Domain 3: Cloud Platform and Infrastructure Security Flashcards

Question

A component failure in the primary HVAC system leads to a high temperature alarm in the datacenter that Kim manages. After resolving the issue, what should Kim consider to prevent future issues like this? A. A closed loop chiller B. Redundant cooling systems C. Swamp coolers D. Relocating the datacenter to a colder climate

Answer 1

**Answer: B. Redundant cooling systems** **A well-designed datacenter should have redundant systems and capabilities for each critical part of its infrastructure.** That means that power, cooling, and network connectivity should all be redundant. Kim should determine how to ensure that a single system failure cannot take her datacenter offline.

Answer 2

**Answer: C. Take a snapshot of the compromised system and use that for the investigation.** Michael should conduct his investigation, but there is a pressing business need to bring the website back online. The most reasonable course of action would be to take a snapshot of the compromised system and use the snapshot for the investigation, restoring the website to operation as quickly as possible while using the results of the investigation to improve the security of the site.

Answer 3

**Answer: D. No access** **The principle of least privilege should guide Joe in this case. He should apply no access permissions by default and then give each user the necessary permissions to perform their job responsibilities.** Read only, editor, and administrator permissions may be necessary for one or more of these users, but those permissions should be assigned based on business need and not by default.

Answer 4

**Answer: B. Hardware security module (HSM)** The question describes an HSM. KMB is a nonsense term used as a distractor, so it is incorrect. TGT is a term associated with Kerberos single sign-on systems and is incorrect. The TCB includes the elements of hardware and software (usually in the operating system) that ensure that a system can only be controlled by those with the proper permissions (i.e., admins with root control), so it is also incorrect.

Answer 5

**Answer: B. API** **Application programming interfaces (APIs) allow for the programmatic interaction with services and platforms.** Jason can use APIs to tie together different technologies and interact with them programmatically. Python scripts may play a role in that automation but they do not, on their own, allow the automation to occur because the script must use the API to interact with services. Cloud access security brokers (CASBs) enforce customer security policies across a variety of services and are not a backend security solution. The hypervisor is a component of the management plane and would be a target of the automation, rather than a facilitator of it.

Answer 6

**Answer: B. Block** **Block storage, also known as volume storage, provides disk volumes for use by servers.** Cloud environments generally do not provide dedicated disks because that approach would be highly inefficient. Object storage is used to store files as individual objects and cannot be used as a disk volume. Any type of storage may be encrypted or unencrypted, but the fact that storage is encrypted does not make it useful for disk volumes.

Answer 7

**Answer: D. CEO** The key to successfully answering this question is noticing that it asks who bears “ultimate responsibility.” The chief executive officer (CEO) bears ultimate responsibility for the success of the organization and, therefore, will be the one held accountable if the business fails. Of course, everyone in the organization has some accountability for their own roles in the program’s success.

Answer 8

**Answer: B. Ensuring that redundant cabling/connectivity enters the facility from different sides of the building/property** **To avoid a situation where severing a given physical connection results in severing its backup as well (such as construction/landscaping, etc.), have redundant lines enter on different sides of the building.** For health and human safety, multiple egress points from each facility are preferred (and often required by law); option A is incorrect. Emergency lighting should receive power regardless of their proximity to the power source, and parking vehicles near generators is a bad idea from a safety perspective; option C is incorrect. Not all facilities need to withstand earthquakes; this may be true of datacenters in California, but not in Sydney, so it is not an industry-wide best practice. Option D is incorrect.

Answer 9

**Answer: B. Hypervisor** The hypervisor is responsible for coordinating access to physical hardware and enforcing isolation between different virtual machines running on the same physical platform.

Answer 10

**Answer: B. Remove the service.** Running unnecessary services on a server increases the attack surface and exposes an organization to unnecessary risk. Therefore, Ursula should work through the organization’s normal change management processes to remove the service.

Answer 11

**Answer: C. VLAN** **Virtual local area networks (VLANs) are used to create logical separation between systems in a datacenter and are the most cost-effective way to provide network segmentation.** Creating separate LANs would require redundant equipment and unnecessary expense. Virtual private networks (VPNs) are used to connect remote users and sites over an insecure network and are not relevant within a secure datacenter. Border Gateway Protocol (BGP) is used to route traffic between network sites and is not relevant to this scenario.

