Why would a traffic log list an application as "not-applicable”?
A. The firewall denied the traffic before the application match could be performed.
B. The TCP connection terminated without identifying any application data
C. There was not enough application data after the TCP connection was established
D. The application is not a known Palo Alto Networks App-ID.
Explanation: traffic log would list an application as “not-applicable” if the firewall denied the traffic before the application match could be performed. This can happen if the traffic matches a security rule that is set to deny based on any parameter other than the application, such as source, destination, port, service, etc1. In this case, the firewall does not inspect the application data and discards the traffic, resulting in a “not-applicable” entry in the application field of the traffic log1.
An engineer is tasked with deploying SSL Forward Proxy decryption for their organization. What should they review with their leadership before implementation?
A. Browser-supported cipher documentation
B. Cipher documentation supported by the endpoint operating system
C. URL risk-based category distinctions
D. Legal compliance regulations and acceptable usage policies
Explanation: The engineer should review the legal compliance regulations and acceptable usage policies with their leadership before implementing SSL Forward Proxy decryption for their organization. SSL Forward Proxy decryption allows the firewall to decrypt and inspect the traffic from internal users to external servers. This can raise privacy and legal concerns for the users and the organization. Therefore, the engineer should ensure that the leadership is aware of the implications and benefits of SSL Forward Proxy decryption and that they have a clear policy for informing and obtaining consent from the users. Option A is incorrect because browser-supported cipher documentation is not relevant for SSL Forward Proxy decryption. The firewall uses its own cipher suite to negotiate encryption with the external server, regardless of the browser settings. Option B is incorrect because cipher documentation supported by the endpoint operating system is not relevant for SSL Forward Proxy decryption. The firewall uses its own cipher suite to negotiate encryption with the external server, regardless of the endpoint operating system. Option C is incorrect because URL risk-based category distinctions are not relevant for SSL Forward Proxy decryption. The firewall can decrypt and inspect traffic based on any URL category, not just risk-based ones.
Which two key exchange algorithms consume the most resources when decrypting SSL traffic? (Choose two.)
A. ECDSA
B. ECDHE
C. RSA
D. DHE
Explanation:
The two key exchange algorithms that consume the most resources when decrypting SSL
traffic are ECDHE and DHE. These are both Diffie-Hellman based algorithms that enable
perfect forward secrecy (PFS), which means that they generate a new and unique session
key for each SSL/TLS session, and do not reuse any previous keys. This enhances the
security of the encrypted communication, but also increases the computational cost and
complexity of the key exchange process. ECDHE stands for Elliptic Curve Diffie-Hellman
Ephemeral, which uses elliptic curve cryptography (ECC) to generate the session key. DHE
stands for Diffie-Hellman Ephemeral, which uses modular arithmetic to generate the
session key. Both ECDHE and DHE require more CPU and memory resources than RSA,
which is a non-PFS algorithm that uses public and private keys to encrypt and decrypt the
session key.
References:
Key Exchange Algorithms, Best Practices for Enabling SSL
Decryption, PCNSE Study Guide (page 60)
An auditor is evaluating the configuration of Panorama and notices a discrepancy between the Panorama template and the local firewall configuration. When overriding the firewall configuration pushed from Panorama, what should you consider?
A. The firewall template will show that it is out of sync within Panorama.
B. The modification will not be visible in Panorama.
C. Only Panorama can revert the override.
D. Panorama will update the template with the overridden value.
Explanation:
When managing firewalls with Panorama, configurations can be pushed from
Panorama templates to managed firewalls. However, there are scenarios where specific
settings need to be overridden on the local firewall level due to unique requirements or
exceptions.
B. The modification will not be visible in Panorama:
When an override is made directly on the firewall, this change is not automatically
reflected back in Panorama's templates or device groups. The local configuration
on the firewall will take precedence over the Panorama pushed configuration for
the overridden settings, but these local changes will not be visible in the Panorama
interface. This means that while Panorama maintains central control and visibility
over the bulk of the configuration, it does not have visibility into local overrides
made directly on the firewalls.
This distinction is crucial for auditors and administrators to understand, as it impacts how
configurations are managed and synchronized between Panorama and the individual
firewalls. Local overrides provide flexibility but require careful management to ensure
consistency and compliance with security policies.
Which two policy components are required to block traffic in real time using a dynamic user group (DUG)? (Choose two.)
A. A Deny policy for the tagged traffic
B. An Allow policy for the initial traffic
C. A Decryption policy to decrypt the traffic and see the tag
D. A Deny policy with the "tag" App-ID to block the tagged traffic
Explanation: Use the dynamic user group in a policy to regulate traffic for the members of the group. You will need to configure at least two rules: one to allow initial traffic to populate the dynamic user group and one to deny traffic for the activity you want to prevent (in this case, questionable-activity). To tag users, the rule to allow traffic must have a higher rule number in your rule base than the rule that denies traffic.
A remote administrator needs access to the firewall on an untrust interface. Which three options would you configure on an interface Management profile to secure management access? (Choose three)
A. HTTPS
B. SSH
C. Permitted IP Addresses
D. HTTP
E. User-IO
Exhibit.
An organization has Palo Alto Networks NGFWs that send logs to remote monitoring and
security management platforms The network team has reported excessive traffic on the
corporate WAN How could the Palo Alto Networks NGFW administrator reduce WAN traffic
while maintaining support for all the existing monitoring/security platforms?
A. Any configuration on an M-500 would address the insufficient bandwidth concerns
B. Forward logs from external sources to Panorama for correlation, and from Panorama send them to the NGFW
C. Configure log compression and optimization features on all remote firewalls
D. Forward logs from firewalls only to Panorama and have Panorama forward logs to other external services
Refer to the exhibit.
Which will be the egress interface if the traffic's ingress interface is ethernet1/7 sourcing
from 192.168.111.3 and to the destination 10.46.41.113?
A. ethernet1/6
B. ethernet1/3
C. ethernet1/7
D. ethernet1/5
Explanation: In the second image, VW ports mentioned are 1/5 and 1/7. Hence it can not be a part of any other routing. So if any traffic coming as ingress from 1/7, it has to go out via 1/5. The egress interface for the traffic with ingress interface ethernet1/7, source 192.168.111.3, and destination 10.46.41.113 will be ethernet1/5. This is because the traffic will match the virtual wire with interfaces ethernet1/5 and ethernet1/7, which is configured to allow VLAN-tagged traffic with tags 10 and 201. The traffic will also match the security policy rule that allows traffic from zone Trust to zone Untrust, which are assigned to ethernet1/7 and ethernet1/5 respectively2. Therefore, the traffic will be forwarded to the same interface from which it was received, which is ethernet1/53.
Page 2 out of 37 Pages |
Previous |