Separate analytics responsibilities
Amazon Kinesis Data Streams captures continuing data streams that consumers can process. AWS Glue provides data integration and preparation capabilities, including ETL and associated catalog metadata. Amazon Athena is an interactive query service that uses standard SQL to analyze data in S3. Amazon Redshift is an analytical data warehouse for structured analytical workloads. These purposes can coexist in one data pipeline. A metadata catalog is not itself the query execution engine, and collecting a stream does not automatically produce a dashboard. Redshift also has a serverless option; do not infer from its warehouse purpose that every deployment requires a permanently provisioned cluster. Choose from the responsibility stated in the case rather than an absolute claim about what another product can never do.
Match the modality before naming AI
Amazon Rekognition analyzes images and video, including identifying objects and scenes. Amazon Transcribe converts speech into text. Amazon Translate translates text between languages. Amazon Comprehend analyzes unstructured text for outputs such as entities and sentiment. Amazon Lex supports conversational interfaces. All can be discussed under machine learning, but that broad category is too coarse to select the missing step. A written notice does not need speech transcription merely because users might later read it aloud. A recorded interview does not yet provide the text required for a text-only analysis step. A chatbot requirement adds an interaction and dialogue responsibility beyond producing a transcript.
Worked example: a community history project
A fictional community archive collects recorded interviews, photographs and written visitor comments. The curator needs a transcript of each recording, a Spanish text version of the transcript, image labels for organizing photographs, and a sentiment summary of comments. First separate the four outputs. Speech-to-text maps to Transcribe, text translation to Translate, visual analysis to Rekognition, and text interpretation to Comprehend. A conversational bot is not required by these outputs, so adding Lex merely because there is speech in the project would not close an identified gap. The archive must still review suitability and mistakes before exposing labels or translations to visitors. The product-purpose comparison does not guarantee that generated output is accurate or culturally appropriate.
A second example: analyze the arrival path
Now the archive receives a continuing feed of device events. One colleague proposes Athena because it supports SQL. Another proposes Kinesis Data Streams because events must be captured continuously for real-time consumers. The immediate missing capability is capture, so the latter addresses that stage. If the case later asks for occasional SQL over files already in S3, Athena becomes relevant to that later requirement. Glue can prepare/catalog data for analysis; Redshift can be the selected warehouse. A larger design may use several services, but a single capability question should be answered for the particular boundary it identifies. Avoid declaring the rejected tool universally useless.
Transfer check: change only one condition
A fictional support service already has written English feedback and wants entities and sentiment, not a translated copy. Name the input and required output before choosing a capability. Then change the case: the input becomes recorded English calls and the first required output is a written transcript. Explain why the original choice no longer fills the first missing step. Finally add a separate requirement for an interactive help conversation. State which additional dialogue responsibility is now present and why transcription alone does not establish it. Try all three explanations before reading the basis in the next section.
Check basis and limits
For written feedback, Comprehend matches the stated text interpretation. Recorded speech first requires speech-to-text transcription, which is the Transcribe purpose in this bounded comparison. An interactive conversation adds the Lex conversational-interface responsibility. These answers use the explicit inputs and outputs; they do not prove that every possible implementation must use those products or that one API call constitutes a complete safe application. Record any uncertainty about formats, permissions, output evaluation and integration as further design questions. Use the linked original practice tasks to explain the strongest competing capability rather than treating recognition of a service name as mastery.
Try it, then explain it.
Complete a short set without looking at the example. After each answer, explain which evidence or rule supports your decision and why the alternatives do not fit.
Sources & authorship
Original OpenExam explanations and examples. References support the concepts; they are not the source of copied questions. Independent educator review is pending.
rd-aws-aird-aws-analytics