Answer 12

**Answer: D. Opportunity** **The only reason organizations accept any level of risk is because of the potential benefit also afforded by a risky activity.** Profit is not the hallmark of every opportunity (or every organization— many organizations are nonprofit or government-based), so option A is incorrect. Likewise, not all risky activities offer a chance to enhance performance, so option B is incorrect. Cost is not a benefit, so that doesn’t even make sense in the context of the question; option C is not correct and a distractor.

Answer 13

**Answer: C. 99.741%** **Tier 2 datacenters are expected to achieve 99.741% availability.** Tier 1 datacenters are expected to achieve 99.671% availability. Tier 3 datacenters are expected to achieve 99.982% availability. Tier 4 datacenters are expected to achieve 99.995% availability.

Answer 14

**Answer: C. Central processing units (CPUs); random access memory (RAM)** The compute nodes of a cloud datacenter can be measured in terms of how many central processing units (CPUs) and how much random access memory (RAM) is available within the center.

Answer 15

**Answer: D. Virtual machines are stored as snapshotted files when not in use.** **VMs are snapshotted and simply stored as files when they are not being used; an attacker who gains access to those file stores could ostensibly steal entire machines in highly portable, easily copied formats. Therefore, these cloud storage spaces must include a significant amount of controls.** It is possible to sanitize cloud file spaces by using crypto-shredding. Virtualization does not prevent the use of application-based security controls. Administrators are free to encrypt cloud-based file stores.

Answer 16

**Answer: B. Recovery point objective (RPO) ** **The recovery point objective (RPO) is a measure of data that can be lost in an outage without irreparably damaging the organization**. Data replication strategies will most affect this metric, as the choice of strategy will determine how much recent data is available for recovery purposes. Recovery time objective (RTO) is a measure of how long an organization can endure an out-age without irreparable harm. This may be affected by the replication strategy, but not as much as the RPO. Option A is incorrect. The maximum allowable downtime (MAD) is how long an organization can suffer an outage before ceasing to be an organization. This is not dependent on the RPO, and the data replication strategy won’t have much effect on it at all. Option C is incorrect. The mean time to failure (MTTF) is a measure of how long an asset is expected to last (usually hardware), as determined by the manufacturer/vendor. The data replication strategy will have no bearing on this whatsoever. Option D is incorrect.

Answer 17

**Answer: A. SQL injection** SQL injection vulnerabilities allow an attacker to send commands through a web application to the database supporting that application. Cross-site scripting attacks execute code on a remote user’s system. Cross-site request forgery and server-side request forgery attacks seek to exploit trust relationships by tricking systems into authorizing unauthorized activity.

Answer 18

**Answer: B. TLS** Transport Layer Security (TLS) is the primary protocol used to implement the HTTPS standard for secure communication between servers and users. Virtual private networks (VPNs) will also encrypt web traffic, but they are a separate service normally provided by a user’s employer and not the remote website. Virtual local area networks (VLANs) are used for network segmentation and not for encryption. IPsec is a protocol used to implement VPNs.

Answer 19

**Answer: B. The BC/DR provider becomes the new normal production environment** Theoretically, all the options are possibly true. However, option B is the most likely to occur; the cost and risk of moving operations from one environment/provider to another is sizable, so staying with the secondary provider (making them the new primary) is a good way to reduce some of the risk involved in returning to normal.

Answer 20

**Answer: B. Least privilege** Gary should follow the least privilege principle and assign users only the permissions they need to perform their job responsibilities. Aggregation is a term used to describe the unintentional accumulation of privileges over time, also known as privilege creep. Separation of duties and separation of privileges are principles used to secure sensitive processes.

Answer 21

**Answer: B. Clearance and need to know** **Before granting access, Gary should verify that the user has a valid security clearance and a business need to know the information.** Gary is performing an authorization task, so he does not have to verify the user’s credentials, such as a password or biometric scan.

Answer 22

**Answer: A. Separation of duties** The matrix shown in the figure is known as a separation of duties matrix. It is used to ensure that one person does not obtain two privileges that would create a potential conflict. Aggregation is a term used to describe the unintentional accumulation of privileges over time, also known as privilege creep. Two-person control is used when two people must work together to perform a sensitive action. Defense in depth is a general security principle used to describe a philosophy of overlapping security controls.

