CSRP — Cyber Security Responder & Practitioner
Lesson 9: Firewall Architecture and Rule Design
Question Bank
Cyber School · BDO Ziv Haft
QUESTION BANK — Lesson 9
Firewall Architecture and Rule Design
20 ISC2 CC-style multiple-choice questions. Used by the L9 lesson quiz (random-10), the Module 2 test (pooled), and the final exam (pooled). Each question has correct-answer rationale + distractor rationales.
Difficulty key: ★ easy · ★★ medium · ★★★ hard Topic tags: [DEF] = definition · [STATE] = stateless/stateful · [ANATOMY] = rule anatomy · [ORDER] = rule order · [DEFAULT] = default-deny · [DMZ] = DMZ pattern · [MISTAKE] = common mistakes · [LOG] = logging · [APPLY] = applied scenario
Q1 — [DEF] · ★
Which statement BEST describes a firewall?
A. A device that scans network traffic for known malware signatures and removes infected files. B. A network policy enforcement point that permits or denies traffic based on configured rules. C. An anti-tampering control that verifies the integrity of operating system files. D. A redundant power supply that ensures network availability during outages.
Correct: B. – B is correct: a firewall is fundamentally a policy enforcement device. – A describes antivirus / content filtering — not a firewall’s primary function. – C describes file integrity monitoring (FIM) — covered in Lesson 1. – D describes a UPS — unrelated to firewalls.
Q2 — [STATE] · ★
A stateless firewall makes its decisions based on:
A. The state of the TCP connection table. B. Each individual packet’s header fields, with no memory of prior packets. C. The user’s authentication status. D. The application-layer payload.
Correct: B. – B is correct: stateless = per-packet decision with no state tracking. – A describes stateful firewalls. – C describes identity-aware firewalls / NGFW user-ID. – D describes deep packet inspection (DPI) — a separate feature.
Q3 — [STATE] · ★★
Which of the following is a key advantage of a stateful firewall over a stateless firewall?
A. Stateful firewalls are always faster than stateless firewalls. B. Stateful firewalls automatically permit return traffic for established connections without an explicit reverse rule. C. Stateful firewalls do not require any configuration. D. Stateful firewalls eliminate the need for default-deny policies.
Correct: B. – B is correct: stateful tracks connections and lets return traffic back automatically. – A is false: stateless is often faster because there’s no connection-table lookup. – C is false: stateful firewalls still require configured rules. – D is false: default-deny is essential on both types.
Q4 — [ANATOMY] · ★
Which 5 fields define a typical firewall rule?
A. User, group, role, time, action. B. Source, destination, port, protocol, action. C. SHA-1, SHA-256, IP, MAC, action. D. URL, headers, cookies, certificate, action.
Correct: B. – B is the canonical 5-field anatomy. – A describes role-based access control (RBAC) attributes. – C mixes hashing algorithms with addressing — nonsensical. – D describes web application firewall (WAF) inspection at L7.
Q5 — [ANATOMY] · ★★
In the Cisco IOS ACL line access-list 100 permit tcp 10.10.20.0 0.0.0.255 host 10.10.50.10 eq 443, what does the 0.0.0.255 represent?
A. A subnet mask. B. A wildcard mask, where 0 means “must match” and 255 means “don’t care.” C. The destination IP. D. The source port range.
Correct: B. – B is correct: Cisco ACLs use inverted (wildcard) masks. – A is a common confusion — wildcard ≠ subnet mask. – C is wrong: destination is host 10.10.50.10. – D is wrong: source port is not specified in this line.
Q6 — [ORDER] · ★★
Most firewalls (including Cisco ACLs) process rules:
A. In random order; the firewall picks the most specific match. B. Top to bottom, first match wins; processing stops at the first matching rule. C. Bottom to top, last match wins. D. Alphabetically by rule name.
Correct: B. – B is correct: top-down, first-match is the universal behavior. – A is false: order is deterministic, not random. – C is false: processing is top-down, not bottom-up. – D is false: rule order is positional, not alphabetic.
Q7 — [ORDER] · ★★★
A security engineer notices that a permit rule allowing HTTPS from the user subnet to the intranet server never matches in the logs, even though users complain they cannot reach the intranet. What is the MOST likely cause?
A. The firewall has crashed. B. An earlier, broader deny rule is shadowing the permit rule — processing stops at the deny before reaching the permit. C. The intranet server is offline. D. The user subnet is using IPv6 instead of IPv4.
Correct: B. – B is correct: classic shadowing. – A is unlikely: other rules in the same ACL still match. – C is possible but the symptom would be different — connection would time out, but the rule WOULD register a hit on the permit line. – D is far-fetched and not implied by the question.
