Microsoft Sentinel for MSSPs: Multi-Tenant Setup Guide
Skip to content

Microsoft Sentinel for MSSPs

Microsoft Sentinel gives MSSPs a way to deliver managed security across multiple client environments, combining Microsoft security signals with logs from the wider technology estate.

How Microsoft Sentinel Works for MSSPs

The usual architecture keeps a Sentinel workspace in each client’s Microsoft Entra tenant, with centralized management across those environments.

Microsoft Sentinel supports an MSSP model through separate client workspaces, delegated Azure management, and multitenant operations in the Microsoft Defender portal. Central management does not require collecting every client’s raw logs into a shared provider workspace.

Each component has a different job:

ComponentRole in the service
Microsoft SentinelCollects and analyzes security data, runs analytics rules, and supports hunting and automation.
Microsoft Defender XDRCorrelates supported endpoint, identity, email, and application signals, with investigation and response capabilities.
Azure LighthouseDelegates supported Azure resource operations across client tenants.
Defender multitenant managementBrings supported incidents, alerts, cases, and hunting across authorized tenants into a shared operational view.

The MSSP builds the service around these capabilities: configuring sources, maintaining detections, investigating activity, coordinating response, and reporting the results.

For example, a suspicious mailbox rule may need evidence from Entra sign-ins, Microsoft 365 audit events, and the user’s endpoint. The client also needs someone to assess the evidence and take the agreed action. A central incident queue helps organize that work; the operating model determines how it gets done.

Choose the Right Multi-Tenant Microsoft Sentinel Architecture

For most MSSPs, start with at least one Microsoft Sentinel workspace in each client’s Entra tenant. Microsoft recommends this approach because important native service-to-service connectors operate within their own tenant boundary.

Deployment modelWorkspace locationBillingBest fit
Client-owned Azure subscriptionClient tenant and subscriptionClient buys Azure directlyClients retaining their Azure resources, configuration, and security history.
Azure supplied through CSPClient tenant, with Azure purchased through the CSP channelPartner or distributor bills through the commercial arrangementProviders combining Azure purchasing with managed security.
Provider-hosted workspaceProvider tenantProvider pays Azure and includes or recovers consumption costsSpecific collection scenarios that have been validated for cross-tenant support.

A CSP billing relationship does not replace delegated access. Decide separately who pays for Azure and who can manage the resources.

Provider-hosted collection needs particular care. Microsoft identifies Defender XDR as an example of a source that cannot connect across tenants in this model. That makes a single provider workspace unsuitable for many Microsoft-focused MSSP services.

Create additional workspaces where region, tenant boundaries, or operational separation require them. Different retention periods alone may be handled through table-level settings.

Record workspace IDs, subscription IDs, billing owners, deployed content versions, and provider-hosted dependencies in a client service inventory. Keeping this information current makes troubleshooting, access reviews, and eventual handover much easier.

Azure Lighthouse vs GDAP vs B2B

Azure Lighthouse, GDAP, and Entra B2B solve different access problems. Choose the access path for the task an analyst or engineer must perform.

Access methodMain purposeWhat to establish for Sentinel operations
Azure LighthouseDelegated management of supported Azure resourcesScope, supported Azure roles, workspace operations, and cross-workspace queries.
GDAPSupported tenant administration and Defender workloads through a CSP relationshipWorkload and role support. GDAP alone does not provide Sentinel data access in the documented Defender multitenant setup.
Entra B2BExternal identities with assigned access in the client tenantClient-side roles, Conditional Access behavior, and workload-specific response rights.
Tenant governance relationshipsDelegated access through provider groups represented as remote tenant groupsClient approval, licensing, role assignments, and support for each required workflow.

Microsoft’s multitenant setup guidance distinguishes GDAP access to Defender data from access to Sentinel data. Its newer governance documentation also describes assigning remote tenant groups Sentinel permissions. Validate the chosen path against the actual portal, API, and response operations your service uses.

Azure Lighthouse and Microsoft 365 Lighthouse are separate products. A working Microsoft 365 administration relationship does not establish all the Azure and Sentinel permissions your SOC needs.

Configure Lighthouse Around Operational Roles

Use security groups for analysts, hunters, and engineers, with delegation at the subscription or resource-group scope the service requires. Lighthouse supports specific built-in roles; it does not support Owner, custom roles, or roles containing DataActions. Its User Access Administrator exception is limited to specified managed-identity role assignments.