Answer 23

**Answer: D. Two-person control** Gary should follow the principle of two-person control by requiring simultaneous action by two separate authorized individuals to gain access to the encryption keys. He should also apply the principles of least privilege and defense in depth, but these principles apply to all operations and are not specific to sensitive operations. Gary should avoid the security through obscurity principle, the reliance upon the secrecy of security mechanisms to provide security for a system or process. Two-person control: A security concept used to maintain high levels of safety and integrity by ensuring that two or more individuals are required to perform certain sensitive or high-risk tasks. This technique prevents a single individual from having sole authority or access and is often applied to prevent fraudulent activities or unauthorized actions.

Answer 24

**Answer: A, B, C.** **Privileged access reviews are one of the most critical components of an organization’s security program because they ensure that only authorized users have access to perform the most sensitive operations. They should take place whenever a user with privileged access leaves the organization or changes roles as well as on a regular, recurring basis.** However, it is not reasonable to expect that these time-consuming reviews would take place on a daily basis.

Answer 25

**Answer: A. Type 1** **Type 1, or “bare-metal” hypervisors, run directly on top of hardware and provide a greater degree of security than Type 2 hypervisors.** This is because Type 2 hypervisors must run on top of another operating system, increasing the total attack surface. Types 3 and 4 hypervisors do not exist.

Answer 26

**Answer: A. Reduced cost** **The major factor driving organizations to lease space in a colocation facility is a reduction in cost achieved through economies of scale.** Leased facilities are not necessarily more, less secure or complex than custom-built facilities, and they do not necessarily have greater capability.

Answer 27

**Answer: B. Dual-power supplies in systems.** Tier 1 datacenters require dedicated space for IT systems, an uninterruptible power supply (UPS) system for line conditioning and backup purposes, sufficient cooling systems to serve all critical equipment, and a power generator for extended electrical outages, with at least 12 hours of fuel to run the generator at sufficient load to power the IT systems. Dual-power supplies are a requirement of a Tier 3 datacenter and are not required in Tier 1 datacenters.

Answer 28

**Answer: B. Digital certificates** **The most important detail in this question is that the access must be automated. This means that systems will connect to each other without any human intervention.** Because of this requirement, biometric controls are not useful because they require that a person be involved in the authentication process. Using passwords would require storing that password on the remote server. This is possible, but not ideal from a security perspective. Since we don’t want to use passwords or biometrics, multifactor authentication is also not feasible. That leaves digital certificates as the most viable option for securing these connections.

Answer 29

**Answer: ** **The NBI usually handles traffic between the SDN controllers and SDN applications.** Options A and C are incorrect because neither of those options lists any of the SDN infrastructure, be that the controllers or the applications. Option D may be arguably correct, as there might be an NBI handling that traffic between those nodes, but option B is more specific and always true for this definition, so it is the better choice.

Answer 30

**Answer: C. Identity federation** This is a very popular function of federated identity. Single sign-on (SSO) is similar to federation, but it is limited to a single organization; federation is basically SSO across multiple organizations. Option A is incorrect. Options B and D are threats listed in the Open Web Application Security Project (OWASP)Top 10; they are incorrect. Federated identity management (FIM) involves linking identities across multiple domains, allowing users to access resources using a single set of credentials. It's a key aspect of Identity and Access Management (IAM) in the cloud, enabling single sign-on (SSO) functionality across different organizations or systems.

Answer 31

**Answer: B. Hardware security module (HSM)** Hardware security modules (HSMs) are security solutions designed to manage the processes surrounding cryptographic keys. Key management boxes (KMBs) provide for the management of physical keys. Ticket-granting tickets (TGTs) are a component of the Kerberos authentication process. The trusted computing base (TCB) provides a secure operating environment inside a computer system.

Answer 32

**Answer: C. Management** Sprawl needs to be addressed from a managerial perspective because it is caused by allowed user actions (usually in a completely authorized capacity).

Answer 33

**Answer: B. Task a moderator well-versed in BC/DR actions to supervise and present scenarios to the participants, including randomized special events.** A trained and experienced moderator can guide the participants through the activity, enhancing their training and noting pitfalls and areas for improvement. Option A is not preferable because having the participants gathered together ensures their full attention and provides interaction that remote participation might not yield. Option C is a baseline; all participants should have copies of the policy as a matter of course. Option D is not useful in a tabletop exercise; only critical participants in the organization should take part in the tabletop.

