Skip to content

SAP-C03 · Domain 3: Cost-Optimized Architecture Design · 15% of the exam

Task 3.1: Design architectures based on AWS pricing models.

Paying the right way: commitments, capacity reservations, Spot and On-Demand matched to usage, architectures that cost less (Graviton, Express workflows, Lambda), rightsizing and scaling from metrics, and AI/ML cost levers such as service tiers, batching and caching.

Study it

  • Pricing models: Savings Plans, Reserved Instances, Spot, capacity reservations and rightsizing

    Partly covered by: AWS Cost Levers

  • Cheaper architectures and AI/ML cost levers: Graviton, Express workflows, batching and caching

    Lesson coming

Sample questions

Try each one before opening the answer. Every option is explained, with the AWS documentation page that proves it.

Question 1 · choose 1

A FinOps team works from a delegated account and wants rightsizing recommendations for Amazon EC2 instances, Auto Scaling groups, EBS volumes, Lambda functions and Amazon ECS services on AWS Fargate across all 90 accounts in the organization. EC2 recommendations must take memory use into account. Which approach meets these requirements?

  1. AUse AWS Compute Optimizer for the organization with a delegated administrator and the CloudWatch agent for memory
  2. BUse the rightsizing recommendations in AWS Cost Explorer from the management account for all member accounts
  3. CReview the AWS Trusted Advisor low utilization check for Amazon EC2 instances in each of the 90 accounts
  4. DTurn on Amazon S3 Storage Lens advanced metrics for the organization and export the metrics to the FinOps account
Show the answer and why
  • AUse AWS Compute Optimizer for the organization with a delegated administrator and the CloudWatch agent for memory

    Correct

    Compute Optimizer gives recommendations for these resource types across an organization, a delegated administrator can view them, and memory utilization from the CloudWatch agent improves EC2 recommendations.

  • BUse the rightsizing recommendations in AWS Cost Explorer from the management account for all member accounts

    Incorrect

    Cost Explorer rightsizing recommendations cover EC2 instances only, not Auto Scaling groups, EBS volumes, Lambda functions or ECS services.

  • CReview the AWS Trusted Advisor low utilization check for Amazon EC2 instances in each of the 90 accounts

    Incorrect

    The low utilization check looks at Amazon EC2 instances only, so it cannot cover Auto Scaling groups, EBS volumes, Lambda functions or ECS services.

  • DTurn on Amazon S3 Storage Lens advanced metrics for the organization and export the metrics to the FinOps account

    Incorrect

    S3 Storage Lens reports on S3 storage usage and activity. It does not cover compute resources.

Rightsizing across compute types and accounts is AWS Compute Optimizer, enabled for the organization; memory needs the CloudWatch agent.

Question 2 · choose 1

A FinOps team wants to compare several Savings Plans commitments, such as a recommended amount and a target coverage percentage, against its last 60 days of usage before buying. Which feature fits?

  1. AThe EC2 instance quotas in Service Quotas
  2. BSavings Plans Purchase Analyzer
  3. CAWS Budgets with a Savings Plans utilization budget
  4. DCost Anomaly Detection monitors
Show the answer and why
  • AThe EC2 instance quotas in Service Quotas

    Incorrect

    Quotas limit how much can be launched, not what a commitment saves.

  • BSavings Plans Purchase Analyzer

    Correct

    Purchase Analyzer runs analyses with a chosen commitment, term, payment option and lookback period based on historical usage.

  • CAWS Budgets with a Savings Plans utilization budget

    Incorrect

    Utilization budgets track plans already bought.

  • DCost Anomaly Detection monitors

    Incorrect

    Anomaly detection watches spend for unusual changes.

Modeling candidate commitments before buying is the Purchase Analyzer.

Question 3 · choose 1

A new platform will have a steady baseline of compute around the clock. In its first three years the team expects to move parts of it from EC2 instances to AWS Fargate and Lambda, change instance families as new generations appear, and possibly move workloads between Regions. Which pricing commitment fits the baseline?