Q8 — [DEFAULT] · ★
Best practice in firewall policy design is:
A. Default-allow with an explicit blocklist of known-bad sources. B. Default-deny with an explicit allowlist of permitted traffic. C. No default; firewall picks the safest action per packet. D. Default-deny only for inbound; default-allow for outbound.
Correct: B. – B is correct: default-deny + allowlist is required by virtually every framework (NIST, PCI DSS, CIS). – A is risky: you cannot enumerate every bad source. – C is not how firewalls work. – D is incorrect: default-deny applies in both directions for sensitive environments.
Q9 — [DEFAULT] · ★★
On a Cisco IOS ACL, what happens to a packet that matches none of the explicit rules?
A. The packet is permitted by default. B. The packet is denied by the implicit deny at the end of every ACL. C. The packet is forwarded with no action logged. D. The router crashes.
Correct: B. – B is correct: Cisco ACLs have an implicit deny. – A is the OPPOSITE of Cisco’s behavior. – C contradicts the implicit deny. – D is absurd.
Q10 — [DEFAULT] · ★★
Why might a security engineer write an EXPLICIT deny ip any any log at the end of a Cisco ACL, even though there is already an implicit deny?
A. To override the implicit deny with a permit. B. To gain visibility — the explicit deny logs every blocked packet, while the implicit deny is silent. C. To make the ACL run faster. D. To comply with the requirement that every ACL have at least one rule.
Correct: B. – B is correct: implicit deny is silent; explicit deny … log provides visibility. – A is wrong: the explicit rule is a DENY, not a permit. – C is false: an additional rule doesn’t speed up processing. – D is fabricated.
Q11 — [DMZ] · ★
What is a DMZ in networking?
A. A subset of the internal network reserved for the CIO and executives. B. A network segment that sits between the internet and the internal network, hosting public-facing services. C. A wireless guest network. D. A backup data center.
Correct: B. – B is the standard definition. – A is a wrong association; DMZ is about exposure, not VIPs. – C describes a guest Wi-Fi VLAN. – D describes a disaster-recovery site.
Q12 — [DMZ] · ★★
In a well-designed DMZ architecture, which traffic flow is typically DENIED?
A. Internet → DMZ:443 (HTTPS to the public web server). B. Internal users → DMZ:443 (employees accessing the public website). C. DMZ → Internal database (web server reaching the internal database). D. Internal management → DMZ:22 (admins SSHing to the DMZ server).
Correct: C. – C is correct: DMZ → Internal is the critical defense; compromise of a DMZ host should not give attackers internal access. – A, B, and D are all typically allowed (with restrictions).
Q13 — [DMZ] · ★★★
Why is it CRITICAL that DMZ-to-internal traffic be tightly restricted?
A. The DMZ has lower bandwidth than the internal network. B. DMZ-hosted services are internet-facing and therefore more likely to be compromised; restricting DMZ-to-internal limits attacker lateral movement. C. DMZ servers use a different operating system than internal servers. D. The DMZ is owned by a third-party vendor.
Correct: B. – B captures the key reasoning: containment of compromised DMZ hosts. – A is irrelevant to security. – C is not necessarily true. – D is not necessarily true.
Q14 — [MISTAKE] · ★
Which of the following firewall rules is MOST dangerous?
A. permit tcp 10.10.20.0/24 host 10.10.50.10 eq 443 B. permit ip any any C. deny ip any any log D. permit tcp 10.10.10.0/24 any eq 22
Correct: B. – B is the infamous “any-any” rule — it allows all traffic from any source to any destination. – A is tightly scoped: source subnet, destination host, specific port. – C is the standard default-deny terminator — safe. – D allows SSH from the management subnet — broader than ideal but not catastrophic.
Q15 — [MISTAKE] · ★★
Which of the following management ports should NEVER be open to 0.0.0.0/0 (the public internet)?
A. TCP 443 (HTTPS). B. TCP 80 (HTTP). C. TCP 3389 (RDP). D. UDP 53 (DNS query).
Correct: C. – C: RDP is a primary ransomware initial-access vector and must never be exposed to the public internet. – A and B are public-facing service ports (websites). – D is the standard DNS port (clients query DNS over UDP/53 to public resolvers).
Q16 — [MISTAKE] · ★★★
A security audit finds that 30% of firewall rules have no documentation comments or change-ticket references. What is the BIGGEST risk this creates?
A. The firewall will run more slowly. B. Engineers cannot determine if a rule is still needed, leading to rule sprawl and potentially obsolete exposure. C. The firewall will be unable to log denies. D. Rule order will be randomized.
Correct: B. – B captures the real operational risk. – A: comments don’t affect runtime. – C: logging is independent of documentation. – D: rule order is unaffected by documentation.