Answer 34

**Answer: C. Whether necessary security controls are in place and functioning properly.** The baseline will contain the suite of security controls applied uniformly throughout the environment. A VM image audit is unlikely to involve any form of physical security; A is incorrect. Baselines won’t predictively show malicious activity; B is incorrect. Baselines also do not have anything to do with user training and awareness; option D is incorrect

Answer 35

**Answer: B. Good cryptographic key management** This is a difficult question that requires a great deal of thought. Option B is correct because appropriate cloud data security practices will require encrypting a great deal of the data, and having the keys will be necessary during contingency operations in order to access the backup; without the keys, you won’t be able to access your data. Option A is not correct because using the cloud for BC/DR will allow personnel to access the backup from anywhere they can get broadband connectivity, not specifically a recovery site. Option C is not correct because the customer will rarely have physical access to servers in the cloud environment. Option D is not correct because forensic analysis is not a significant consideration in BC/DR; it is much more important for incident response.

Answer 36

**Answer: A. Remote access for a customer to racked devices in the datacenter; electrical utilities; connectivity to an internet service provider (ISP)/the internet** Ping is a term used to describe the ability of customers to access their systems remotely. Power is shorthand for electrical power to the systems. Pipe refers to the network connectivity that supports servers’ connections to the internet.

Answer 37

**Answer: B. Multitenancy** Sharing resources with other, unknown customers (some of whom may be competitors of or even hostile to the organization) is a risk not faced by organizations that maintain their own, on-premises datacenters. All the other answers are threats that exist in both environments and are therefore incorrect.

Answer 38

**Answer: D. Deliver streaming media content in an efficient manner by placing it closer to the end user.** **Software-Defined Networks (SDN) allow administrators to perform a variety of automated functions. These include rerouting traffic based on current customer demand, creating logical subnets without having to change any physical connections, and filtering access to resources based upon specific rules or settings.** Delivering streaming media content in an efficient manner by placing it closer to the end user is a function of content delivery networks (CDNs), not SDNs.

Answer 39

**Answer: A. Failover cluster** **The illustration shows an example of a failover cluster, where DB1 and DB2 are both configured as database servers. At any given time, only one will function as the active database server, while the other remains ready to assume responsibility if the first one fails.** Although the environment may use UPS, tape backup, and cold sites as disaster recovery and business continuity controls, they are not shown in the diagram.

Answer 40

**Answer: B. Avoiding vendor lock-in/lockout** **Having an additional backup with a different provider means that if your primary provider becomes unusable for any reason (including bankruptcy or unfavorable contract terms), your data is not held hostage or lost.** Custom VMs may or may not work in a new environment; this is actually a risk when porting data out of the production environment; Option A is incorrect. Performance probably will not increase if data is replicated to another cloud provider; in fact, you will probably lose some load-balancing capability you might have had if you kept the data and backups together. Option C is incorrect. Having two providers will always be more costly than a single provider; option D is incorrect.

Answer 41

**Answer: B. Limiting access to sensitive information** The “sensitive information,” in this case, is whatever knowledge of the datacenter’s security controls and processes might be gathered by physically visiting the datacenter. Even though a cloud customer cannot get access to the facility, this also means that other cloud customers (some of whom may be hostile to another customer’s interests) also will not have access, so none would have an advantage over the other(s). Option A is incorrect because qualified personnel are still required whether or not a cloud environment has limited access to their datacenter. In fact, security may be degraded by having unqualified personnel rather than qualified personnel working in the cloud datacenter. Option C is incorrect because reducing jurisdictional exposure does not enhance security. There may be a correlation between ensuring statutory compliance and enhancing security as it applies to limiting access to the cloud datacenter. However, option B is a better answer because it is certainly true. Therefore, option D is not the best answer to the question.

Answer 42

**Answer: C. Object storage** Virtual server instances and containers provide direct computing resources to cloud service users. **Function as a service provides a platform upon which computing may be performed. Object storage does not provide any compute capability, as it is solely a storage service.**

Answer 43

**Answer: C. With VMs, the cloud provider does not have to deploy an entire hardware device for every new customer.** While options A and B are both also true, C is the most significant reason cloud datacenters use VMs. If the cloud provider had to purchase a new box for every user, the cost of cloud services would be as much as running a traditional environment (or likely cost even more), and there would be no reason for any organization to migrate to the cloud, especially considering the risks associated with disclosing data to a third party. Option D is simply untrue. VMs are not easier to operate than actual devices.