  1. AEC2 Instance Savings Plans for the instance family that the platform uses at launch
  2. BStandard Reserved Instances for the instance types in the launch design
  3. CCompute Savings Plans sized to the steady baseline of hourly compute spend
  4. DSpot Instances for the baseline, with On-Demand Instances as a fallback
Show the answer and why
  • AEC2 Instance Savings Plans for the instance family that the platform uses at launch

    Incorrect

    EC2 Instance Savings Plans apply to one instance family in one Region, and they do not cover Fargate or Lambda.

  • BStandard Reserved Instances for the instance types in the launch design

    Incorrect

    Standard Reserved Instances are tied to instance attributes and do not apply to Fargate or Lambda usage.

  • CCompute Savings Plans sized to the steady baseline of hourly compute spend

    Correct

    Compute Savings Plans apply to EC2 usage regardless of instance family, size, operating system, tenancy or Region, and also to Fargate and Lambda usage.

  • DSpot Instances for the baseline, with On-Demand Instances as a fallback

    Incorrect

    Spot Instances can be interrupted, so they do not suit a baseline that must run around the clock, and they are not a commitment-based discount.

The more the platform will change, the more flexible the commitment must be: Compute Savings Plans follow usage across families, Regions and compute services.

Question 4 · choose 1

A new Amazon EMR cluster runs nightly Spark jobs that can tolerate lost workers and still finish by morning. The team wants to cut compute cost without risking the cluster's HDFS data. Which purchasing plan fits?

  1. AOn-Demand Instances for every node in the cluster
  2. BSpot Instances for the primary node
  3. COn-Demand primary and core nodes, Spot task nodes
  4. DSpot Instances for all core nodes to save the most
Show the answer and why
  • AOn-Demand Instances for every node in the cluster

    Incorrect

    All On-Demand pays full price for work that tolerates interruptions.

  • BSpot Instances for the primary node

    Incorrect

    Losing the primary node would end the cluster.

  • COn-Demand primary and core nodes, Spot task nodes

    Correct

    Task nodes do not store data, so they can use Spot Instances at a reduced cost while core nodes keep HDFS data.

  • DSpot Instances for all core nodes to save the most

    Incorrect

    Core nodes store HDFS data, which could be lost if Spot capacity is reclaimed.

On EMR, run task nodes on Spot and keep data-holding nodes On-Demand.

Question 5 · choose 1

A risk analytics grid of 40 nodes in one Availability Zone runs on previous-generation EC2 instances that do not use enhanced networking. The nodes exchange millions of small TCP messages, and monitoring shows the network hitting a packets-per-second ceiling with high and uneven latency between nodes, while CPU stays low. The team must raise packet throughput and lower node-to-node latency without changing the TCP-based application, without paying any extra charge for a networking feature, and with roughly the same instance count and hourly spend. Which change meets these requirements?

  1. ALaunch the existing instance types into a cluster placement group so that the nodes sit close together in the Availability Zone
  2. BPut an AWS Global Accelerator accelerator in front of the nodes so that their traffic travels over the AWS global network
  3. CAttach a second elastic network interface to every node and split the grid's messages across the two interfaces
  4. DMove to current-generation instance types of similar size that support ENA, and make sure enhanced networking is turned on
Show the answer and why
  • ALaunch the existing instance types into a cluster placement group so that the nodes sit close together in the Availability Zone

    Incorrect

    Cluster placement groups suit low-latency traffic between instances, but AWS notes that the lowest latency and the highest packets per second also require an instance type that supports enhanced networking, which these instances lack.

  • BPut an AWS Global Accelerator accelerator in front of the nodes so that their traffic travels over the AWS global network

    Incorrect

    Global Accelerator improves the path from internet clients to applications through AWS edge locations and adds hourly and data transfer charges. It does nothing for traffic between nodes in one zone.

  • CAttach a second elastic network interface to every node and split the grid's messages across the two interfaces

    Incorrect

    Multiple network interfaces serve needs such as management networks, network appliances and dual-homed instances. They do not give these instances enhanced networking or lift their packets-per-second ceiling.

  • DMove to current-generation instance types of similar size that support ENA, and make sure enhanced networking is turned on

    Correct

    Enhanced networking provides higher bandwidth, higher packets per second and consistently lower latency between instances, at no additional charge. All Nitro-based instances use ENA for it, and the change needs no application code changes.