A practical starting pattern is:

FunctionStarting Azure roleAdditional considerations
Incident analystsMicrosoft Sentinel ResponderSeparate permissions for actions in connected products.
Threat huntersMicrosoft Sentinel ReaderAccess to other evidence sources where required.
Detection engineersMicrosoft Sentinel ContributorControlled release process and privileged access management.
Playbook operatorsMicrosoft Sentinel Playbook OperatorPlaybook identity and downstream permissions.
Deployment and investigation servicesPermissions for their specific functionSeparate identities for deployment, investigation, and containment.

Reader, Responder, and Contributor provide progressively broader Sentinel permissions. Incident management rights do not automatically authorize endpoint isolation or directory changes.

Use eligible Lighthouse authorizations for temporary privileged work where appropriate. They require applicable PIM licensing and cannot be activated by service principals, so unattended deployments need a separately controlled permission model.

Protect the provider tenant accordingly. Client Conditional Access policies do not apply to users accessing resources through Lighthouse; the managing tenant’s policies apply to that access path.

Manage Microsoft Sentinel Clients in the Defender Portal

Build new operating procedures around the Defender portal. Microsoft states that Sentinel will no longer be supported in the Azure portal after March 31, 2027.

Defender multitenant management provides a combined view across authorized client tenants. Microsoft currently documents a limit of 100 target tenants, so larger providers need an operating design that accounts for that boundary.

Where a client has multiple workspaces, distinguish their roles:

Primary workspace

When integrated with Defender XDR, receives synchronized Defender XDR incidents and alerts and participates in the combined correlation experience.

Secondary workspaces

Keep separate incident correlation. They can ingest supported Defender table data when configured, but do not inherit primary-workspace incident synchronization.

Document the primary workspace and test any change to it. Detection rules, playbooks, and ticket integrations may depend on the previous configuration.

Before accepting a client into the shared view, confirm that analysts can open its environment directly and complete the required tasks. Test MFA, role assignments, and permitted response actions. A client missing from the dashboard should trigger an access or health check.

Build Detection Coverage Around Client Risk

Start with what the service must detect and investigate. Then identify the sources, rules, and response capabilities needed to support it.

A useful coverage matrix connects attack scenarios to evidence and operational tests:

Security scenarioEvidence to considerExample validation
Account compromiseSign-ins, directory changes, session activity, identity alertsCan the analyst explain the sign-in and perform an approved session revocation?
Business email compromiseMailbox rules, email telemetry, sign-ins, application consentCan the investigation connect an account event to suspicious forwarding?
Endpoint intrusionEDR alerts, process activity, relevant operating-system logsCan the team reconstruct execution and isolate a test device?
Cloud abuseControl-plane audit, workload activity, access changesCan the team identify who changed a resource and assess the impact?
Lateral movementEndpoint, authentication, firewall, DNS, and network telemetryIs there enough evidence to follow movement between systems?

Treat this as a planning framework and adapt it to each client’s estate and licenses.

For every source, record which accounts, devices, systems, and event categories it covers. Define expected latency and what should happen if collection stops. A connector showing recent activity does not establish that the whole environment is represented.

Use a shared detection baseline with client-specific settings. An education customer, a law firm, and an industrial operator may share core identity detections while needing different thresholds, exclusions, critical assets, and response approvals.

Track gaps alongside incidents. Missing audit events, failing rules, and unavailable response permissions belong in the service improvement plan, with an owner and a next action.

Deploy Microsoft Sentinel Content as Code

Manual rule changes become difficult to control as the client base grows. Content as code gives the MSSP a reviewed source for shared detections and a repeatable way to release them.

Keep common analytics rules, parsers, workbooks, and automation separate from client configuration. Parameterize thresholds, enabled sources, exclusions, and deployment scope. Record dependencies so a rule is deployed only where the required tables and supporting content exist.

Select a Deployment Method

MethodUseful forMain consideration
Sentinel repositoriesGitHub or Azure DevOps deployment of supported contentCreating the native connection requires a home-tenant account and Owner access to the workspace resource group; B2B guests and delegated access are not supported for setup.
Defender distribution profilesCopying supported content across target tenantsLicensing, content support, and target permissions must match the deployment.
Provider-operated CI/CDCustom release logic and client-specific deploymentThe provider maintains authentication, scripts, testing, and failure handling.

Distribution profiles support several Sentinel content types, but automation rules that trigger playbooks cannot be distributed through that feature. Check outcomes by client and item when a publish is partially successful.

