T Techclick ← All lessons
ISC2 CISSP · Domain 7 · Security Operations

CISSP Domain 7 — contain first, then restore with proof

The ticket says “wipe the box.” The stem says FIRST or IMMEDIATELY. The engineer brain reimages. The CISSP brain validates the alert, stops the bleed, bags the evidence, then restores to a number the BIA actually named. This page is that shift — IR, logging, forensics, BCP/DR, need-to-know, job rotation — using the official ISC2 Domain 7 outline and NIST SP 800-61, not a vendor playbook.

22 min read · L2 primary · Quiz at end

⚡ Quick Answer

CISSP Domain 7 manager lesson: validate then contain, preserve evidence, pick need-to-know vs job rotation, match RTO/RPO to the BIA. Official ISC2 outline + NIST 800-61.

After this page you can

Quick answer

An event is noise until you prove harm. An incident is a confirmed adverse event. FIRST on a SIEM fire is validate and assess scope. Mitigation in the official 7.6 list is containment — stop the bleed — not a rebuild. Capture volatile evidence first, hash it, and keep a chain of custody. Need-to-know is which information. Least privilege is which rights. Job rotation (and mandatory vacation) is the detective control for one-person fraud. BIA first, then RPO/RTO, then a site and a test. People safety outranks the server.

Say this out loud

I do not wipe first. I confirm it is an incident, contain the blast radius, preserve what I will need in court, then recover to the BIA numbers — and I write the lesson so the next one is cheaper.

1. Why IMMEDIATELY is not “wipe it”

Interview, 35 minutes in. “Ransomware alert on a file server. What does the security professional do first?” The engineer answers “reimage from last night’s backup.” That is recovery. It is not first. It also destroys the only copy of the attacker’s foothold.

ISC2 Domain 7 — Security Operations — is 13% of the official April 2024 exam outline. The outline is not asking you to click Rebuild in vCenter. It is asking whether you can run investigations, logging, incident management, and continuity so the business survives and the evidence still stands up.

Hero · the SOC is a decision desk, not a wipe button
SOC desk with Detect, Respond, Recover labels and an incident ticket
Notice: the signed ticket is the work. Detect / Respond / Recover are phases, not three products you buy.

Three reasons this domain fails strong operators:

Hard words before the runbook

Event — any observable occurrence. Most are noise. NIST SP 800-61 Rev. 2 uses this split.

Incident — a violation or imminent threat of violation of security policy, or an event that jeopardizes CIA. Outline 7.6 starts once you have one.

Mitigation (7.6) — containment. Stop further damage. Not the same as eradication or a rebuild.

Need-to-know — access to specific information required for the job. Content.

Least privilege — the minimum rights needed to perform the function. Actions.

RPO — how much data you can lose (backup frequency). RTO — how fast IT must return to a defined service level. MTD / MAO — how long the business can stand the outage. Typically RTO + WRT ≤ MTD.

2. Mental model · detect, decide, preserve, restore

Hold four parts. Interviews fail when people treat SIEM as “the answer,” treat IR as “wipe and patch,” or shop for a hot site before the BIA.

Journey · detect, contain, eradicate, recover
Four glass panels labeled Detect, Contain, Eradicate, Recover
Read left to right. Contain sits before eradicate. Recover is last among the four — and lessons learned still come after recover.

Part 1 · two official IR lists, one sequence

Objective 7.6 names seven activities in this order: detection, response, mitigation, reporting, recovery, remediation, lessons learned. That is the list to recite on the exam.

NIST SP 800-61 Rev. 2 (the CBK workhorse, August 2012) uses four phases: Preparation → Detection and Analysis → Containment, Eradication, and Recovery → Post-Incident Activity. Mitigation in 7.6 is the containment bucket. Remediation plus lessons learned is post-incident.

NIST SP 800-61 Rev. 3 (April 2025) remaps the same work onto CSF 2.0. Govern, Identify, and Protect are preparation — they are not the incident itself. The incident is Detect, Respond, Recover. Improvement is continuous, not a meeting you skip. Know r3 exists; the stem will still reward the 7.6 verbs and “contain before eradicate.”

