# Free SOD practice questions

> 10 free SOD practice questions with the correct answer and an explanation.

Page: https://devopscodelab.com/exams/security-operations-decisions/free-practice-questions/
Practice exams: https://devopscodelab.com/exams/security-operations-decisions/

### Question 1 of 10

The SOC's phishing-report mailbox receives 150 user submissions per day, and analysts spend most of the shift opening each one by hand. Most submissions are marketing mail or duplicates of the same campaign. What is the MOST effective way to reduce analyst workload while still catching real phishing?

- A. Auto-close every submission whose sender domain has a valid SPF, DKIM and DMARC pass, since authenticated mail cannot be a phishing message
- B. Automate enrichment and cluster submissions by sender, URL or attachment hash so analysts review one case per campaign
- C. Stop accepting user phishing reports and rely only on the secure email gateway verdicts, which already scan every inbound message for malicious content
- D. Ask users to forward only messages they are certain are malicious, after a short training module

**Answer:** B

**Explanation:** Clustering and enrichment collapse duplicate reports into a single case and push obvious marketing mail down the queue, so analyst time goes to distinct campaigns. Dropping user reports removes a valuable detection source for mail that bypassed the gateway, authentication passes are routinely obtained by attackers using their own or compromised domains, and discouraging reports increases the chance a real phish is never reported.

### Question 2 of 10

A SIEM rule fires 400 times a day for 'PowerShell launched with an encoded command'. Review shows 97% of hits come from a single software-deployment tool running signed scripts from one management server. The SOC must cut noise without losing visibility into attacker use of encoded PowerShell. What is the BEST tuning change?

- A. Exclude only the deployment tool's signed scripts from its management server, keeping the rule active for all other sources
- B. Exclude every event where the PowerShell command line contains the -EncodedCommand flag from a host in the server subnet
- C. Raise the rule's threshold so it only alerts when more than ten encoded PowerShell launches occur on one host within an hour
- D. Disable the rule and rely on the EDR vendor's built-in PowerShell protections, which already block known malicious scripts

**Answer:** A

**Explanation:** A narrow exclusion keyed on the known-good parent process, signer and source host removes the benign volume while attackers who use encoded PowerShell elsewhere still trigger the rule. A count threshold lets a single malicious launch slip through, disabling the rule removes visibility entirely, and a subnet-wide exclusion blinds the SOC on every server, which is exactly where attackers move laterally.

### Question 3 of 10

A threat actor used a stolen service-account credential to create persistence on 30 Windows servers. The SOC has removed the scheduled tasks it found and wants to declare eradication complete. Which step is MOST important before closing eradication?

- A. Rebuild only the server where the initial access happened, since the other servers merely received pushed scheduled tasks
- B. Increase the logging level on the 30 servers for two weeks and close eradication if no new scheduled tasks appear
- C. Run a full antivirus scan on the 30 servers and close eradication if no malware signatures are detected on any of them
- D. Rotate the stolen service-account credential and secrets it could read, then hunt for other persistence types

**Answer:** D

**Explanation:** As long as the attacker still holds a valid credential, they can recreate persistence at will, so rotating it and anything it exposed is essential, followed by hunting for persistence beyond scheduled tasks. An antivirus scan misses living-off-the-land techniques, rebuilding one server leaves the others untrusted, and waiting for new scheduled tasks ignores other persistence mechanisms and the live credential.

### Question 4 of 10

An EDR alert shows a malicious script that disables shadow copies and stops backup services on a server. Encryption has not started. What is the BEST immediate action?

- A. Take a full backup of the server immediately so that files are protected before any encryption
- B. Restart the backup services and wait to see whether the script stops them again later tonight
- C. Disable shadow copies on all other servers so the attacker gains nothing by deleting them later
- D. Isolate the server, kill the script's process tree and check other hosts for the same activity

**Answer:** D

**Explanation:** Deleting shadow copies and stopping backups are classic pre-encryption steps, so isolating the host, stopping the process and hunting for the same behaviour elsewhere can prevent the ransomware stage. Restarting services and waiting gives the attacker time, a fresh backup from an attacker-controlled host may be compromised and slows containment, and disabling shadow copies elsewhere helps the attacker.

### Question 5 of 10

Hunters look for command-and-control beaconing in proxy logs. Many legitimate applications, such as update agents, also call home at regular intervals. Which method BEST separates likely beacons from benign periodic traffic?

- A. Exclude all connections over port 443, since encrypted traffic cannot be analysed for timing patterns without decryption
- B. Score connection regularity per host and destination, then filter out destinations contacted by most of the fleet
- C. Review the top 100 destinations by total bytes transferred, because beacons usually move large amounts of data
- D. Alert on every destination contacted at an exact interval of 60 seconds, because malware typically uses one-minute sleep timers

**Answer:** B

**Explanation:** Beacons show low-variance timing to a destination that few hosts contact, so combining regularity scoring with fleet prevalence removes common update services. Fixed-interval rules miss jittered beacons, timing analysis works on encrypted traffic metadata, and beacons are typically small, so byte volume points to backups and downloads instead.

