Microsoft Sentinel Azure portal retirement: what it means for your SIEM
After March 31, 2027, Microsoft Sentinel will no longer be supported in the Azure portal and will be available only in the Microsoft Defender portal. Once your workspace runs there, Microsoft Defender XDR data privacy and retention policies apply in place of Sentinel's own, and the Fusion analytics rule is replaced by Defender XDR correlation. Microsoft states that moving to the Defender portal has no extra cost, so the change is mainly about who sets the rules for retention, privacy and correlation in your environment.
Start by auditing what your current Sentinel configuration depends on. Check your retention settings against the Defender XDR policies you would inherit, list every analytics rule that relies on Fusion, and confirm with your compliance owner whether the Defender retention and privacy terms still meet your data residency and log retention obligations. Microsoft's transition guide for the Defender portal is the natural next read for the migration steps and the points where your workspace configuration changes.
Staying in the Microsoft stack suits teams that already depend on Defender XDR and want its correlation model. Self-hosted Wazuh with OpenSearch fits better when you need to control retention and data location yourself, when your logs must stay inside a network that cannot reach Azure, or when you want a SIEM whose license does not depend on the portal a vendor requires you to use. Wazuh is licensed under GPLv2, and OpenSearch under Apache 2.0. Before deciding, confirm the sizing, staffing and support model for your daily volume, because those costs recur and are yours to carry in a self-hosted design.
What the retirement changes for data residency, retention and detections
Once Sentinel runs only in the Microsoft Defender portal, the governing rules for your logs, their retention and the detections that fire on them change with it. The cards below cover the four requirements that move with the portal, and each one is something to confirm in your own tenant before the cutover date.
Where logs are governed
After the move, Microsoft Defender XDR data privacy and retention policies apply to your Sentinel data instead of Sentinel's own policies. Compliance teams that documented data handling for Sentinel will need to review that documentation against the Defender XDR terms. Confirm which regions your workspace data is stored in and which Defender XDR policies now govern it, and record the answers in your data inventory.
How retention is set now
Retention for Sentinel data in the Defender portal follows the Defender XDR retention policies rather than the settings you may have configured in the Azure portal. Retention that was set per table or per workspace may not carry over unchanged. Export your current retention settings from each workspace and compare them with the Defender XDR policies before the portal switch, so any gap between what you keep and what you are required to keep is found early.
Fusion replaced by XDR correlation
The Fusion analytics rule is replaced by Defender XDR correlation when Sentinel runs in the Defender portal. Detections that depended on Fusion to link multi-stage attacks will behave differently, and the alert grouping you rely on may change. List every Fusion-driven detection, test how Defender XDR correlation groups the same activity, and document any detection you need to keep through a custom rule.
Portal access for analysts and auditors
Analysts, auditors and external reviewers who open Sentinel today in the Azure portal will work in the Microsoft Defender portal after the change. Role assignments, workspace access and the evidence that auditors export may need to be reissued under the new portal. Review who holds access now, confirm which roles map to the Defender portal, and check that audit evidence collection still works before the March 31, 2027 date.
How the Defender portal move changes your existing rules and policies
The core change is that data privacy and retention policies stop being set by Sentinel and start being set by Microsoft Defender XDR once Sentinel runs in the Defender portal. Microsoft's guidance states that Defender XDR data privacy and retention policies apply instead of Sentinel's own. For a regulated organization, this means the retention periods, data handling terms and access settings you documented for Sentinel no longer describe the platform you operate. Before the cutover, compare your current Sentinel retention configuration with the Defender XDR settings your tenant will apply, and record the differences in the compliance evidence you already keep. Ask your Microsoft account team and your privacy office to confirm which policy governs each data type you ingest, because the answer may differ between tables and workspaces in your environment.
Analytics rules need the most attention. Microsoft states that the Fusion analytics rule is replaced by Defender XDR correlation, so any detection logic, incident grouping or reporting that depended on Fusion must be re-validated rather than assumed to carry over. Review every enabled rule, including scheduled, near-real-time and Microsoft-provided templates, and confirm which ones are supported in the Defender portal and which need rewriting. Check connector status too, since data sources that feed a rule can stop producing alerts silently if their connector does not carry over. Test each rule against known-good and known-bad events in a non-production workspace where possible, and keep a dated inventory of what changed. Microsoft's transition guide lists the exact steps, and you should verify in your own environment which rules and connectors carry over. If a meaningful share of your detections depends on Fusion or on Sentinel-specific retention, that is a signal to scope the self-hosted option in parallel, since those dependencies are the parts you would otherwise rebuild twice.