Part 2 · logging is how detection earns its keep

Objective 7.2 is not “buy a SIEM.” It is IDPS, SIEM, continuous monitoring and tuning, egress monitoring, log management, threat intelligence (feeds and hunting), and UEBA. A SIEM that is not tuned is an alarm clock nobody believes. The official Domain 7 AI note is the same idea: use ML to cut alert fatigue and raise fidelity — then a human still validates.

Logs only correlate if clocks agree. Protect log integrity (append-only, separate store) or the investigation is hearsay with a timestamp.

Part 3 · people controls are operations, not IAM trivia

Objective 7.4 lists need-to-know / least privilege, SoD, privileged account management, job rotation, and SLAs. These are how you run the shop every day. Domain 5 designs the access model. Domain 7 is whether the SOC, the admin team, and the finance clerk actually live it.

Flow 1 · four people controls, four jobs
Outline 7.4 · pick the control the stem actually named Need-to-know Which files? Analyst sees only their queue — not HR salary data. Content Least privilege Which rights? Helpdesk resets passwords — cannot export the directory. Actions SoD Split the process Requester cannot also approve the wire and reconcile. Fraud prevention Job rotation Swap the chair Detective + backup skill. Vacation is the close cousin. Detect + resilience

Read left → right. If the stem is about seeing a file, it is need-to-know. If it is about being able to do a thing, it is least privilege. If one person can complete a crime alone, it is SoD. If you need to catch ongoing fraud or remove a single point of failure, it is rotation.

Say this out loud

Need-to-know is the file. Least privilege is the button. SoD splits the process. Job rotation finds the fraud the SIEM never saw.

3. Decision flow · the IR path

Decision · contain vs eradicate is not the first diamond
A glowing decision diamond splitting into paths labeled Contain and Eradicate
Caption, not the art: you only reach contain vs eradicate after the alert is validated. Skipping analysis is the engineer-brain miss. Eradicate is the later branch.
Flow 2 · from SIEM fire to lessons learned
7.6 verbs on the 800-61 path Alert / event (SIEM, UEBA) Validated incident with scope? no · tune Close as event document + tune yes · 7.6 Response Activate IR + impact Preserve volatile evidence Mitigation = contain Isolate. Do not rebuild yet. Eradicate → recover Remediate + lessons Fix root cause. Write it down. 800-61 r2 phases 1 Preparation (before the fire) 2 Detection & Analysis 3 Contain · Eradicate · Recover 4 Post-incident activity r3: Detect / Respond / Recover Govern · Identify · Protect = prep

Read top → bottom on the right. Diamond = decision. Green = you can leave the incident path. Reporting in 7.6 runs throughout — legal, leadership, regulators — not only at the end.

800-61 Rev. 2 is explicit: document containment strategies before you need them, and preserve evidence while you contain. Eradication without a picture of scope leaves a second foothold. Recovery without a clean image puts the same malware back.

4. How to choose people controls, sites, tests

Use the tables. Qualitative judgment is enough for most FIRST questions. Numbers (RPO hours, RTO hours) show up when the stem hands them to you.

