Mixed practice exam 5

Explanation

Answer: A. On-Demand is the baseline EC2 pricing model: capacity is billed as it is consumed, with no upfront fee and no term commitment, so the customer can start and stop instances freely. AWS bills EC2 usage in one-second increments with a 60-second minimum. The trade-off is that On-Demand carries the highest per-hour rate of the shared-tenancy pricing models, higher than Spot, Reserved Instances, and Savings Plans, which is why AWS positions it for short-term, spiky, or unpredictable workloads that cannot be interrupted.

• B. Spot Instances have no term commitment, but they are not simply pay-as-you-go: they draw on spare EC2 capacity and AWS can reclaim the instance with a two-minute warning, so they are not the option AWS describes as uncommitted pay-as-you-go capacity.
• C. Reserved Instances are a billing commitment to a consistent amount of usage for a 1-year or 3-year term. They lower the rate but are the opposite of 'no long-term commitment'.
• D. Dedicated Hosts allocate an entire physical server to one customer and are billed per host. They can be purchased On-Demand, but the defining characteristic is single-tenant hardware, not uncommitted per-second pricing.

Explanation

Answer: A. Spot Instances use spare Amazon EC2 capacity and AWS advertises savings of up to 90% off the On-Demand price. The discount is that large because the capacity is not guaranteed: when AWS needs the capacity back it reclaims the instance after a two-minute interruption notice. That risk is the reason Spot is priced below every committed-use option.

• B. Dedicated Instances address tenancy requirements by running on hardware dedicated to a single customer account. They carry a price premium over shared tenancy rather than a discount.
• C. Standard Reserved Instances provide the largest discount of the Reserved Instance types, but the maximum AWS advertises for committed-use compute is up to 72%, not 90%.
• D. Compute Savings Plans provide savings of up to 66% off On-Demand in exchange for a 1-year or 3-year hourly spend commitment. That is a substantial discount, but it is not the deepest one AWS offers.

Explanation

Answer: C. A Savings Plan is a commitment to spend a chosen dollar amount per hour on eligible usage for a fixed term, which is 1 or 3 years for Compute, EC2 Instance, and SageMaker AI Savings Plans and 1 year for Database Savings Plans. Usage up to that hourly commitment is charged at the discounted Savings Plans rate; anything above it is charged at On-Demand rates. Expressing the commitment in dollars per hour rather than in instance hours is what gives Savings Plans their flexibility, because the same commitment can be met by different instance types, sizes, and in the case of Compute Savings Plans, different services and Regions.

• A. Prepaid credits are a separate billing construct. Compute, EC2 Instance, and SageMaker AI Savings Plans offer All Upfront, Partial Upfront, and No Upfront payment options, and Database Savings Plans are No Upfront, but in every case the commitment itself is a rate of spend per hour rather than a credit balance.
• B. Committing to a specific instance type for a set number of hours describes the older Reserved Instance model, not Savings Plans. Savings Plans deliberately avoid tying the commitment to a single instance type.
• D. AWS does not offer a discount expressed as a percentage of the total bill. Savings Plans discounts apply only to the eligible compute or database usage that the plan covers.

Explanation

Answer: B. Spot Instances can be reclaimed by AWS at any time with a two-minute notice, so the deciding question is always whether the workload tolerates interruption. A checkpointed batch job does: if an instance is reclaimed, the job resumes from the last checkpoint on new capacity, and the workload captures the discount without risking data loss. Fault-tolerant, stateless, flexible, and time-insensitive workloads are the classic Spot fit.

• A. A predictable workload running continuously for three years is the textbook case for a Savings Plan or a 3-year Reserved Instance, which convert that predictability into a committed-use discount.
• C. Physical core visibility is a tenancy and licensing requirement met by Dedicated Hosts, which is an independent dimension from Spot pricing.
• D. A run that must complete in one uninterrupted window is exactly what Spot cannot guarantee, because AWS may reclaim the capacity mid-run. On-Demand or a committed-use option is appropriate here.

Explanation

Answer: B. A Dedicated Host is a physical EC2 server allocated entirely to one customer, and it exposes the sockets, cores, and host ID. That visibility is what makes it possible to satisfy licenses that are counted per socket, per core, or per VM, such as Windows Server, SQL Server, SUSE Linux Enterprise Server, and Red Hat Enterprise Linux licenses bound to physical hardware. Dedicated Hosts are billed per host rather than per instance.