Staying in Defender or moving to self-hosted Wazuh and OpenSearch
Organizations already standardized on Microsoft Defender, Entra ID and Microsoft 365 often get the most value from staying with Sentinel in the Defender portal. Moving to the portal has no extra cost according to Microsoft, and the Microsoft data connectors, Defender XDR incident handling and shared identity model remain in place. Wazuh with OpenSearch becomes the better choice when the team wants to reduce per-gigabyte exposure as log volume grows, or when it needs to own the detection logic and the storage layer. The Defender path keeps:
- Native Microsoft telemetry: Defender XDR signals and Microsoft data connectors continue to feed the workspace without custom collectors.
- Single incident queue: Correlation across Defender XDR products replaces the Fusion analytics rule, so triage stays in one portal.
- Vendor-managed platform: Microsoft operates the SIEM service, scaling and patching.
- Shared Microsoft skills: Analysts already trained on Defender can work without relearning a second query and alerting model.
- Microsoft retention policies: Defender XDR privacy and retention rules apply, which may be acceptable if they match your obligations.
Before deciding, confirm with Microsoft how Defender XDR retention and correlation differ from your current Sentinel rules, and ask for the current billing model for your ingestion volume.
Some organizations must keep security logs under retention and access rules they set themselves, which the Defender portal does not allow once Defender XDR privacy and retention policies apply. Self-hosting Wazuh with OpenSearch gives the security team direct control over index lifecycle, retention periods, access roles and where each copy of the data sits. Wazuh is licensed under the GNU General Public License version 2, and OpenSearch under Apache 2.0, so the platform carries no per-gigabyte license fee. The Defender path keeps:
- Defender-managed retention: Data follows Microsoft's Defender XDR policies, which the security team does not configure per index.
- Microsoft-hosted storage: Logs remain in Microsoft's service, with data handling governed by Microsoft's terms.
- Built-in integrations: Microsoft connectors and Defender sensors stay supported without custom ingest pipelines.
- Cloud-scale ingestion: Microsoft handles capacity planning for large log volumes.
- Documented correlation: Defender XDR correlation is supported by Microsoft's documentation.
Wazuh offers role-based access control in its dashboard, while OpenSearch Security provides document-level and field-level security, field masking and audit logs. Confirm with your data protection officer which retention periods and access controls your obligations require, then check whether those controls exist in each option.
Sentinel runs only in Azure, so networks that cannot reach Azure cannot use it, even before the portal change. Wazuh can be deployed from an offline installation guide, and its vulnerability detection can use an offline vulnerability repository instead of the Wazuh Cyber Threat Intelligence platform. OpenSearch can be installed from tarball or RPM packages, which suits sites that must stage software through a controlled transfer process. The Defender path keeps nothing for these environments, because the service depends on connectivity to Microsoft's cloud. The better choice here is Wazuh with OpenSearch when:
- Disconnected sites: Classified, OT or disconnected networks need the full SIEM to run without internet access.
- Offline updates: Patch and rule updates must be brought in through an approved transfer process, which you should plan and test.
- Local vulnerability data: An offline vulnerability repository replaces the online threat intelligence feed.
- Local identity sources: OpenSearch Security supports LDAP, Active Directory and Kerberos, so sign-in can stay on site.
- Agent coverage: Wazuh agents run on Linux, Windows, macOS, Solaris, AIX and HP-UX, with agentless syslog monitoring for devices such as firewalls.
Confirm the offline update process with your security authority before deployment, and check that the operating systems and hardware in the isolated zone match the versions Wazuh and OpenSearch support.