Answer 44

**Answer: A. Full test** **Full tests, also known as full interruption tests, shut down the primary operating facility and shift operations to the backup facility.** These tests are very likely to have a serious impact on production operations. The parallel test activates the backup facility but does not move production responsibility to it. Walkthroughs and simulations do not activate the backup facility or impact production operations in any way.

Answer 45

**Answer: A. Escape** This is an example of an escape attack because the attacker was able to leave the confines of their own virtual machine and access resources belonging to another customer. There is no indication in the scenario that the attack used any specific overflow, injection, or scripting vulnerability.

Answer 46

**Answer: B. Hypervisor** **Escape attacks always occur as the result of a vulnerability or malfunction in the hypervisor because the hypervisor is responsible for performing the separation that prevents one customer from accessing resources belonging to another customer.**

Answer 47

**Answer: D. Object** **Disk volumes used to support virtual machines are typically stored on block storage.** However, when snapshotting is used to create backups of those disks, the backups are commonly stored in less expensive object storage. Dedicated disks are not generally used in cloud environments. The backups may be encrypted while in object storage, but this is not a technical requirement.

Answer 48

**Answer: D. Need for a physical hot site/warm site** **Having your data backed up and accessible in the cloud eliminates any need for having a distinct hot site/warm site separate from your primary operating environment; instead, your personnel can recover operations from anywhere with a good broadband connection.** Cloud BC/DR capability does not remove the necessity of security personnel and appropriate policies; both options A and B are incorrect. Option C makes no sense as an answer to the question. It is unclear how you can cut costs by eliminating your old access credentials. In fact, it is difficult to imagine how that is a true statement. Therefore, option C is a poor choice, and option D is the best choice.

Answer 49

**Answer: B. Auditing** A specified configuration built to defined standards and with a controlled process can be used to demonstrate that all VMs within an environment include certain controls; this can greatly enhance the efficiency of an audit process. The VM’s image has very little to do with physical security or training; options A and C are incorrect. Baseline images are the opposite of customization; option D is incorrect.

Answer 50

**Answer: D. Packet capture** Log entries do provide some insight into user activity but they generally do not provide the full context of user communication. Netflow records only provide the “telephone bill” level detail of communications and not the content. While those sources would be useful, full packet capture provides the most accurate reconstruction of user activity, but it is costly to implement due to data storage requirements.

Answer 51

**Answer: C. Use another cloud provider for the BC/DR backup.** **It’s best to have your backup at another cloud provider in case whatever causes an interruption in service occurs throughout your primary provider’s environment; this will be more complicated and expensive, but it provides the best redundancy and resiliency.** Using the same provider for production and backup is not a bad option, but it entails the risk of the same contingency affecting both copies of your data. Having either the backup or the production environment localized does not provide the best protection because you no longer operate secure cloud datacenters.

Answer 52

**Answer: C. Copies of the laws/regulations/standards governing specific elements of the plan.** This is not an easy question, because every plan/policy should include mention of the governance documents that drive the formation of the plan/policy; however, these can be included by reference only— you don’t need to include full copies of these governance documents. All the other options should be included in the BC/DR plan/policy.

Answer 53

**Answer: A. Extensible Markup Language (XML)** Security Assertion Markup Language (SAML) is based on XML. HTTP is used for port 80 web traffic; HTML is used to present web pages. ASCII is the universal alphanumeric character set.

Answer 54

**Answer: A. Regulatory compliance.** **Depending on your industry and the nature of your data, moving information into another jurisdiction may affect or invalidate your regulatory compliance.** Cloud providers, wherever they are located, should compensate for environmental and physical security factors, so this should have no impact on your potential risk; options B and C are incorrect. Option D is incorrect because it is a blanket statement that is not always true. In fact, for some organizations, the physical location where their data is stored can have serious regulatory consequences.

Answer 55

**Answer: C. Some risks cannot be transferred to a cloud provider** **Under current legal frameworks, some risks (such as legal liability for privacy data breaches) cannot be transferred to a contracted party, so the data owners (that is, cloud customers) will still retain those risks. It is important to note that customers are always responsible for managing risk in some way, even if risk is transferred to a cloud provider.** Option A is not correct; risks can and should be mitigated, even in the cloud. Option B is not correct; cloud migration will require some risk acceptance, but that is true for everything except avoided risk. Option D is incorrect; cloud providers can choose not to offer services or not to accept certain clients.

