Identifying Keywords and Service Signals in Questions

SAA-C03 scenario questions often describe a business problem without explicitly naming the correct AWS service. Instead, they embed keyword signals — words and phrases that, when recognised, point directly to the correct service or approach. Expert test-takers read every scenario twice: first for the overall context, then for the specific keywords that narrow the answer. Missing a single keyword like 'serverless' or 'lowest cost' can lead you to a plausible but wrong answer.

When the question emphasises cost, specific keywords map to specific AWS patterns. 'most cost-effective' often signals Spot Instances, S3 Glacier, or Savings Plans. 'reduce cost without code changes' → Reserved Instances or right-sizing. 'pay only for what you use' → Lambda or Fargate (serverless). 'minimise data transfer cost' → CloudFront, VPC endpoints, or same-Region access. Always prioritise the cheaper, fully managed option when cost is the primary constraint.

Performance keywords map to specific architecture choices. 'millisecond latency' → DynamoDB or ElastiCache. 'global low latency' → CloudFront or Global Accelerator. 'high IOPS' → EBS io2 Block Express or instance store. 'shared file system across multiple EC2' → EFS. 'read-heavy workload' → Read Replicas or ElastiCache. 'GPU / machine learning' → P or G EC2 instance families. Match the performance requirement to the service designed for that specific workload.

Security keywords are frequent in SAA-C03 given Domain 1's 30% weight. 'encrypt data at rest' → KMS (SSE-KMS, S3-KMS, EBS/RDS encryption). 'rotate credentials automatically' → Secrets Manager. 'detect threats in real time' → GuardDuty. 'scan for vulnerabilities' → Amazon Inspector. 'discover PII in S3' → Amazon Macie. 'block SQL injection / XSS' → AWS WAF. 'DDoS protection' → AWS Shield Standard (free) or Shield Advanced (paid, with DRT support).

'least operational overhead' is one of the most common exam qualifiers. It always steers you toward managed/serverless services over self-managed alternatives. EC2 requires OS patching → move to Fargate or Lambda. Self-managed MySQL → move to RDS. Self-managed Redis → ElastiCache. Custom load balancer → ALB. Custom DNS → Route 53. Custom cron jobs → EventBridge Scheduled Rules. If two answers are architecturally equivalent but one is managed and one is self-managed, 'least operational overhead' selects the managed option.

'variable load', 'sudden traffic spikes', or 'unpredictable demand' signal the need for auto-scaling and elasticity. The answer almost always involves Auto Scaling Groups (for EC2), Lambda (for event-driven compute that scales to zero), DynamoDB On-Demand (for unpredictable database load), or Kinesis Data Streams with auto-scaling shards. Avoid fixed-capacity answers (Reserved Instances, provisioned throughput) when the scenario describes variable or unpredictable load.

Migration scenario keywords map directly to the 7 Rs and AWS migration tools. 'move quickly with no changes' → Rehost / AWS MGN. 'reduce DBA overhead' → Replatform to RDS. 'heterogeneous database engine migration' → DMS + SCT. 'move VMware workloads' → VMware Cloud on AWS (Relocate). 'discover on-premises servers' → Application Discovery Service. 'track migration progress' → AWS Migration Hub. 'transfer terabytes of data offline' → AWS Snowball Edge.

Database questions appear in Domains 2 and 3. Map requirements to the right database service. 'relational, managed, ACID' → RDS or Aurora. 'key-value, millisecond at scale' → DynamoDB. 'in-memory cache / session store' → ElastiCache. 'analytics / columnar queries on petabytes' → Amazon Redshift. 'fully managed search' → Amazon OpenSearch Service. 'graph relationships' → Amazon Neptune. 'time series' → Amazon Timestream. One wrong keyword on the database type usually eliminates the option immediately.

Decoupling scenarios test SQS vs SNS vs EventBridge vs Kinesis. 'decouple producers from consumers' → SQS standard queue. 'exactly-once, ordered processing' → SQS FIFO. 'fan-out to multiple subscribers' → SNS. 'route events by content' → EventBridge. 'real-time streaming with replay' → Kinesis Data Streams. 'deliver to S3/Redshift without code' → Kinesis Data Firehose. The key differentiators are: persistence/replay (Kinesis), ordering (FIFO), fan-out (SNS), and content-based routing (EventBridge).

The exam frequently asks for serverless solutions to eliminate EC2 management overhead. Map these keywords: 'no server management' → Lambda, Fargate, API Gateway. 'event-driven' → Lambda triggered by S3, SQS, DynamoDB Streams. 'HTTP API backend' → API Gateway + Lambda. 'run containers without managing infrastructure' → Fargate. 'serverless SQL on S3' → Athena. 'serverless ETL' → AWS Glue. 'serverless orchestration' → Step Functions Express Workflows. When a question says 'no servers', eliminate all EC2-based answers immediately.

Networking questions test your ability to match connectivity requirements to the right AWS service. 'connect on-premises to AWS privately' → Direct Connect or Site-to-Site VPN. 'route traffic globally with static anycast IPs' → AWS Global Accelerator. 'reduce latency for a global fleet of users' → CloudFront (content) or Global Accelerator (dynamic). 'private connectivity to AWS services without internet' → VPC Endpoints (Gateway for S3/DynamoDB, Interface for others). 'connect multiple VPCs in a hub-and-spoke model' → AWS Transit Gateway. Match the connectivity requirement to the service designed for that specific network pattern.

In this lesson you learned: exam keywords map directly to AWS services — cost signals → managed/serverless, performance signals → caching/CDN/high-IOPS, security signals → KMS/GuardDuty/WAF, 'least operational overhead' always selects managed services over self-managed EC2, and serverless keywords eliminate all EC2-based answers immediately. Next up we explore the process of elimination and recognising distractor answer patterns.