If the ticket looks like…ControlWhy not the cousin
Analyst can open only their queue, not payroll files. Need-to-know Least privilege is about the action (reset vs export), not the file set.
Helpdesk can reset passwords but cannot create domain admins. Least privilege (+ PAM for the admin path) Need-to-know would hide a document. This hides a capability.
One clerk requests, approves, and reconciles the same payment. Separation of duties Rotation might later detect the fraud. SoD tries to make the fraud impossible.
Long-tenured operator, no backup, auditor suspects silent fraud. Job rotation or mandatory vacation A new SIEM does not watch a paper process. Rotation is detective + resilience.
Vendor must restore email in four hours or pay. SLA An SLA is a contract metric. It is not a substitute for your own BIA.
BIA numberWhat it measuresWhat you buy
RPO Maximum acceptable data loss (time). Backup frequency, replication, journal. Hourly RPO is not a weekly tape.
RTO Maximum time to restore a defined service level. Hot / warm / cold / cloud / reciprocal site, HA, spare capacity agreements (outline 7.10).
WRT Time to verify integrity and resume the business after IT is up. Runbooks, people, comms. A restored database that finance cannot trust is not recovered.
MTD / MAO How long the process can be down before irrecoverable harm. If RTO + WRT > MTD, the strategy fails the BIA. Declare disaster when MTD is at risk.
Flow 3 · time numbers and site heat
BIA clock · then pick the site that can hit it Disaster t = 0 RTO IT service back WRT business verified ≤ MTD or the firm dies RPO data look-back Hot Live or near-live replica Hours or less. Expensive. Tight RTO / RPO Warm Hardware in place, data / staff to load. Medium RTO Cold Empty space / power. Cheap. Days to weeks. Loose RTO only

Read left → right on the clock, then pick heat. Outline 7.10 says “cold vs. hot” and resource capacity agreements. Cloud and reciprocal sites are valid strategies if they hit the BIA. Geographic separation so one flood does not take both sites.

Official 7.12 testWhat you actually doHits production?
Read-through / tabletop Sit with the plan. Talk the scenario. Find missing names and phone trees. No
Walkthrough Role-owners walk their own steps and correct the document. No
Simulation Act the disaster. Stop before you touch live systems or data. No
Parallel Bring the recovery site up while production stays up. Compare results. No (recovery side only)
Full interruption Take production down and run the real fail-over. Yes — highest proof, highest risk
Communications Stakeholders, test status, regulators — named in 7.12 on purpose. Depends on the exercise
BCP is not DR

Objective 7.11 is Disaster Recovery (IT, facilities, comms, restoration). Objective 7.13 is Business Continuity (the business process keeps running). Domain 1.7 already told you the BIA comes first. Domain 7 is where you implement and test. People safety (7.15 — travel, 2FA fatigue, emergency, duress) outranks the server every time.

5. Runbook · Side A log, B IR/forensics, C BCP/DR

This is not a vendor console path. It is the Domain 7 operating path you walk on an alert, a laptop seizure, or a declared disaster. Each side cites one primary source.

Side A · logging and detection (due diligence of the SOC)

Primary source: ISC2 outline 7.2 and NIST SP 800-92, Guide to Computer Security Log Management.

  1. Collect the right sources, not every syslog on earth

    IDPS, firewalls, endpoints, identity, cloud control plane, egress. 7.2 names egress monitoring on purpose — exfil is how you find the quiet incident.

  2. Normalize, correlate, tune

    SIEM aggregates and correlates. UEBA watches user and entity baselines. Threat intel (feeds + hunting) tells you what “interesting” looks like this week. Untuned rules are alert fatigue — the official Domain 7 AI paragraph is about raising fidelity, not auto-closing tickets.

  3. Protect the log itself

    Time sync. Integrity. Retention that matches legal hold and policy. If the attacker can edit the only copy, you have no investigation.

  4. Treat the first fire as an event

    Validate. Deduplicate. Ask “what else is in scope?” A single host alert that is really a subnet is a different incident. This is Detection and Analysis, not Recovery.

Side B · incident, investigation, forensics

Primary source: ISC2 outline 7.1 / 7.6, NIST SP 800-61 Rev. 2 (four-phase handling) and Rev. 3 (CSF 2.0 Detect / Respond / Recover).

  1. Secure the scene, then decide live vs dead

    Once evidence is contaminated, you cannot decontaminate it. Live systems hold RAM, cache, and crypto keys. Power-off destroys them. A dead-box disk image is right when volatility is already gone or the box is off. Do not “just pull the plug” because a blog said so.

  2. Collect in order of volatility

    Well-known order (RFC 3227 / NIST SP 800-86): registers and cache → routing/ARP/process table/memory → temp file systems → disk → remote logs → physical config and topology → archives. Hash before and after the image. Use a write blocker on media. Bit-for-bit copy; work from the copy.

