Which statement correctly distinguishes scalability from elasticity?
Hard · Choose one
Explanation
Answer: A. Scalability is a capability: the architecture can grow to handle more work, whether by adding instances horizontally or by moving to larger instances vertically. Elasticity is the dynamic behavior layered on top: capacity is added automatically when demand rises and, crucially, removed when it falls. A system can be scalable without being elastic, for example if scaling requires a manual change.
B. Self-managed fleets can be made elastic with automatic scaling, and managed services vary in how they scale. The distinction is behavioral, not service-category based.
C. Both concepts apply across compute, storage, databases, and networking. Neither is restricted to one resource type.
D. Never failing and recovering after failure describe fault tolerance and high availability. Neither term is about handling changes in load.
02
An architecture runs two application servers behind a load balancer in two Availability Zones. If one server fails, the load balancer routes all traffic to the survivor and users see a brief error for in-flight requests before retrying successfully. Which term most precisely describes this behavior?
Hard · Choose one
Explanation
Answer: A. High availability accepts a small, bounded disruption while the system detects the failure and redirects traffic, and it keeps overall downtime minimal. That is exactly what happens here: in-flight requests fail, health checks remove the bad target, and service continues. Fault tolerance would be the stronger claim that no request was affected at all.
B. Elasticity is capacity changing in response to demand. Here capacity changed because of a failure, and the surviving fleet was not resized to match load.
C. Fault tolerance means the failure is absorbed with no impact on any request. Because in-flight requests errored and had to be retried, the stronger claim does not hold.
D. Durability concerns whether stored data survives over time. It is a storage property and does not describe the availability of the compute tier.
03
Amazon S3 storage classes are described as designed for 99.999999999 percent durability and for a separate, lower availability figure. Which statement explains why these two numbers differ?
Hard · Choose one
Explanation
Answer: C. The two measure different risks. Durability answers "will my object still exist and be intact years from now," and is raised by storing redundant copies across multiple devices and, in most storage classes, across multiple Availability Zones. Availability answers "can I reach the service right now," and is affected by transient conditions such as maintenance or network events. Data can be perfectly durable while temporarily unreachable, which is why the availability figure is always the lower of the two.
A. Both are published design targets, and availability is additionally backed by a service level agreement. The difference is what each one measures, not how binding it is.
B. Both figures describe the stored data and the service that fronts it. Neither is a statement about the encryption or reliability of data while it is moving over the network.
D. Neither figure is defined by counting Regions against Availability Zones. Redundancy contributes to both numbers; the distinction remains preservation versus reachability.
04
A CIO states that the main value of moving to AWS is "agility." A colleague argues that the main value is "elasticity." Which statement best clarifies the difference between the two claims?
Hard · Choose one
Explanation
Answer: A. Agility is an organizational outcome: because resources appear in minutes and failed experiments cost only the hours they ran, the business can try more ideas and ship more often. Elasticity is a technical behavior: running capacity grows and shrinks with load. Elasticity contributes to agility, but a team can have elastic infrastructure and still release slowly if its processes are the bottleneck.
B. Neither is a security measure, and elasticity is not primarily financial, although it has strong cost consequences.
C. They are related but distinct. One describes how fast the organization can move; the other describes how running capacity responds to load.
D. Both apply in either environment. Production benefits from agility through faster, safer releases, and non-production benefits from elasticity by scaling to zero when idle.
05
Which TWO statements about the AWS Cloud value proposition are TRUE? (Choose TWO.)
Hard · Select 2 answers
Explanation
Answer: A and B. Two of AWS's stated advantages are captured here. Increased speed and agility means new resources are minutes away, so the cost and time of experimenting drop dramatically and unsuccessful ideas can be abandoned cheaply. Stopping guessing capacity means you access as much or as little capacity as you need and scale up and down with only a few minutes' notice, rather than committing to a size before deployment.
C. A workload lifted unchanged onto oversized, always-on instances can easily cost more than it did on premises. Savings follow from rightsizing, elasticity, and the use of managed services.
D. AWS is responsible for the availability of its services. Whether the customer's application stays up depends on how the customer architects and operates it.
E. Security is a shared responsibility. AWS secures the infrastructure, but data classification, access control, and encryption choices remain with the customer.