### Question 6 of 10

Hunters suspect DNS tunneling from a small number of hosts. The DNS resolver logs are available in the SIEM. Which analytic is MOST likely to surface tunneling with a manageable false-positive rate?

- A. Alert on every DNS query whose response is a TXT record, since tunneling tools always use TXT records to move their data
- B. Per-host query volume to one parent domain with long, high-entropy subdomains, excluding known CDN and vendor domains
- C. Flag any host that performs more than 1,000 DNS lookups per hour, because normal workstations rarely exceed that query rate
- D. List all domains that were first seen in the environment today and send each one to an analyst for manual reputation review

**Answer:** B

**Explanation:** Tunneling shows up as many long, random-looking subdomains under one parent domain from the same host, and excluding legitimate high-entropy services such as CDNs and security vendors keeps the noise down. TXT records are used by many legitimate services and tunnels can use other record types, first-seen domains are far too numerous to review manually, and raw query volume flags busy but benign hosts.

### Question 7 of 10

A scanner reports 4,000 critical-CVSS findings across the estate. The patch team can remediate about 200 per week. Which prioritisation approach BEST reduces real risk first?

- A. Patch the hosts with the largest number of findings first, since they contribute the most to the total open vulnerability count
- B. Rank by known exploitation (CISA KEV), exploit-likelihood scores, internet exposure and asset criticality
- C. Remediate the oldest findings first, because vulnerabilities that have been open the longest are the most likely to be exploited
- D. Remediate strictly in descending CVSS base score order, starting with every 10.0

**Answer:** B

**Explanation:** Combining evidence of active exploitation and exploit likelihood with exposure and business criticality focuses limited capacity on the vulnerabilities attackers will actually use against important assets. CVSS base scores describe severity but not likelihood, raw counts per host can be dominated by low-impact findings, and age alone says nothing about whether exploits exist.

### Question 8 of 10 (choose 2)

A cloud threat-detection service alerts that an EC2 instance's role credentials are being used from an external IP. The instance runs a web application with a server-side request forgery bug reported last week. Which TWO actions BEST address the root cause and the immediate risk? (Choose two.)

- A. Enforce IMDSv2 on the instance and fix the SSRF flaw so credentials cannot be fetched through the application
- B. Move the instance to a private subnet without changing anything else in the application or the metadata settings
- C. Attach an additional deny policy to the users group in the account so human users cannot assume the role
- D. Revoke the role's active sessions issued before now so the stolen credentials stop working
- E. Rotate the SSH key pair of the instance so the attacker can no longer log in to the operating system

**Answer:** A, D

**Explanation:** Revoking sessions issued before the current time invalidates temporary credentials already stolen, and requiring IMDSv2 plus fixing the SSRF closes the path by which the application leaked them. SSH keys are unrelated to credentials exfiltrated via the metadata service, moving subnets does not stop SSRF through the application, and restricting human users does not affect the instance role being abused.

### Question 9 of 10

An alert shows that cloud audit logging was disabled in a production account at 02:14 by an admin role, and re-enabled at 02:51. No change ticket exists. What should the SOC do?

- A. Treat it as a routine maintenance action: re-enabled logging shows the admin finished planned work and corrected the setting afterwards
- B. Treat it as a likely attacker evasion: investigate the role's other actions and use alternative logs to reconstruct the 37-minute gap
- C. Treat it as a logging-service outage: open a support case with the provider and wait for their incident report before investigating
- D. Treat it as an audit finding only: record it for the quarterly compliance review, since logging is already re-enabled now

**Answer:** B

**Explanation:** Disabling audit logs without a change is a classic defence-evasion step, so the role's activity must be investigated and the gap reconstructed from other sources such as flow logs, service logs or provider-side events. Assuming maintenance ignores the missing ticket, a provider outage would not appear as a user API action, and deferring to a compliance review leaves a possible intrusion uninvestigated.

### Question 10 of 10

During a major incident the CIO asks for status every 30 minutes, which keeps pulling the lead analyst away from the investigation. What is the BEST way to handle stakeholder communication?

- A. Assign a communications lead who issues scheduled updates, so the technical lead stays on the investigation
- B. Ask the CIO to stop requesting updates until the investigation is finished, so that the analysts can concentrate on the incident
- C. Give the CIO direct access to the analysts' working chat channel so they can follow every finding in real time
- D. Send status only when there is a confirmed finding, which keeps updates accurate and avoids repeated messages with no news

**Answer:** A

**Explanation:** Separating the communications role from the technical lead is a core incident-command practice: leaders get predictable updates and responders keep their focus. Refusing updates damages trust and decision-making, raw analyst chat contains unverified hypotheses that can mislead executives, and silence until a finding arrives leaves stakeholders guessing during long investigations.

Ready for a timed run? Take the free 15-question SOD mini exam: https://devopscodelab.com/app/exams/security-operations-decisions/mini/

DevOpsCode Lab is not affiliated with General. SOD and related names are trademarks of their owners. We write every question ourselves; these are not real exam questions.