Answer 56

**Answer: C. Maximum allowable downtime** The administrative offices of a cloud datacenter rarely are part of the critical functions of the operation; a datacenter could likely endure the loss of the administrative offices for a considerable length of time, so redundancy here is probably not cost-effective.

Answer 57

**Answer: B. Recovery point objective** **The recovery point objective (RPO) is a measure of data that can be lost in an outage without irreparably damaging the organization. Data replication strategies will most affect this metric, as the choice of strategy will determine how much recent data is available for recovery purposes.** Recovery time objective (RTO) is a measure of how long an organization can endure an outage without irreparable harm. This may be affected by the replication strategy, but not as much as the RPO. Option A is incorrect. The maximum allowable downtime (MAD) is how long an organization can suffer an outage before ceasing to be an organization. This is not dependent on the RPO, and the data replication strategy won’t have much effect on it at all. Option C is incorrect. The mean time to failure (MTTF) is a measure of how long an asset is expected to last (usually hardware), as determined by the manufacturer/vendor. The data replication strategy will have no bearing on this whatsoever. Option D is incorrect.

Answer 58

**Answer: ** **Authorization is the process of granting users and other security principals access to resources in an environment.** Identification and authentication are part of the overall identity and access management (IAM) process, as is authorization, but they do not specifically describe granting access to resources. Federation is a means of conducting IAM across organizations; authorization is a more specific answer, so federation is incorrect.

Answer 59

**Answer: B. Platform as a service (PaaS)** **When using two different cloud providers, a cloud customer runs the risk that data/software formats used in the operational environment can’t be readily adapted to the other provider’s service, thus causing delays during an actual failover.** Risks of physical intrusion are neither obviated nor enhanced by choosing to use two cloud providers; option A is incorrect. Using a different cloud provider for backup/archiving actually reduces the risks of outages due to vendor lock-in/lockout and natural disasters, so options C and D are not correct.

Answer 60

**Answer: C. Vendor lock-in/lockout** As the models increase in level of abstraction and service, the customer’s control over the environment decreases. Therefore, the customer has the most control over the configuration of IaaS services, a moderate degree of control over PaaS/FaaS services, and the least control over SaaS services.

Answer 61

**Answer: C. Perform a full-scale test.** **Without a full test, Warren can’t be sure the BC/DR plan/process will work the way it is intended.** Audits are good, but they will not demonstrate actual performance the way a test will, so options A and B are incorrect. It is important that the BC/DR capacity and performance be included in the contract, but that will not truly ensure that the functionality exists; a test is required, so option D is incorrect.

Answer 62

**Answer: C. Returning to normal operations too soon.** **A premature return to normal operations can jeopardize not only production, but personnel; if the contingency that caused the BC/DR action is not fully completed/addressed, there may still be danger remaining.** The BC/DR plan/process should take into account both the absence of essential personnel and telecommunications capabilities, so options A and D are incorrect. Option B does present a serious problem for the organization, but option C is still a greater risk, so B is incorrect.

Answer 63

**Answer: C. Wet pipe** **Systems that use water always pose a greater failure risk to electronic equipment than those that use gas because water can destroy equipment.** Of the systems listed, wet pipe systems pose the greatest risk because water is always present in the pipes.

Answer 64

**Answer: B. Private cloud** **Guest escape (a malicious user leaving the confines of a VM and able to access other VMs on the same machine) is less likely to occur and to have a significant impact in an environment provisioned for and used by a single customer.** In a public cloud, this is more likely and would be more significant, so option A is incorrect. The service model doesn’t specifically dictate the likelihood of occurrence or impact (both PaaS and IaaS could be in a private or public cloud, which is the more important factor), so both options C and D are incorrect.

Answer 65

**Answer: A. Prohibitions on port scanning and penetration testing ** **Many cloud providers restrict activities that are common for administrative and security purposes but can also be construed/used for hacking; this includes port scanning and penetration testing. These restrictions can reduce the customer’s ability to perform basic security functions. Customers should review these practices with service providers and confirm that they are allowed to conduct routine security activities.** While geographical dispersion of cloud assets might make securing those assets more difficult in the notional sense (customer administrators can’t physically visit the devices that host their data), remoteness does not necessarily inhibit good security practices, which can be performed at a remove. This is not as detrimental as rules against port scanning/pen testing, so option B is incorrect. There are no rules against user training or laws against securing your own assets, in the cloud or otherwise; options C and D are incorrect.