  3. Start and keep the chain of custody

    Who handled what, when, where, why. Seal it. Outline 7.1: evidence collection and handling, reporting and documentation, artifacts (data, computer, network, mobile). Five classic rules of evidence: authentic, accurate, complete, convincing, admissible. Chain of custody helps admissibility; it does not guarantee it.

  4. Contain, then eradicate, then recover

    Mitigation = isolate, sinkhole, disable the account, segment the VLAN. Reporting runs in parallel to legal, leadership, and any regulator the incident triggers. Recovery returns service. Remediation removes the root cause (patch, key rotate, kill the bad rule). Lessons learned is a dated meeting, not a Slack shrug.

Side C · continuity and recovery

Primary source: ISC2 outline 7.10–7.13 and NIST SP 800-34 Rev. 1, Contingency Planning Guide.

  1. BIA first — then strategy

    Critical processes, impact over time, RPO / RTO / WRT / MTD, resource dependencies. Shopping for a hot site before the BIA is Domain 1’s “buy the WAF first” mistake in a different jacket.

  2. Pick backup and site to hit the numbers

    7.10: cloud / on-site / off-site storage; cold vs hot; resource capacity agreements; multiple processing sites; HA, QoS, fault tolerance. Incremental vs differential is CBK restore-math (incremental = last full + every incremental; differential = last full + one differential). Mirror is fastest and fattest. Test a restore, not just a backup job.

  3. Declare when MTD is at risk, restore in BIA order

    7.11: response, personnel, communications, assessment, restoration, training, lessons learned. Most critical first on the way to the recovery site. Least critical first when you fail back, so you prove the primary site before you bet the crown jewels on it.

  4. Test with the official 7.12 name

    Tabletop if you have never walked it. Parallel if you must prove the site without taking production. Full interruption only when leadership accepts the risk. Communications to stakeholders and regulators is a test type, not an afterthought. 7.9 change management still wraps patches (7.8) — a Friday-night emergency change is still a change.

Green path — you finished Domain 7 work when

The event was validated. Scope is written. Containment happened before the rebuild. Evidence is hashed and chained. Recovery matches the BIA. Root cause is fixed. A lessons-learned note exists. The next tabletop uses that note.

6. Runtime path after the incident

Recovery is not the end. 800-61 Rev. 3’s whole point is that Detect / Respond / Recover feed Identify.Improvement, which retunes Govern / Protect / Detect. A closed ticket with the same detection gap is a care failure.

Proof · chain of custody and a clean restore
Operations desk with sealed evidence bag, custody clipboard, and green health checks
Notice: the sealed bag and the signed clipboard are the investigation. The green screen is only useful if it matches a tested restore.

After IR

Indicators go into detection. Patch and vuln (7.8) close the hole through change (7.9). Configuration baselines (7.3) so the rebuilt host is not the same soft image.

After DR

Update the plan, the call tree, and the BIA if the business changed. A parallel test that failed is a gift — treat it as a finding, not a secret.

Physical + people

7.14 perimeter and internal physical controls. 7.15 travel, awareness (insider, social media, 2FA fatigue), emergency, duress. Life safety still wins over lock-down.

Media

7.5: manage and protect media at rest and in transit. Encrypt, track, sanitize at end of life. A backup tape in a rideshare is an incident, not a courier story.

