Pair your devices with a code and playback position follows you: pause on this device, hit resume on the other. Position is saved to the site every minute and on pause.
Open this panel on your other device and enter the same code.
Starts this lesson and continues through 0 more to the end of certification prep.
The exam guide says distractors are "plausible responses that match the content area" and that they are what "a candidate with incomplete knowledge or skill might choose" (exam guide). So recognition, meaning "that one looks right", is exactly what the distractors are built to exploit.
Elimination turns it round. You don't hunt for the right answer. You prove three answers wrong, each for a named reason. That gives you three advantages:
Everything below is Method. The AWS facts used to eliminate options are quoted and sourced.
STEP 0 Read the LAST sentence first — what is being asked, one answer or several?
STEP 1 Read the stem. Write down (mentally): constraints (pass/fail) + the ONE optimiser.
─────────────────────────────────────────────────────────────────────────────────
PASS A IMPOSSIBLE — the option describes something AWS doesn't do or allow. ✗ out
PASS B FAILS A CONSTRAINT — works, but breaks a stated must/cannot/number. ✗ out
PASS C LOSES ON THE OPTIMISER — meets everything, but a survivor does it better. ✗ out
─────────────────────────────────────────────────────────────────────────────────
ONE LEFT → answer. TWO LEFT → line them up, read only the difference (lesson 2).
ZERO LEFT → you misread a constraint. Re-read the stem, not the options.
Two habits make it fast:
⚠️ All four questions below are original, written for this course from the task statements. They aren't from the AWS Official Practice Question Set and aren't calibrated against real exam items.
An e-commerce company's order service emits an event for every order. Three internal systems must receive every order event: fulfillment, fraud scoring, and a loader that feeds the analytics warehouse. The analytics loader is taken offline for maintenance for up to eight hours each week, and no order events can be lost for any system during that window. The systems are owned by different teams and must be able to fail independently. Which solution meets these requirements with the LEAST operational overhead?
- A. Publish each event to a single Amazon SQS standard queue. Configure all three systems to poll the queue.
- B. Publish each event to an Amazon SNS topic. Subscribe each system's HTTPS endpoint directly to the topic.
- C. Publish each event to an Amazon SNS topic. Subscribe one Amazon SQS queue per system to the topic, and have each system poll its own queue.
- D. Deploy an Apache Kafka cluster on self-managed Amazon EC2 instances across three Availability Zones. Give each system its own consumer group.
Step 1: constraints and optimiser. Constraints: every system gets every event. Nothing lost during an 8-hour outage of one consumer. Consumers fail independently. Optimiser: least operational overhead.
Pass A (impossible). Nothing here is impossible as described. Move on. That's normal, because many questions have no impossible option.
Pass B (fails a constraint).
Pass C (loses on the optimiser).
C ✓. SNS fanout is AWS's named pattern: "The Fanout scenario is when a message published to an SNS topic is replicated and pushed to multiple endpoints, such as … Amazon SQS queues" (SNS). Each queue holds messages for its own consumer, and retention is configurable "from 1 minute to 14 days. The default is 4 days" (SQS FAQ). That's comfortably longer than eight hours.
The one-word lesson: B and C differ by one phrase, "one Amazon SQS queue per system". That phrase is the durability.
A company runs a containerized nightly reconciliation job that takes about 40 minutes to complete. It currently runs on an Amazon EC2 instance that runs 24 hours a day but is idle outside the job. The company's existing scheduler can start any of the options below. The company wants to stop managing servers and wants the MOST cost-effective solution that requires no changes to the container image.
- A. Package the job as an AWS Lambda function and invoke it once a night.
- B. Run the container as an Amazon ECS task on AWS Fargate once a night.
- C. Keep the EC2 instance but move it to a larger instance type so the job finishes faster.
- D. Create an Amazon EKS cluster with a managed node group of EC2 instances and run the container as a scheduled Kubernetes job.