Answer 66

**Answer: B. Software-defined network (SDN)** **The question describes a Software-Defined Network (SDN). A VPN is used for creating an encrypted communications tunnel over an untrusted medium, so option A is incorrect.** ACLs are used as centralized repositories for identification, authentication, and authorization purposes, so option C is incorrect. RBAC is an access control model used to assign permissions based on job functions within an organization, so option D is incorrect.

Answer 67

**Answer: C. Getting physical access can be difficult.** **Cloud providers may be reluctant to grant physical access, even to their customers, on the assumption that allowing access would disclose information about security controls.** In some cases, cloud customers won’t even know the location(s) of the datacenter(s) where their data is stored. The other options are all untrue. Data in the cloud and controls in the cloud can most certainly be audited. So, options A and B are incorrect. D is untrue; there are regulators for all industries, including those that operate in the cloud.

Answer 68

**Answer: A. Business drivers** The business requirements will determine the crucial aspects of BC/DR. All the other options may constitute some input that will influence the BC/DR, but they are not the prevailing factors, and so are incorrect.

Answer 69

**Answer: C. Physical location of the launch point** **Because cloud access is remote access, pen tests will be remote tests; it doesn’t really matter what the physical origin of the simulated attack is.** Cloud providers will want notice before the customer launches the test, and that notice should include a description of the scope of the test, knowledge of the timeframe for the test, and the logical (not physical!) addresses of the testing systems.

Answer 70

**Answer: D. Checklist review** **The checklist review is the least disruptive type of disaster recovery test. During a checklist review, team members each review the contents of their disaster recovery checklists on their own and suggest any necessary changes.** During a tabletop exercise, team members come together and walk through a scenario without making any changes to information systems. During a parallel test, the team actually activates the disaster recovery site for testing, but the primary site remains operational. During a full interruption test, the team takes down the primary site and confirms that the disaster recovery site is capable of handling regular operations. The full interruption test is the most thorough test but also the most disruptive.

Answer 71

**Answer: D. Integrity** DDoS prevents all these things except for data integrity. DDoS only prevents communication; it does not usually result in modified data.

Answer 72

**Answer: C. Access control list (ACL) of authorized personnel** Health and human safety is a paramount goal of security; all facilities must have multiple emergency egress points. All the other options are distractors as they are included in option C.

Answer 73

**Answer: B. Within the datacenter** **Because VMs don’t take updates when they are not in use (snapshotted and saved as image files) and updates may be pushed while the VMs are saved, it’s important to ensure that they receive updates when they are next instantiated.** Systems may be configured to perform automatic updates. A physical tracking mechanism won’t be of much aid for virtual devices because they aren’t physically stolen like hardware boxes, so option A is incorrect. Having an ACL in the image baseline would create a situation where every user from every cloud customer could access every VM in the datacenter; option C is incorrect. Write protection is used in forensic analysis of machines (virtual or otherwise); it would not be useful in an operational baseline. Option D is incorrect.

Answer 74

**Answer: ** **Performing live deception and trickery against employees of the cloud provider (or its suppliers/vendors) could be construed as unethical and possibly illegal, especially without their knowledge and/or consent. Social engineering probably won’t be involved in penetration tests run by customers.** All the other options are legitimate activities a customer might perform during a penetration test (with provider permission).

Answer 75

**Answer: D. Recover quickly from minor incidents** Having the backup within the same environment can allow easy rollback to a last known good state or to reinstantiate clean VM images after minor incidents (e.g., a malware infection in certain VMs). Ease of compliance will not be determined by the location of the backup, so option A is incorrect. Traveling should not be a major cost for cloud usage; option B is incorrect.

Answer 76

**Answer: C. Log file generation** Event logging is essential for incident management and resolution; this can be set as an automated function of the CM tools. Not all systems need or can utilize biometrics; option A is incorrect. Usually, tampering refers to physical intrusion of a device; since the question is about VMs, it is probably not applicable. Option B is incorrect. Hackback is illegal in many jurisdictions; option D is incorrect.