06
A company that already runs on AWS is deciding whether to build its own custom message queue on Amazon EC2 or to consume a managed queuing service. Which reasoning best reflects the AWS Cloud value proposition?
Hard · Choose one
Explanation
Answer: A. A message queue is plumbing: customers do not choose the company's product because of it. AWS's guidance is to stop spending money and effort on undifferentiated heavy lifting and to focus on the projects that differentiate the business. Consuming the managed service also transfers patching, scaling, and availability engineering to AWS.
B. Owning code carries ongoing cost in maintenance, on-call burden, security patching, and scaling work. Those costs usually exceed the service charge for commodity infrastructure.
C. Running on the same infrastructure does not make the two equivalent. They differ sharply in who carries the operational and availability burden.
D. Managed services integrate with identity, encryption, and logging controls, and are frequently easier to secure well than a hand-built equivalent.
07
Which TWO characteristics are features of a cloud consumption model that a traditional on-premises model lacks? (Choose TWO.)
Hard · Select 2 answers
Explanation
Answer: B and E. The two defining shifts are self-service on-demand provisioning, which removes the procurement and installation cycle from the critical path, and consumption-based charging, which replaces the up-front purchase of fixed capacity with a bill that tracks what actually ran. Together they are what AWS means by on-demand delivery of IT resources with pay-as-you-go pricing.
A. Refresh planning is an on-premises obligation. On AWS the underlying hardware is refreshed by AWS, and customers migrate to newer instance generations when it suits them.
C. A multi-year purchase that fixes environment size is the traditional model. It is the constraint the cloud model removes.
D. Commissioning a local facility is the on-premises path. On AWS, a new geography is reached by deploying into an existing Region.
08
Which TWO of the following are design principles of the AWS Well-Architected security pillar? (Choose TWO.)
Hard · Select 2 answers
Explanation
Answer: C and D. Keep people away from data and maintain traceability are both security pillar design principles. Keeping people away from data means using tools and automation to remove the need for direct human access to sensitive data, which reduces mishandling and human error. Maintaining traceability means monitoring, alerting, and auditing actions and changes in real time, and integrating logs and metrics with systems that can investigate and act automatically.
A. Considering mechanical sympathy is a performance efficiency pillar design principle.
B. Adopting a consumption model is a cost optimization pillar design principle.
E. Scaling horizontally to increase aggregate workload availability is a reliability pillar design principle.
09
A workload must satisfy a strict recovery time objective, and the team is deciding whether to invest in automated failover or in reducing the monthly bill. Which statement best reflects how the Well-Architected Framework treats this situation?
Hard · Choose one
Explanation
Answer: A. The framework is a set of perspectives, not a ranking. Strengthening one pillar frequently costs something against another, and the framework exists to make those trade-offs explicit and deliberate rather than accidental. The right balance is driven by the workload's own business requirements, such as the stated recovery time objective and what an outage actually costs.
B. Cost optimization means delivering business value at the lowest price point, not minimizing spend regardless of the consequences.
C. Reliability is critical for many workloads, but not all. A short-lived internal batch job does not warrant the same investment as a payment system.
D. There is no pass mark and no requirement for equal scores. A review identifies risks and produces an improvement plan the business can prioritize.
10
A retailer's architecture stores every raw image upload indefinitely in the highest-performance storage tier, keeps three permanently running copies of a batch environment that is used for two hours per night, and applies no lifecycle rules. Which TWO Well-Architected pillars are most directly violated? (Choose TWO.)
Hard · Select 2 answers
Explanation
Answer: A and D. Paying premium storage rates for data that is never accessed, and running batch capacity 24 hours a day for a two-hour job, directly contradicts cost optimization's consumption model and its call to measure overall efficiency. The same waste has an environmental dimension: sustainability's maximize utilization principle targets idle resources and unnecessary storage precisely because they consume energy for no productive output.
B. Operational excellence concerns how the workload is run and improved. Poor storage tiering and idle capacity are cost and efficiency failures rather than procedural ones, although better operational practice would surface them.
C. Nothing in the description weakens identity, encryption, traceability, or network protection. The architecture is wasteful rather than insecure.
E. Over-provisioning does not reduce reliability. If anything the extra capacity is harmless to availability; the problem is that it is unnecessary.