Step 1. Constraints: runs ~40 minutes. No server management. No change to the container image. Optimiser: most cost-effective. Note what's shared: "the company's existing scheduler can start any of the options". Scheduling isn't being tested, so ignore it.
Pass A (impossible).
Pass B (fails a constraint).
B ✓. Fargate "removes the need to choose server types, decide when to scale your clusters, or optimize cluster packing" (Fargate). You run the existing container image as a task, and it runs only when started.
The lesson: Pass A did the heavy lifting. One hard number in the stem (40 minutes) plus one hard
number you know (15 minutes) removes an option outright. Collect the numbers from SAA1–SAA4's
cheat sheets, because numbers eliminate faster than anything else.
A company runs a MySQL database on a Single-AZ Amazon RDS DB instance. Business analysts run heavy reporting queries against it during the day, and order processing slows down while the reports run. Separately, an audit found that the database is a single point of failure if its Availability Zone becomes unavailable. The company wants to fix both problems without changing the database engine or the application's data model. Which combination of actions should a solutions architect take? (Choose two.)
- A. Convert the DB instance to a Multi-AZ deployment.
- B. Create a read replica and point the reporting tool at the replica's endpoint.
- C. Convert to Multi-AZ and point the reporting tool at the standby instance.
- D. Move the database to Amazon DynamoDB to remove the connection limits.
- E. Increase the DB instance size.
Step 0. "(Choose two.)" This is a multiple-response item, so judge each option independently against the stem. You're not looking for a "best pair".
Step 1. Two separate problems: (1) reporting load hurting writes, (2) an AZ single point of failure. Constraints: same engine, same data model.
Pass A (impossible).
Pass B (fails a constraint).
A ✓ and B ✓. Two problems, two features. Multi-AZ is the availability feature ("enhanced database availability and durability"). Read replicas are the read-scaling feature. You "distribute your application's read traffic amongst them", and they're kept current with "asynchronous replication" (RDS FAQ).
The one-word lesson: A and C are near-twins. C adds "point the reporting tool at the standby", and that clause is impossible. The distractor is the right answer plus one false clause.
A financial services company runs a web application in one AWS Region and must add a disaster recovery Region. Requirements: data loss must be limited to seconds. After the decision to fail over, the DR Region must be able to accept requests immediately, even at reduced capacity, without first starting application servers, and then scale up. The solution must be the MOST cost-effective that meets these requirements.
- A. Take nightly backups, copy them to the DR Region, and redeploy the full stack from infrastructure as code when needed.
- B. Replicate data continuously to the DR Region and provision core infrastructure there. Keep application servers switched off until failover.
- C. Replicate data continuously to the DR Region and run a scaled-down but fully functional copy of the production stack there. Scale it up at failover.
- D. Run the full production stack at full capacity in both Regions and serve users from both.
Step 1. Constraints: RPO in seconds. The DR Region serves immediately without starting servers. Optimiser: most cost-effective. This is lesson 2's two-sided elimination.
Pass B (fails a constraint). The "too slow" end.
Pass C (loses on the optimiser). The "far more than needed" end.
C ✓. Warm standby: "a scaled down, but fully functional, copy of your production environment in another Region" (same page).
The lesson: the stem paraphrased AWS's own distinguishing sentence. When a requirement reads like a definition, find the term it defines.
| Situation | Do this |
|---|---|
| Two left, both look right | Line them up word by word (lesson 2). Ask what each differing word commits you to, then check it against the constraints. |
| Two left, still tied | Break the tie on operational overhead, since managed beats self-managed. Then cost. |
| Zero left | You misread a constraint. Re-read the stem, especially "not", "without", "only", "must". |
| You can't name a reason for any elimination | It's a knowledge gap, not a reading gap. Guess among the options you couldn't fault, flag it, move on. |
| Past ~2.5 minutes | Answer with your best survivor, flag it, move on (lesson 1's time plan). |