Release and Verify Changes

Validate templates and queries, test against representative data, and release to a small client group before expanding. Check whether the deployed rule runs successfully and produces useful investigations.

Record each client’s intended version, deployed version, exceptions, and failures. Use stable resource identifiers so an update changes the existing item.

Capture emergency portal edits back into source control before the next deployment. Otherwise, a later release can overwrite them. Handle retirement explicitly: deleting a repository file does not delete its deployed Sentinel resource.

Hunt Across Client Workspaces

Cross-client hunting helps answer whether a newly identified indicator or behavior affects other managed environments.

Two query patterns matter:

Defender multitenant hunting

Runs a query independently in selected tenants and combines the results.

Explicit cross-workspace KQL

Uses workspace() references to query named workspaces, including supported cross-tenant access through Lighthouse.

For example, this query summarizes non-success sign-ins in two client workspaces:

union (workspace("<acme-corp-workspace-id>").SigninLogs | where TimeGenerated >= ago(24h) | extend Client = "acme-corp"), (workspace("<acme-health-workspace-id>").SigninLogs | where TimeGenerated >= ago(24h) | extend Client = "acme-health") | where ResultType != "0" | summarize NonSuccessSignIns = count(), ApproxAccounts = dcount(UserPrincipalName) by Client, IPAddress | order by NonSuccessSignIns desc 

Replace the placeholders with workspace identifiers and run with an authorized identity. The example requires SigninLogs in both workspaces. Non-success results need interpretation; they are not automatically attacks.

Keep the client identity attached to every result. Similar private IP addresses, usernames, or hostnames across customers do not establish a shared incident.

Scheduled cross-workspace analytics rules have different constraints from interactive hunting: they support up to 20 workspaces per query, with five or fewer recommended for performance. Sentinel must be enabled on each referenced workspace, and alerts are created only in the rule’s home workspace.

For a common detection that should operate independently for each customer, deploying the rule into each eligible client workspace is usually easier to govern.

Track Microsoft Sentinel Costs and Service Margin per Client

An MSSP needs to understand both Azure consumption and the work required to deliver the service.

Track costs at the agreed workspace, resource, subscription, or billing scope. Separate Microsoft charges from SOC fees, engineering work, and shared platform allocations.

Cost areaWhat to track
Analytics ingestionBillable volume, major contributing tables, duplicate collection, and growth.
Retention and data lakeRetained data, storage, query activity, and the capabilities used.
Automation and additional servicesLogic Apps and other separately charged components.
Pricing arrangementsCommitment scope, applicable benefits, negotiated rates, and actual billed charges.
Service deliveryInvestigation effort, tuning, response coordination, reporting, and specialist support.

Sentinel’s analytics tier offers pay-as-you-go and commitment pricing. Other capabilities have their own meters, so ingestion volume alone does not describe the full bill.

Use the Usage table to investigate consumption trends and Azure Cost Management and billing records to reconcile charges. A cross-workspace Usage query can help identify which client and table contributed to growth, but its output is not an invoice.

Review spending alongside coverage. An increase may come from a newly monitored system, a connector error, repeated events, or duplicate collection. Decide what changed before filtering data.

Lighthouse delegation does not automatically pool unrelated clients into one commitment tier. Size commitments against their actual billing scope, and account for the 31-day restriction on reducing a commitment tier.

Choose a Clear Commercial Model

Common arrangements include a service fee with Azure billed separately, a bundled CSP service, or per-user or per-device pricing with defined consumption assumptions.

Whichever model you use, specify covered technologies, monitoring hours, investigation depth, response actions, and responsibility for material cost changes. Measure analyst effort as well as gigabytes: a small, complex client can require substantial operational work.

Assign a named owner to spend alerts and cost reviews. For detailed filtering, routing, retention, and pricing decisions, see Microsoft Sentinel Cost Optimization.

Onboard Microsoft Sentinel Clients With a Repeatable Runbook

Start with a pilot client and prove the complete service before making the deployment pattern standard.

1. Define the Service

Agree the tenants, systems, sources, coverage hours, retention, response authority, and escalation contacts. Identify who approves source consent, spending changes, and containment.

Record the security scenarios the service must cover and any known gaps.

2. Prepare the Client Workspace

Confirm tenant, subscription, region, billing arrangement, and required workspaces. Establish the intended primary workspace where Defender XDR integration applies.

Verify Microsoft.SecurityInsights and Microsoft.OperationalInsights registration in the client subscriptions and at least one subscription in the provider tenant, as required by Microsoft’s MSSP guidance.