• A. An On-Demand Capacity Reservation reserves EC2 capacity in a specific Availability Zone. It says nothing about tenancy or hardware visibility, so it does not address a per-socket licensing requirement.
• C. A zonal Reserved Instance is a billing and capacity construct. It does not change instance tenancy or expose the underlying physical hardware.
• D. Dedicated Instances also run on single-tenant hardware, but they give the customer no visibility into sockets, cores, or host ID, so they cannot satisfy licenses that must be counted against physical hardware. Their BYOL support is limited to scenarios such as SQL Server with License Mobility.

Explanation

Answer: B. AWS states that Amazon EC2 usage is billed in one-second increments with a 60-second minimum, and this applies across the purchasing options. Fine-grained billing matters for cloud economics because it means short-lived and elastic workloads pay only for the time they actually consume, which is a core part of the pay-as-you-go value proposition.

• A. There is no five-minute minimum for EC2. The documented minimum is 60 seconds.
• C. Hourly rounding was the original EC2 billing model, but AWS moved to per-second billing. A question keyed to hourly rounding reflects out-of-date pricing behavior.
• D. AWS does not bill EC2 compute per day. Daily granularity appears in cost reporting, not in how the usage itself is metered.

Explanation

Answer: C. Convertible Reserved Instances can be exchanged for another Convertible Reserved Instance with different attributes, such as a different instance family, operating system, or tenancy, as long as the exchange results in a reservation of equal or greater value. That flexibility is bought at the cost of a smaller discount than a Standard Reserved Instance provides. Convertible is the correct choice whenever the workload's shape is expected to change within the term.

• A. Zonal versus regional describes the scope of a Reserved Instance, which controls capacity reservation and flexibility of application. Scope is a separate attribute from whether the reservation is Standard or Convertible.
• B. Regional scope gives Availability Zone flexibility and, under certain conditions, instance size flexibility, but it does not allow the reservation to be exchanged for a different instance family.
• D. Standard Reserved Instances provide the deepest Reserved Instance discount but explicitly cannot be exchanged. They can be modified within limits, for example changing Availability Zone or scope, but not exchanged for a different instance family.

Explanation

Answer: B. A Reserved Instance purchased with zonal scope reserves capacity in the Availability Zone specified, in addition to providing the billing discount. This is the only one of the listed options that does both. The trade-off is rigidity: a zonal Reserved Instance applies only to the exact instance family, size, and Availability Zone specified, with no size or Availability Zone flexibility.

• A. A regional Reserved Instance provides a billing discount and applies across any Availability Zone in the Region, but AWS documentation is explicit that it does not reserve capacity. This is one of the most commonly missed distinctions on the exam.
• C. Savings Plans of every type are purely billing constructs. A Compute Savings Plan lowers the rate on eligible EC2, AWS Fargate, and AWS Lambda usage but never reserves capacity.
• D. An EC2 Instance Savings Plan commits to a specific instance family in a specific Region for a lower rate, but like all Savings Plans it provides no capacity reservation.

Explanation

Answer: D. Instance size flexibility lets a regional Reserved Instance discount apply across sizes in the same family using normalization factors, so a reservation for one size can cover several smaller instances or part of a larger one. AWS limits this benefit to Amazon Linux/Unix Reserved Instances with default tenancy. Reservations for other platforms, such as Windows, or for dedicated tenancy do not receive size flexibility.

• A. Zonal scope removes flexibility rather than granting it. A zonal Reserved Instance applies only to the exact family, size, and Availability Zone specified.
• B. The payment option affects only how much of the cost is paid in advance and therefore the size of the discount. It has no bearing on whether the discount flexes across instance sizes.
• C. Convertible Reserved Instances allow the reservation to be exchanged for different attributes, which is a different mechanism from automatic size flexibility within a family.

Explanation

Answer: C. EC2 Instance Savings Plans commit to a specific instance family in a specific Region and, in exchange for that narrower scope, offer the deepest Savings Plans discount at up to 72% off On-Demand. Within that family and Region the plan still flexes across instance size, operating system, and tenancy, which is enough flexibility for a workload whose family and Region are fixed.

• A. Compute Savings Plans are the most flexible type, automatically applying across EC2, Lambda, and Fargate usage regardless of instance family, size, Region, operating system, or tenancy. That flexibility costs discount depth, capping at up to 66%.
• B. SageMaker AI Savings Plans apply to Amazon SageMaker AI machine learning instance usage and would not cover this company's general-purpose EC2 workload at all.
• D. Database Savings Plans apply to managed database services such as Amazon Aurora, Amazon RDS, and Amazon DynamoDB. They do not cover EC2 compute.