Skip to content

Sentinel Architecture and Workspace Design

Atlanta, USA

GitHub Cloud2BR OSS - Learning Hub

Last updated: 2026-07-27


References

Back to the documentation hub

The Log Analytics workspace is Sentinel's primary data, access, cost, and retention boundary. Choose its location and ownership before data arrives; moving an established security operations center (SOC) later is harder than designing the boundary deliberately.

Select the workspace model

Model Best for Tradeoff
Single workspace Central SOC, consistent residency, common detection content Broad access and chargeback need careful governance
Regional workspaces Residency or sovereignty requirements Cross-workspace hunting and content deployment add complexity
Subsidiary workspaces Separate organizations or managed-service boundaries Central incident operations need Lighthouse and clear ownership
Workload workspace Highly regulated or isolated telemetry Duplicates platform overhead and reduces correlation by default

Use one workspace for a defined operating boundary, not simply for every subscription. A workspace should have named owners for data, detection content, incident response, cost, and platform operations.

Design the data path

  1. Define the security questions and required data sources.
  2. Place the workspace in the approved region and subscription.
  3. Enable Sentinel, retention, archive, and access settings.
  4. Connect a pilot data source and verify the intended table, timestamp, volume, and entity fields.
  5. Add analytics, automation, and incident routing only after data quality passes.
  6. Record the workspace architecture, dependencies, and recovery procedures.

Set up the workspace

  1. In the Azure portal, create a Log Analytics workspace in the approved subscription and region, then apply owner, environment, and cost-allocation tags.
  2. Grant the platform owner and security operations center (SOC) roles through Microsoft Entra groups, using the least Sentinel and workspace permissions required for each role.
  3. Open Microsoft Sentinel, select the workspace, and enable Sentinel for that workspace.
  4. Configure the retention, archive, budget, resource lock, and diagnostic settings required by the data contract.
  5. Enable one pilot connector and query its target table before deploying analytics rules or playbooks.
  6. Record the workspace identifier, region, owners, connectors, and recovery contacts in the operating runbook.

Architecture boundaries

  • Workspace: stores collected telemetry and Sentinel content references.
  • Data collection rule: controls collection and transformation for Azure Monitor Agent sources where applicable.
  • Data connector: provisions the supported ingestion path for a product or service. A connector's enabled state is not proof that current data is arriving.
  • Analytics rule: reads tables and produces alerts; it must have a known schema, query period, entity mapping, suppression behavior, and owner.
  • Playbook: executes under a Logic App identity and should have a separate authority and rollback model from the detection that triggers it.

Validate the design

  • Confirm workspace region, resource lock, tags, budget, and service owner.
  • Confirm least-privilege access for analysts, engineers, automation identities, and break-glass administrators.
  • Verify a controlled test event reaches the expected table and preserves its timestamp, source identity, and required entity fields.
  • Measure daily data volume by table before enabling broad collection.
  • Test incident creation, assignment, closure, and evidence retention.

Operational decisions

  • Set a retention and archive policy per table based on investigation and legal needs, not an assumed universal period.
  • Document cross-workspace query owners and performance expectations before relying on them for high-severity detections.
  • Treat new table schemas and data transformations as change-managed dependencies for analytics rules and workbooks.

Business example

A retailer operates in two data-residency regions but uses one central SOC. It deploys a Sentinel workspace in each approved region, grants the SOC delegated access through Lighthouse, and standardizes analytic rules through source control. Each workspace retains local telemetry while the SOC uses approved cross-workspace queries for incident triage and monthly coverage reporting.