7. Traps + proof checklist

Stem patternWhat ISC2 is testingEngineer trapManager move
SIEM / AI alert fires. FIRST? Detection and analysis (7.2, 7.6). Validate. Escalate, wipe, or page the CEO. Validate the alert and assess scope.
Confirmed malware. IMMEDIATELY / BEST next? Mitigation = contain (7.6, 800-61). Eradicate / restore now. Contain, preserve evidence, then eradicate.
Live laptop, insider theft 7.1 forensics. Volatility. Pull the plug or run full AV. Capture volatile data, write-block, hash, chain of custody.
One person can complete a payment 7.4 SoD vs rotation. Buy monitoring. Split the process (SoD). Rotation detects what SoD missed.
Cannot see payroll, can still admin Need-to-know vs least privilege. Call both “least privilege.” Hidden files = need-to-know. Hidden buttons = least privilege.
RPO 4 hours, RTO 1 hour 7.10 / 7.11 metrics. Swap the two letters. Buy a cold site. RPO is data loss. RTO is downtime. One-hour RTO is not a cold site.
Prove the DR plan without risking Friday traffic 7.12 test types. Full interruption because it sounds serious. Parallel (or tabletop if the plan has never been read).
Fire alarm vs locked data-centre doors 7.15 life safety. Fail-safe. Keep doors locked for confidentiality. People out. Fail-safe unlocks. Fail-secure is for the firewall, not the fire exit.
Common trap

Do not recite PICERL as if it were the ISC2 list. SANS PICERL is useful in a SOC. The exam outline’s 7.6 verbs are detection, response, mitigation, reporting, recovery, remediation, lessons learned. Mitigation is contain. Remediation is the lasting fix. They are not synonyms.

Pilot / interview proof checklist
Interview angle

Weak: “Domain 7 is the SOC and backups.” Strong: “I validate before I contain, I contain before I eradicate, and I do not recover on top of a dirty image. Need-to-know is the file; least privilege is the right; rotation is how I catch the fraud the SIEM never saw. RPO is data, RTO is time, and the BIA decides the site — not the brochure.”

Knowledge check

Six judgment items. Map each one to a FIRST/BEST stem, not a definition. Check answers, then reset and retry the misses.

Q1

A high-severity ransomware signature fires on one file server. An engineer opens a change to wipe and restore from last night. What does the security professional do FIRST?

Correct: b. FIRST on a SIEM fire is Detection and Analysis — confirm it is an incident and learn the blast radius. A wipe is recovery and destroys evidence. Press and fleet-wide patching come after scope. Re-read Decision flow and Side A.
Q2

SOC analysts can open only tickets in their own queue. They cannot open HR salary files even though they have a domain account. Which 7.4 control is this?

Correct: a. Hidden files are need-to-know. Least privilege would be about which admin actions they can perform. Rotation and DR tests are different 7.4 / 7.12 controls. Re-read Mental model and the choose table.
Q3

Legal seizes a still-running laptop suspected of insider data theft. What is the FIRST forensic move?

Correct: c. Order of volatility (RFC 3227 / NIST 800-86): memory and live state die first. Pulling power destroys them. AV changes the disk. Unencrypted email breaks confidentiality and chain of custody. Re-read Side B.
Q4

Scope is confirmed: one subnet is beaconing to a known C2. Leadership asks for the BEST next action.

Correct: d. 7.6 mitigation is containment. 800-61 wants you to stop the bleed and keep evidence, then eradicate and recover. Lessons learned is last. Waiting for a perfect report while C2 runs is not IMMEDIATELY. Re-read Decision flow.
Q5

Finance says they can lose at most four hours of ledger data, but the payment service must be back in one hour. What did they set, and which site class can even try?

Correct: b. RPO is data loss (4 hours). RTO is downtime (1 hour). A cold site cannot make a one-hour RTO. MTD is the business ceiling, not a substitute for naming RPO/RTO. Re-read How to choose and Side C.
Q6

An auditor flags a 12-year finance operator who is the only person who understands a manual reconciliation. No SIEM will ever see it. What is the MOST appropriate operations control?

Correct: a. Outline 7.4 names job rotation. Mandatory vacation is the well-known cousin. Both detect fraud and remove a single point of failure. A SIEM, a DR test, and a firewall allow-list do not watch a paper desk. Re-read Mental model and traps.

Sources

Related: CISSP overview (all 8 domains) · Domain 1: BIA and risk treatment · Domain 6: Assessment and Testing · Domain 8: Software Security · 8-week roadmap