Q17 — [LOG] · ★
Why is logging firewall denies important?
A. Denied packets must be retransmitted by the firewall. B. Denies provide intrusion detection visibility — every blocked attempt is signal about who is probing the network. C. The firewall vendor charges based on the number of denies. D. Without logs, the firewall cannot enforce rules.
Correct: B. – B is the core reason: deny logs are a primary input to SOC monitoring. – A is false: firewalls drop, they don’t retransmit. – C is fabricated. – D is false: logging is observation, not enforcement.
Q18 — [LOG] · ★★
A SOC analyst sees a sudden spike of denied TCP/22 (SSH) attempts from a single internet IP. What is the MOST appropriate initial response?
A. Open SSH (TCP/22) at the perimeter to see what the source wants. B. Investigate the source IP via threat-intel feeds; if confirmed malicious, block at the firewall and document the action. C. Ignore the spike; SSH brute-force is common. D. Shut down all internal SSH servers.
Correct: B. – B is the proportionate response: investigate, then act. – A is the OPPOSITE of correct — never open exposure to investigate. – C is wrong: even routine noise can correlate with targeted activity; ignore-by-default is bad practice. – D is disproportionate and breaks legitimate internal operations.
Q19 — [APPLY] · ★★★
A defender writes the following ACL on a perimeter router (top to bottom):
1. permit tcp any host 10.10.100.10 eq 443
2. permit ip 10.10.10.0/24 any
3. deny ip any 10.10.40.0/24
4. permit tcp 10.10.20.0/24 host 10.10.40.20 eq 443
5. deny ip any any log
Will rule 4 ever match traffic from 10.10.20.0/24 to 10.10.40.20:443?
A. Yes — rule 4 is specific and will match the relevant traffic. B. No — rule 3 denies all traffic to 10.10.40.0/24, and the firewall stops processing at the first match. C. Yes — but only if the destination is on port 443. D. No — the source 10.10.20.0/24 is invalid.
Correct: B. – B identifies the shadowing: rule 3 is a broad deny that prevents rule 4 from ever firing. – A misses the shadow. – C is wrong: even on port 443, rule 3 fires first. – D is incorrect: 10.10.20.0/24 is a valid private range.
Q20 — [APPLY] · ★★★
You discover the following rule on your perimeter firewall: permit tcp any host 10.10.40.10 eq 3389. The Management subnet (10.10.10.0/24) legitimately needs RDP access to 10.10.40.10. Which is the BEST fix?
A. Leave the rule unchanged; the Management subnet is included in any. B. Delete the rule entirely; RDP access should be removed permanently. C. Replace the rule with permit tcp 10.10.10.0/24 host 10.10.40.10 eq 3389 to scope the source to the Management subnet only. D. Change the destination port to 22 instead of 3389.
Correct: C. – C is the BEST fix: tighten the source to preserve legitimate access while closing public exposure. – A leaves the public-internet exposure intact — unacceptable. – B breaks legitimate Management access. – D changes the protocol entirely; doesn’t solve the underlying exposure problem and breaks the intended RDP function.
Bank summary
| Topic | Easy ★ | Medium ★★ | Hard ★★★ | Total |
| Definition | 1 (Q1) | — | — | 1 |
| Stateless/Stateful | 1 (Q2) | 1 (Q3) | — | 2 |
| Rule anatomy | 1 (Q4) | 1 (Q5) | — | 2 |
| Rule order | — | 1 (Q6) | 1 (Q7) | 2 |
| Default-deny | 1 (Q8) | 2 (Q9, Q10) | — | 3 |
| DMZ | 1 (Q11) | 1 (Q12) | 1 (Q13) | 3 |
| Common mistakes | 1 (Q14) | 1 (Q15) | 1 (Q16) | 3 |
| Logging | 1 (Q17) | 1 (Q18) | — | 2 |
| Applied scenarios | — | — | 2 (Q19, Q20) | 2 |
| Total | 7 | 8 | 5 | 20 |
Usage rules
- L9 lesson quiz: Randomly draw 10 from this bank per attempt.
- Module 2 test: This bank is pooled with L10–L20 banks. Module test draws 30 random questions, weighted to ensure all lessons are represented.
- CSRP final exam: This bank is pooled with all 40 lessons’ banks. Final exam draws 80 questions covering all 5 domains.
- Question reuse interval: A learner who fails the L9 quiz can retry; LearnDash shuffles which 10 of the 20 are shown.
Hebrew localization
A Hebrew translation of every question + every distractor + every rationale must be produced and reviewed by a native Hebrew speaker before publish. Hebrew tracking field: localization_he_complete: true/false.