3. Configure Delegated and Workload Access

Deploy Lighthouse at the agreed scope and configure the supported Defender and directory access paths.

Test the production analyst and service identities. Connector setup can require additional tenant-side permissions or consent; Lighthouse alone is not sufficient for every connector.

4. Connect Sources and Validate Coverage

Confirm actual events, entity identifiers, latency, and source scope. Compare the observed data with the agreed coverage matrix.

Test collection failures as well as successful collection so the team knows how a silent source will be detected and assigned.

5. Deploy the Baseline

Apply approved content and client parameters. Validate dependencies, rule health, alert quality, playbook identities, and downstream actions.

Record the deployed version and any exceptions.

6. Connect the Service Workflow

Map incidents to the correct ticket board, escalation contacts, client portal, and reporting schedule.

Use tenant, workspace, and incident identifiers together when synchronizing records. Display names and incident numbers alone are insufficient across clients.

7. Test Detection Through Response

Generate an agreed event on a test asset. Follow it through detection, investigation, ticket synchronization, escalation, and an authorized response.

Verify tenant and asset identity at every step.

8. Accept and Review

Confirm what is covered, which gaps remain, and who owns them. Capture ingestion, cost, and alert baselines.

Schedule an early service review to assess noise, investigation effort, source health, and actual costs. Use what you learn to improve the shared onboarding pattern.

Deliver 24x7 Monitoring and Response

A multitenant console reduces navigation between environments. Delivering 24x7 coverage still requires qualified people, working response paths, and reliable handovers.

Operating modelResponsibility
In-house SOCThe MSSP supplies every covered shift, engineering, escalation, and specialist support.
After-hours partnerThe MSSP handles agreed business hours and a partner covers the remaining period.
Co-managed SOCResponsibilities are shared by hours, investigation level, or function.
White-label SOCA partner supports delivery under the MSSP’s brand while the MSSP retains the client relationship.

Define monitoring, investigation, approval, and containment availability separately. State what happens when the client’s primary contact cannot be reached and who can authorize urgent action.

AI can assist with evidence gathering, triage, and investigation. Introduce it gradually: begin with investigation support, review findings, and authorize specific actions as the workflow proves reliable. Keep client context, access, evidence, and approvals attached to each case.

For implementation detail, see Agentic SOC on Microsoft Sentinel. For service options, see Managed SOC for MSPs and MSSPs.

Coverage by clientlocal time
acme-corpFully managed
acme-healthAfter hours
acme-legalOverflow
acme-retailCo-managed
acme-logisticsAfter hours
acme-financeOverflow
0006121824
ThreatDefence SOCyour teamsame cases, both teamsSOC on escalation

Offboard Clients With a Complete Handover

Client-owned workspaces make continuity easier, but the service may still depend on provider credentials, central rules, playbooks, repositories, or ticket integrations.

Use the service inventory to plan the transition:

  1. Hand over open incidents, configuration, rule inventories, unresolved gaps, and operational documentation.
  2. Confirm which detection content remains licensed and replace essential provider-hosted dependencies.
  3. Remove the client from deployment targets, scheduled hunts, reports, and ticket jobs.
  4. Transfer automation to approved identities, then revoke outgoing application grants, guest access, direct roles, and delegated relationships.
  5. Resolve CSP billing and commitments separately from technical access.
  6. Verify that the incoming service works and the outgoing provider’s identities are denied access.

Coordinate access removal with validation of cross-tenant rules. Their continued operation can depend on the creating identity’s permissions in other subscriptions or tenants.

Agree retention and disposal for provider-held case records and exports. Removing delegation does not remove copies already created outside the workspace.

Microsoft Sentinel for MSSPs With ThreatDefence

ThreatDefence helps MSPs and MSSPs build managed security services around Microsoft Sentinel and the wider client environment.

Keep client Sentinel workspaces in place and add an AI SOC platform with 24x7 SOC support. AI agents gather evidence and investigate activity across connected Microsoft and non-Microsoft sources, while analysts validate findings and coordinate response within each client’s approved limits.

Your team can work with ThreatDefence through shared cases, defined escalation paths, and client-specific procedures. White-label options bring the portal, reports, and notifications under your brand, with ticket synchronization for ConnectWise, Autotask, HaloPSA, or ServiceNow.

Extend coverage where the client’s security program needs it, including attack surface management, brand protection, and network detection. Choose fully managed, after-hours, or co-managed support according to your delivery model.