Which open-source options replace Sentinel after the retirement
This table compares three self-hosted options against the capabilities Sentinel customers rely on today. Read each row as a question to test in your own environment, since feature parity depends on your data sources, detection content and staffing. The main takeaway is that Wazuh covers the most ground out of the box, OpenSearch gives the most control but needs the most build work, and Graylog sits between them.
| Capability | Wazuh | OpenSearch | Graylog Open |
|---|---|---|---|
| License | GPLv2 with an OpenSSL linking exception | Apache 2.0 across the project, no paid edition described in the FAQ | Server Side Public License, free to self-manage with no ingestion caps |
| Architecture | Indexer, server, dashboard and agents on endpoints | Search and analytics engine plus Dashboards; you assemble collection and detection | Server, data node and Sidecar for collection; search and dashboards included |
| Endpoint coverage | Agents on Linux, Windows, macOS, Solaris, AIX and HP-UX | No agents; you rely on Beats, Fluent Bit or other shippers you choose | Sidecar and inputs; agent coverage depends on the shipper you deploy |
| Agentless devices | Syslog collection and periodic API probing for firewalls and similar devices | Depends on the ingest pipeline you build | Syslog and cloud inputs |
| Compliance views | Prebuilt dashboards for PCI DSS, GDPR, HIPAA and NIST 800-53 | Not prebuilt; you map controls to queries and dashboards yourself | Content depends on the packs and rules you configure |
| Access control and audit | Single sign-on documented; confirm role model against your needs | SAML, OpenID Connect, LDAP, Kerberos, RBAC, field-level security and audit logs via the Security plugin | Role-based access; confirm SSO and audit options for your edition |
| Query language and content | Rules and decoders in Wazuh format; no KQL compatibility | Query DSL and Dashboards; KQL queries must be rewritten | Graylog search syntax; KQL must be rewritten |
| Microsoft data connectors | Not built in; use agents, syslog or exported logs from Microsoft sources | Not built in; build ingest pipelines for each Microsoft source | Not built in; configure inputs for each Microsoft source |
| Operating burden | Lower for a single cluster, but the indexer and server still need sizing and patching | Highest; you own cluster sizing, upgrades and security configuration | Moderate; you run the server and storage nodes and manage Sidecar |
How to move from the Azure portal to self-hosted Wazuh and OpenSearch
This sequence turns a Sentinel estate into a documented, tested Wazuh and OpenSearch deployment in which every data source, detection and access rule has a known replacement before any traffic is switched.
Inventory connectors, rules and retention
Your team lists every Sentinel data connector, analytics rule, scheduled query, workbook and Logic Apps playbook, with its log source, owner, retention setting and the compliance framework it supports. Compliance and security leads confirm which sources are mandatory for evidence. The step is done when each item has a named replacement path (a Wazuh agent, syslog collection, an API probe or a rewritten rule) or a documented decision to retire it. Export the analytics rule definitions now, since they are the input for rewriting detections later.
Choose the Wazuh deployment option
Platform engineering selects one of the documented Wazuh deployment options: all-in-one or step-by-step installation, the prebuilt OVA virtual machine, an Amazon Machine Image, Docker, Kubernetes, Ansible or Puppet, or the offline installation guide for networks without internet access. Choose offline if the environment is isolated, and confirm with your own infrastructure team which option matches your hosting platform and change process. The step closes with a written architecture note naming where the indexer, server and dashboard will run and who operates each component.
Size the indexer and server cluster
Architects size the Wazuh indexer, which stores alerts and runs full-text search, and the Wazuh server, which runs decoders and rules. The server can run as a cluster for horizontal scaling, so the plan should state the expected endpoint count, log volume and retention window, then ask the vendor and your own test results to confirm node counts. Storage and memory are sized from a pilot load, not from assumptions. The step is complete when a sizing sheet, approved by operations, shows capacity headroom for the Sentinel-equivalent volume and a tested rollback point for the cluster.
Deploy agents and syslog collection
Endpoint teams install Wazuh agents on Linux, Windows and macOS systems, and configure agentless collection through syslog and periodic API probes for firewalls and other devices that cannot run an agent. Agent configuration and upgrades are managed centrally from the server. Each replaced Sentinel connector is mapped to its new source and tested with sample events. The step is done when every source in the inventory appears in the dashboard, with timestamps and fields checked against the original Sentinel records.
Set OpenSearch Security authentication and roles
Security engineers connect the OpenSearch Security plugin to your identity provider using SAML, OpenID Connect, Active Directory or LDAP, then define role-based access for analysts, administrators and auditors. Field-level and document-level security and field masking limit exposure of personal data, and audit logs and TLS for node and client traffic are switched on and verified. Access reviews should mirror the Sentinel role assignments you exported. The step closes when a test user from each role can sign in and only sees the data their role allows, and the audit log records those actions.
