbac/data_access.md
+13-11Lines changed: 13 additions & 11 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -43,10 +43,10 @@ Log in as a user assigned the Datadog Admin role, or any user with a role in you
43
43
44
44
In order to create a Restricted Dataset, identify the data to be restricted with a query.
45
45
46
-
{{< img src="/account_management/rbac/restricted_dataset.png" alt="Create a Restricted Dataset dialog. Selects data in RUM, APM, Logs, and Metrics matching the tag service:hr. Grants access to a Privileged access team.">}}
46
+
{{< img src="/account_management/rbac/restricted_dataset-2.png" alt="Create a Restricted Dataset dialog. Selects data in RUM, APM, Logs, and Metrics matching the tag service:hr. Grants access to a Privileged access team.">}}
47
47
48
48
Name Dataset
49
-
: A descriptive name to help users understand what data is contained in the dataset.
49
+
: A descriptive name to help users understand what data is contained in the dataset.
50
50
51
51
Select data to be included in this Dataset
52
52
: The boundary definition that describes which data to restrict to a specific set of users. Boundaries are query statements with limitations that allow an access manager to define the scope of sensitive data to be protected. The [supported telemetry types][10] are custom metrics, RUM sessions, APM traces, logs, cloud costs, error tracking issues, and CI Visibility pipelines.
@@ -61,7 +61,7 @@ After completing all the fields to define the dataset, click **Create Restricted
61
61
You may create a maximum of 100 Restricted Datasets. If you need a higher limit, reach out to Support.
62
62
63
63
### API
64
-
The Data Access Control API is under development and should be considered unstable. Future versions may be backward incompatible.
64
+
The Data Access Control API is under development and should be considered unstable. Future versions may be backward incompatible.
65
65
66
66
Terraform support will be announced after Data Access Control is generally available.
67
67
@@ -75,6 +75,7 @@ Terraform support will be announced after Data Access Control is generally avail
75
75
- Error Tracking issues
76
76
- Logs
77
77
- RUM sessions
78
+
- LLM Observability
78
79
79
80
## Usage constraints
80
81
@@ -92,6 +93,7 @@ Playlists are collections of Session Replays you can aggregate in a folder-like
92
93
### Logs
93
94
Data Access Control is separate from the existing [Logs RBAC permissions][11] feature, also known as log restriction queries. To use Data Access Control with Log Management, first request access to Data Access Control. Next, manually migrate your configuration from Log Management permissions to Data Access Control.
94
95
96
+
95
97
## Select tags for access
96
98
97
99
Each Restricted Dataset can control access to multiple types of data, such as metrics. You are free to use the same or different tags across multiple types of telemetry. Within each telemetry type, you must use a _single_ tag or attribute to define your access strategy.
@@ -101,7 +103,7 @@ If you have too many combinations of tags or attributes to fit within these cons
101
103
### Supported example
102
104
103
105
#### Restricted Dataset 1
104
-
- Telemetry Type: RUM
106
+
- Telemetry Type: RUM
105
107
- Filters: `@application.id:ABCD`
106
108
107
109
#### Restricted Dataset 2
@@ -112,23 +114,23 @@ If you have too many combinations of tags or attributes to fit within these cons
112
114
113
115
### Not supported example
114
116
115
-
#### Restricted Dataset 1:
116
-
* Telemetry type: RUM
117
+
#### Restricted Dataset 1:
118
+
* Telemetry type: RUM
117
119
* Filters: `@application.id:ABCD`
118
120
119
121
#### Restricted Dataset 2:
120
-
* Telemetry type: RUM
122
+
* Telemetry type: RUM
121
123
* Filters: `env:prod`
122
124
123
125
Restricted Dataset 1 uses `@application.id` as the tag for RUM data, so a new Restricted Dataset can't change to a different tag. Instead, consider reconfiguring Restricted Dataset 2 to use `@application.id`, or changing all of your Restricted Datasets with RUM data to use another tag.
124
126
125
127
### Not supported example
126
128
127
-
#### Restricted Dataset 1:
128
-
* Telemetry type: RUM
129
+
#### Restricted Dataset 1:
130
+
* Telemetry type: RUM
129
131
* Filters: `@application.id:ABCD`
130
132
131
-
#### Restricted Dataset 2:
133
+
#### Restricted Dataset 2:
132
134
* Telemetry type: RUM
133
135
* Filters: `@application.id:IJKL``env:prod`
134
136
@@ -150,7 +152,7 @@ For example, if you have a single application that is instrumented with Real Use
150
152
* Telemetry type: RUM
151
153
* Filters: `@application.id:<rum-app-id>`
152
154
***Grant access:**
153
-
* Teams or roles of users who can see this RUM data
155
+
* Teams or roles of users who can see this RUM data
154
156
155
157
This configuration example would protect the RUM data from this application, and keep other data from this application available to existing users in your organization.
{{< nextlink href="https://datadoghq.com/legal/hipaa-eligible-services">}}<u>HIPAA-Eligible Services</u>: Datadog Legal's list of HIPAA-eligible services{{< /nextlink >}}
10
+
{{< nextlink href="/llm_observability/evaluations/#sensitive-data-scanner-integration">}}<u>Sensitive Data Scanning for LLM Observability</u>: Redact sensitive information in your LLM application{{< /nextlink >}}
11
+
{{< /whatsnext >}}
12
+
13
+
## Data Access Control
14
+
15
+
LLM Observability allows you to restrict access to potentially sensitive data associated with your ML applications to only certain teams and roles in your organization. This is particularly important when your LLM applications process sensitive information such as personal data, proprietary business information, or confidential user interactions.
16
+
17
+
Access controls in LLM Observability are built on Datadog's [Data Access Control][1] feature, which enables enables you to regulate access to data deemed sensitive. You can use the `ml_app` tag to identify and restrict access to specific LLM applications within your organization.
0 commit comments