Start with a Microsoft Sentinel cost and coverage review to identify missing evidence, detection gaps, and unnecessary consumption. The review provides a practical basis for improving the service before expanding it across more clients.

MSP consoleall tenants
Tenants20
Open alerts31
Ingest, 30 days2.3 TB
Silent sources1
Portal brand
ThreatDefenceNorthgate SOCHarbor Cyber
portal.yourbrand.com
TenantVolume, 30 daysAlertsReport
acme-corp
742 GB12Oct 3
northwind
480 GB4Oct 3
harbor-health
391 GB9Oct 4
bluegum-legal
238 GB1Oct 3
summit-retail
196 GB3Oct 5
kestrel-mining
152 GB2Oct 4
lumen-dental
128 GB–Oct 5
ironbark-civil
104 GB1Oct 4
others (12)
98 GB–Oct 3
Search all tenantsevent.action:"New-InboxRule" AND forward*3 tenants, 7 hits

Request a Demo

See how ThreatDefence supports Microsoft Sentinel operations across multiple clients, from AI investigation and shared cases to branded reporting and 24x7 SOC delivery.

    I'm an MSP, MSSP, or reseller

    By submitting this form, you accept the terms in our privacy policy.

    Frequently Asked Questions

    Is Microsoft Sentinel suitable for MSSPs?

    Yes. Sentinel supports client workspaces across tenants, delegated Azure management, cross-workspace queries, and operations through Defender multitenant management. The MSSP still needs a repeatable approach to detection content, access, response, cost reporting, and service delivery.

    Should each customer have a separate Microsoft Sentinel workspace?

    Usually, at least one workspace per client Entra tenant. This supports native connector boundaries and separate configuration. Some clients need additional workspaces for geography or operational requirements. A provider-hosted workspace requires source-by-source validation.

    Does Azure Lighthouse replace B2B or GDAP?

    No. Lighthouse delegates supported Azure resource operations. B2B, GDAP, and tenant governance relationships address other access requirements. Your service may need more than one method to support Azure management, Defender investigations, connector consent, and response.

    Can GDAP provide Microsoft Sentinel access?

    Microsoft’s Defender multitenant setup guidance states that GDAP does not provide Sentinel data access. It supports eligible Defender workloads and administration. Microsoft separately documents tenant governance relationships for Sentinel permissions. Validate the access method for each operation.

    Do clients need to replace their existing EDR?

    Not necessarily. Sentinel can collect data from supported Microsoft and third-party sources. Whether the SOC can investigate and respond through the existing EDR depends on connector support, available telemetry, licensing, API permissions, and the service design.

    Is Microsoft Sentinel the same as Defender XDR?

    No. Defender XDR correlates supported security signals across Microsoft workloads. Sentinel adds SIEM capabilities and broader log collection and analysis. They can work together in the Defender portal, with integration behavior determined partly by the primary workspace.

    Does moving to the Defender portal merge client workspaces?

    No. The portal brings operational views together while Sentinel workspaces remain distinct. Primary and secondary workspaces retain different correlation and synchronization behavior. Treat portal onboarding as a configuration change and test affected integrations.

    Can MSSPs hunt across all their clients?

    Across clients they are authorized to query. Defender multitenant hunting and Lighthouse-based workspace queries use different execution patterns. Defender multitenant advanced hunting has a 50,000-record total limit divided across selected tenants, so check completion and result limits before drawing conclusions.

    Can Microsoft Sentinel be delivered as a white-label service?

    Yes, through an MSSP service model and supporting platform. Define which portal, reports, notifications, and workflows carry your brand, and how client Sentinel workspaces connect to the service. Azure delegation alone does not create a branded client experience.

    Who pays for Microsoft Sentinel?

    The client can pay Azure directly, or the provider can supply Azure through a CSP arrangement. The contract should state whether consumption is included in the SOC fee, billed separately, or subject to agreed allowances.

    Does client-owned Microsoft Sentinel keep all data in the client tenant?

    The workspace location establishes an important storage boundary, but central tickets, exports, reports, and external AI processing can create additional copies. Microsoft also documents processing exceptions for some Sentinel capabilities. Assess the complete data flow against the client’s requirements.

    Can clients remove an MSSP's access?

    Yes. Clients can revoke Lighthouse delegation, but a complete offboarding process must also address separate guest accounts, application grants, direct roles, and other access relationships. Replace operational dependencies before withdrawing essential monitoring or response.