Run parallel alerting
The security operations team runs Wazuh rules and the rewritten detections alongside the live Sentinel analytics rules for an agreed observation period. Each alert is compared with the Sentinel result, and differences are classified as missed detections, false positives or timing gaps. Tuning continues until the team accepts the alert quality for each high-priority use case. The step is complete when the comparison log shows every critical detection reproduced in Wazuh, with sign-off recorded by the detection owner and the SOC lead.
Cut over and retire Sentinel
The team switches incident routing, dashboards and compliance reporting to the Wazuh dashboard, confirms that Sentinel exports are archived in the agreed format, and disables the Sentinel connectors in sequence. Backups of the indexer and configuration are verified, and an upgrade plan follows the Wazuh and OpenSearch release cadence. The step is done when a documented cutover record exists, the archived Sentinel data has been restored and read successfully in a test, and the compliance owner has confirmed the evidence set. Confirm any Microsoft billing or retention terms with your account team before you shut anything down.
Running Wazuh offline when Azure access is not an option
Sentinel cannot be installed in an isolated network because it runs only as an Azure service, so any site that cannot reach Azure has to look at a self-hosted platform from the start. Wazuh documents an offline installation guide for systems without internet connectivity, and its vulnerability detection can use an offline vulnerability repository in place of the Wazuh Cyber Threat Intelligence (CTI) platform. That combination is what makes Wazuh the relevant option for classified, OT and disconnected sites, where the question is not whether to move off the Azure portal but whether the monitoring stack can run at all in the enclave.
The vulnerability feed is the part to settle before any installation work begins. Wazuh vulnerability detection matches the software inventory that agents collect against vulnerability data from either the CTI platform or your offline repository, so without a feed the inventory still builds but findings stop reflecting current advisories. Confirm with your security team how that repository gets updated, who approves each import, how the update is moved across any cross-domain or data diode path, and how often the team will check that the imported data is current. Check the update procedure in the offline installation guide against your own transfer rules rather than assuming the default workflow fits.
Plan the offline build as a managed transfer exercise. Installation packages, the offline vulnerability repository, OpenSearch-backed indexer components and any container or operating system dependencies must be staged on media that your accreditation process allows, and each artifact should be hashed and recorded so the receiving team can prove what was imported. Ask your Wazuh contact or DeployOpen whether the chosen Wazuh release line and its patch cadence fit your change windows, because offline sites receive patches only when the transfer completes. Test the full cycle on a staging enclave first so that the first production update does not become the first time anyone has run it.
Evidence to keep for auditors during the platform change
What retention settings should I document before and after the Sentinel to Defender move?
Record the retention configuration of every log source before the move and the retention configuration that applies after it, so an assessor can see the change and its date. In Sentinel, the Log Analytics workspace retention and any archive or data lake settings are the baseline. When Sentinel runs in the Microsoft Defender portal, Microsoft Defender XDR data privacy and retention policies apply instead of Sentinel's, so the new values must be written down and compared against your own retention requirements. In Wazuh, the equivalent evidence is the index lifecycle policy and snapshot schedule in the Wazuh indexer, along with the retention period you set for alerts and archives. Keep exported configuration files and dated screenshots as part of the change record, and confirm each value with the vendor documentation for your tenant.
Which OpenSearch Security plugin records prove who can access log data?
The OpenSearch Security plugin produces the access control records that show who can read, search and change log data. Export the role definitions, role mappings, internal users list and backend configuration, which cover role-based access control, document-level and field-level security and field masking. Keep the authentication configuration for SAML, OpenID Connect, Active Directory, LDAP, Kerberos or JSON Web Token backends, because it shows how identities from your directory are tied to those roles. Dated exports taken at each change, together with the approval tickets that authorized them, let an auditor trace each permission back to a decision. Confirm the exact export procedure and file format against the OpenSearch security documentation for your version.
How should audit log settings be kept for the platform change?
Keep the audit log configuration and the audit logs themselves for the period before, during and after the change. The OpenSearch Security plugin provides audit logging, so record which categories are enabled, where the logs are written and who can read them. Preserve the Sentinel audit history and the Defender portal activity records you export during cutover, since they show who changed analytics rules, connectors and workspace settings. Store the logs in a location that the administrators performing the change cannot alter, and document the retention period you chose for them. Ask your auditor which audit trail they expect for the platform change, because requirements differ by framework and organization.
