Cover of Microsoft Sentinel Field Guide
Microsoft Defender Field Guides

Microsoft Sentinel Field Guide

Cloud-native SIEM security operations, from workspace onboarding and the data lake to analytics rules, KQL threat hunting, and Microsoft Defender XDR

This is the companion page for Microsoft Sentinel Field Guide. If the book is in your hands, start with the placement quiz: it tells you which chapters to read first. The chapter map, the hardest SC-200 questions answered in full, and the topics sitting just outside the book are all on this page.

Get the book on Amazon See where you are

What this guide covers

Microsoft sells Sentinel as a cloud-native SIEM and security operations platform, and the Learn overview will tell you exactly that in one sentence. What it does not tell you is that the whole product has moved into the Microsoft Defender portal, that the Azure portal it was built in stops working on March 31, 2027, and that onboarding to the unified portal can silently disable Fusion (Sentinel's built-in machine-learning rule that fuses lower-fidelity signals into a single multistage-attack incident), rewrite the scope of your automation rules, and strip the incident-provider condition out from under detections you already shipped. The documentation describes each of those changes in its own article. None of them warns you which one breaks what you already built.

This is the connected account the docs never put in one place. It walks the product from the workspace up: the two-tier data lake and what mirroring costs, the licensing math behind Sentinel being generally available without an E5 or a Defender XDR entitlement, ingestion mechanics that double-bill you when a syslog facility is shared, and the RBAC and automation gotchas that fail with cryptic errors instead of clear ones. Every limit, license boundary, and portal path is grounded against current Microsoft Learn and validated in a live Microsoft 365 E5 lab.

You do not need a production SOC to follow along. A trial tenant and a single connected data source are enough to reproduce nearly every behavior in this book.

Chapter map

  1. Microsoft Sentinel, What It Is and Where It Lives
  2. Licensing, Prerequisites, and Cost Planning
  3. Workspace Design and Onboarding
  4. Roles, RBAC, and Access Control
  5. Data Tiers, Retention, and the Sentinel Data Lake
  6. The Content Hub and Data Connectors
  7. Ingestion Mechanics, AMA, DCRs, Windows Events, Syslog, and CEF
  8. Analytics Rules, Detection at Every Speed
  9. Watchlists, Threat Intelligence, and Hunting
  10. UEBA, User and Entity Behavior Analytics
  11. Automation Rules, Playbooks, and SOAR Patterns
  12. Workbooks, SOC Optimization, and Operational Visibility
  13. Sentinel and Defender XDR Unified, Before You Go to Production

Appendices: PowerShell and Graph Quick Reference; Build a Lab Tenant; Production Readiness Checklist.

Who it is for

This book is for administrators and engineers who run Microsoft Sentinel and want the moving parts wired together rather than listed, for IT pros ramping onto the product who need one coherent production-focused reference, and for anyone sitting the SC-200 (Microsoft Security Operations Analyst) exam who wants depth beyond exam cramming; the material tracks the SC-200 skills measured as of July 28, 2026, which explicitly cover the Sentinel data lake, Sentinel Graph, KQL jobs, notebooks connected to the Sentinel MCP server, and agentic AI tools including embedded Microsoft Security Copilot.

See where you are

7 questions from this guide's territory (3 easy, 4 medium), with the explanation after each. Nothing is stored and nothing is sent anywhere.

The rest of Microsoft Defender Field Guides

One product per book, a lab appendix in every one, and a companion page like this one for each.

  • Cover of Microsoft Defender for Endpoint Field Guide
    Field guide

    Microsoft Defender for Endpoint Field Guide

    Most MDE deployments fail quietly, not loudly.

  • Cover of Microsoft Defender XDR Field Guide
    Field guide

    Microsoft Defender XDR Field Guide

    A thin, low-priority incident is far more often a coverage gap than a clever adversary, and the portal warns you about neither. This is not a product you turn on.