The constraints are more packets per second, lower latency, no application change and no extra networking charge. A cluster placement group helps only together with enhanced networking, Global Accelerator serves internet clients at a cost, and extra interfaces do not change the instance's network performance. Current-generation instances with ENA meet every constraint.

Question 6 · choose 1

A trading firm runs 60 EC2 instances around the clock in us-east-1a, next to its exchange connection, and expects to keep them for three years. It needs capacity guaranteed in that Availability Zone during market spikes, wants a commitment discount on the same usage, and plans to move to a newer instance family halfway through the term. Which purchasing option meets these requirements?

  1. AZonal Convertible Reserved Instances in us-east-1a, exchanged for the new instance family when the firm upgrades
  2. BZonal Standard Reserved Instances in us-east-1a, modified to the new instance family when the firm upgrades
  3. CA three-year Compute Savings Plan sized to the instances' hourly spend
  4. DRegional Convertible Reserved Instances for us-east-1, exchanged for the new instance family when the firm upgrades
Show the answer and why
  • AZonal Convertible Reserved Instances in us-east-1a, exchanged for the new instance family when the firm upgrades

    Correct

    A zonal Reserved Instance reserves capacity in its Availability Zone, and a Convertible Reserved Instance can be exchanged during the term for one with a different instance family.

  • BZonal Standard Reserved Instances in us-east-1a, modified to the new instance family when the firm upgrades

    Incorrect

    Zonal Standard Reserved Instances reserve capacity, but Standard Reserved Instances cannot be exchanged, so a move to another instance family is not possible within the term.

  • CA three-year Compute Savings Plan sized to the instances' hourly spend

    Incorrect

    A Compute Savings Plan follows a change of instance family, but Savings Plans give a discount only; they do not reserve capacity in an Availability Zone.

  • DRegional Convertible Reserved Instances for us-east-1, exchanged for the new instance family when the firm upgrades

    Incorrect

    Regional Reserved Instances apply the discount in any zone of the Region but do not reserve capacity.

Only zonal Reserved Instances combine a discount with a capacity reservation, and only Convertible Reserved Instances can be exchanged for another instance family during the term.

Question 7 · choose 1

A telemetry platform runs 150 Kinesis data streams in one account, all in On-demand Standard mode. Traffic per stream varies widely during the day, ingest and retrieval charges dominate the bill, and the fixed per-stream charge adds up across so many streams. The team does not want to plan shard capacity or manage scaling. Which change lowers the cost while meeting these requirements?

  1. ASwitch every stream to provisioned mode with enough shards for its peak traffic
  2. BReplace the streams with Amazon SQS standard queues, which have no per-stream charge
  3. CTurn on On-demand Advantage mode for the account
  4. DReduce the data retention period of every stream from seven days to the default 24 hours
Show the answer and why
  • ASwitch every stream to provisioned mode with enough shards for its peak traffic

    Incorrect

    Provisioned mode requires the team to choose and manage shard counts, and sizing each stream for its peak would leave capacity idle most of the day.

  • BReplace the streams with Amazon SQS standard queues, which have no per-stream charge

    Incorrect

    SQS is a different service without Kinesis features such as ordered shards, replay from retention and multiple consumers reading the same records, so this is a redesign rather than a pricing change.

  • CTurn on On-demand Advantage mode for the account

    Correct

    On-demand Advantage is an account-level mode for on-demand streams that drops the fixed per-stream charge and lowers data ingest, retrieval and extended retention prices by at least 60% compared with On-demand Standard, with no capacity planning. Its commitment of at least 25 MiB/s of ingest and of retrieval across the on-demand streams is far below what a bill dominated by data charges implies.

  • DReduce the data retention period of every stream from seven days to the default 24 hours

    Incorrect

    Shorter retention removes extended retention charges only. It does not lower the ingest, retrieval and per-stream charges that dominate the bill.

Kinesis Data Streams offers On-demand Standard, On-demand Advantage and Provisioned modes. For many variable on-demand streams, On-demand Advantage keeps automatic scaling and removes the per-stream charge at lower data prices. The mode commits the account to at least 25 MiB/s of data ingest and 25 MiB/s of data retrieval across its on-demand streams, charging any shortfall at the discounted rate, and it must stay on for at least 24 hours, so it pays off for accounts with steady usage near or above that level, such as this one.

Practice domain 3 →Practice all domains →