What evidence shows TLS is configured for node and client traffic?
Keep the TLS configuration for both node-to-node traffic inside the cluster and client traffic from dashboards, agents and ingest sources. For OpenSearch, the Security plugin configuration defines the certificates, certificate authorities and whether client certificates are required. Record the certificate issuer, validity dates, key lengths and the hostnames each certificate covers, and store the configuration files with a hash or version tag. For Wazuh, document the certificates used between agents and the server, between the server and the indexer, and for the dashboard. Include a test result from a TLS scan of each listening port so the configuration is verified, not only described. Confirm the cipher and protocol settings against your own policy before you record them as compliant.
How do compliance dashboards map to PCI DSS, GDPR, HIPAA and NIST 800-53?
Map each control your assessor cares about to the dashboard view, rule set or report that provides its evidence, and keep the mapping as a dated document. Wazuh's dashboard includes prebuilt views for PCI DSS, GDPR, HIPAA and NIST 800-53, which give a starting point for this mapping. Verify each view against the control wording in your own copy of the standard and state where a view does not cover a control, so the gap is visible. Export the dashboard state and the rule groups used, and keep them with the change record. Refer to the controls by their IDs and titles in the mapping. Use your compliance program or the assessor to confirm which controls apply and what counts as sufficient evidence.
What change records are needed for rewritten detection rules?
Each rewritten detection rule needs a change record that links the original Sentinel analytics rule, the new rule in the target platform, the reviewer and the test results. Keep the exported ARM template or rule definition from before the move, the new rule file, and a log of the tests that showed the rule fires on the same test data. Note any change in logic, severity, mapped technique or alert routing, because a silent change in detection coverage is the finding auditors are most likely to raise. Store the records in your change management system with approval dates. Ask your assessor how they want rule mappings presented, and confirm the expected format before the cutover.
Who prepares this evidence and who issues the opinion?
DeployOpen prepares the systems, controls and evidence for the platform change, including configuration exports, access records, TLS documentation, dashboard mappings and rule change records. DeployOpen does not issue audit reports, attestations or certifications. Independent auditors and certification bodies review the evidence and form any opinion on it. Agree in writing, before work starts, which evidence items are in scope, who receives each one and how it is stored. Ask your assessor to confirm the evidence format and sampling approach in advance, so the work matches what they will test.
Can my own team keep this evidence current after the move?
Yes, if the evidence is produced on a schedule and assigned to named owners. The recurring items are access exports after each role change, TLS certificate records when certificates renew, retention and audit log reviews, and rule change records for every edit. Each of these is a small, repeatable task, but each needs an owner and a calendar entry. Automate exports where the platform allows it, and store them in a location separate from the systems they describe. Ask your internal audit or compliance lead which review cycle they require, and confirm that cycle with the auditor before the platform goes live.
How DeployOpen plans and deploys this move under your constraints
DeployOpen plans, implements, hardens, documents and supports self-hosted open-source platforms inside your own cloud accounts, data centers or air-gapped networks. For the Sentinel portal move, we start by mapping which decisions stay in Microsoft's Defender portal and which would move to a self-hosted Wazuh and OpenSearch deployment, then build only what the written scope covers.
How we scope
- Scope, deliverables and timelines are agreed in writing before any work starts.
- The written scope lists which Sentinel data sources, analytics rules and workbooks are in or out of the move.
- Ask us to confirm how Defender XDR retention and privacy policies affect the data you must keep.
How we build
- Wazuh components are deployed in your environment using the installation method your constraints allow.
- Detection logic is rebuilt as Wazuh rules and OpenSearch queries, then tested against agreed sample events.
- Access is controlled through OpenSearch Security roles, TLS and your chosen single sign-on backend.
How we hand over
- You receive documentation covering architecture, configuration, rule inventory and operating procedures.
- Parallel running and cutover steps are documented so your team can verify alert parity before Sentinel is retired.
- Compliance evidence we prepare supports your assessment, but independent auditors issue any report or certification.
How we support
- Ongoing support, upgrades and continuous testing are available as retainers after handover.
- Patch and upgrade planning follows the Wazuh and OpenSearch release cycles, which you should confirm for your version.
- Ask for a quote that separates one-time migration work from recurring support and retainer hours.