  • Cover of KQL Threat Hunting Field Guide
    Field guide

    KQL Threat Hunting Field Guide

    A hunting query returns zero rows and nobody in the room can say whether the environment is genuinely clean or the table was never populated because the connector behind it was never turned on.

The hard set (3 questions)

The hardest SC-200 questions this guide publishes, none of them repeated from the placement quiz above: the answer first, why every other option is wrong, and the Microsoft Learn page behind it. The practice book for this exam holds the full bank.

During a UEBA investigation you want to know which other users Microsoft Sentinel considers behavioral "peers" of a suspected compromised user, so you can judge whether an action is unusual relative to that peer group. Sentinel ranks the top peers using a TF-IDF weighting in which smaller groups carry higher weight. Which UEBA table holds this dynamically calculated peer ranking?

Correct answer: D. UserPeerAnalytics

UserPeerAnalytics contains dynamically calculated peer groups, ranking the top 20 peers based on security group membership, mailing lists, and other associations. It uses the TF-IDF algorithm to normalize weights, so smaller groups carry higher weight.

Why the other options are wrong:

  • A. BehaviorAnalytics stores enriched behavioral events with the InvestigationPriority score, not peer rankings.
  • B. UserAccessAnalytics is not a Microsoft Sentinel UEBA table; it is a distractor.
  • C. IdentityInfo stores entity profile attributes (roles, blast radius, group membership), not the ranked peer list.

Memory hook: Peers? UserPeerAnalytics: top 20, ranked by TF-IDF.

Microsoft Learn: Ueba reference

Your organization onboarded Microsoft Sentinel to the Defender portal and integrated Microsoft Defender XDR incidents, and you are also ingesting Defender for Cloud alerts. You now see duplicate incidents for the same Defender for Cloud alerts. Which set of actions eliminates the duplicates?

Correct answer: B. Configure the Tenant-based Defender for Cloud connector, disconnect the Subscription-based (Legacy) connector, and turn off any analytics/incident-creation rules that create incidents from Defender for Cloud alerts

When Defender for Cloud incidents already flow in through the Defender XDR incident connector, you should use the Tenant-based Defender for Cloud connector to sync alerts, disconnect the Subscription-based (Legacy) connector, and disable any Sentinel analytics/incident-creation rules built on Defender for Cloud alerts. Together these steps prevent the same alert from generating duplicate incidents.

Why the other options are wrong:

  • A. Disconnecting the XDR connector abandons XDR incident integration and isn't the recommended fix.
  • C. Bi-directional sync isn't the source of duplicate incidents; the legacy connector and incident rules are.
  • D. Running both connectors is what causes the duplication.

Memory hook: XDR path: go Tenant-based, drop the Legacy connector and its incident rules.

Microsoft Learn: Concept integration 365

Your team relied on Microsoft Sentinel livestream to run a query continuously and push notifications when new matches appeared. Per current Microsoft guidance, livestream is being retired. Which alternatives does Microsoft now recommend for persistent query results and notifications to messaging platforms?

Correct answer: B. KQL jobs, analytics rules, or playbooks.

Current Learn guidance states Sentinel livestreams are no longer available and directs you to use KQL jobs, analytics rules, or playbooks instead. These alternatives offer persistent query results and support notifications to various messaging platforms.

Why the other options are wrong:

  • A. Guided hunting and sample queries help you author queries; they don't automate persistent results or alerts.
  • C. Bookmarks and the investigation graph preserve and visualize findings; they don't provide continuous querying or notifications.
  • D. Restore historical data and search jobs retrieve archived/historical data; they aren't continuous notification mechanisms.

Memory hook: Livestream retired; use KQL jobs, analytics rules, or playbooks.

Microsoft Learn: Hunting

Exam prep? The SC-200 sampler scores you by topic and points you to the guide behind each miss.

Take the SC-200 sampler Every SC-200 practice question, with explanations