Analytics services: choosing the right one

The services at a glance

The Analytics category of the exam covers eleven services. Most analytics designs combine several of them, so it helps to know the job each one does before comparing them. Each service name links to its own tutorial, which covers its security, resilience, performance, and cost in depth.

Service	What it does
Amazon Kinesis Data Streams	Collects streaming records in real time and keeps them so that several applications can read them
Amazon Data Firehose	Delivers streaming data to destinations such as Amazon S3, Amazon Redshift, and Amazon OpenSearch Service, with no applications to write
Amazon MSK	Runs Apache Kafka clusters for you, so existing Kafka applications work unchanged
Amazon Athena	Runs standard SQL (Structured Query Language)The standard language for querying and changing data in relational databases. Analytics services such as Amazon Athena and Amazon Redshift also use SQL.Glossary entry queries directly on data in Amazon S3, with no infrastructure to manage
Amazon Redshift	A fully managed, petabyte-scale Data warehouseA database designed for analytics, holding cleaned, structured data from many sources, so that reports and queries across years of history run fast. Amazon Redshift is the AWS data warehouse.Glossary entry
AWS Glue	Serverless data integration: crawlers, the Data Catalog, and ETL (Extract, transform, and load)Copying data from sources, cleaning and reshaping it, and loading it into a target such as a data warehouse. AWS Glue is the AWS serverless ETL service.Glossary entry jobs
Amazon EMR	Managed clusters for big data frameworks such as Apache Spark and Apache Hadoop
Amazon OpenSearch Service	Managed OpenSearch clusters for log analytics, full-text search, and monitoring
AWS Lake Formation	Central governance and fine-grained access control for a Data lakeA central store for large amounts of structured and unstructured data kept in its original form, usually in Amazon S3, which many tools can analyze. AWS Lake Formation helps build and secure one.Glossary entry
AWS Data Exchange	Finds, subscribes to, and uses data from other organizations
Amazon Quick	Business intelligence dashboards and visualizations (Amazon Quick Sight, formerly Amazon QuickSight)

Amazon Kinesis Video Streams shares the Kinesis name, but the exam lists it under Media Services.

How they fit together

A typical data lake pipeline moves through five stages:

Ingest: Kinesis Data Streams, Data Firehose, or Amazon MSK bring streaming data in.
Store and catalog: data lands in Amazon S3, and AWS Glue crawlers record its schema in the AWS Glue Data Catalog.
Process: AWS Glue jobs or Amazon EMR clusters clean and transform the data. Amazon Managed Service for Apache Flink processes the streams themselves as the data arrives.
Query and analyze: Athena queries the data in place, Redshift serves warehouse workloads, and OpenSearch Service handles search and log analytics.
Visualize and govern: Amazon Quick builds dashboards, and Lake Formation controls who can see which databases, tables, columns, and rows.

Real-time analytics on streams: Amazon Managed Service for Apache Flink

Amazon Managed Service for Apache Flink (formerly Amazon Kinesis Data Analytics) runs Apache Flink applications with no servers to manage. An application reads a Kinesis data stream or an Amazon MSK topic and computes results continuously as records arrive: windowed counts and aggregations (such as "orders per store in the last 5 minutes"), filtering, joins, and streaming ETL. You write the application in Java, Scala, Python, or SQL, or query streams interactively in a Studio notebook. The service scales the application and checkpoints its state.

Data Firehose only buffers and delivers, and Athena or Glue run on Encryption at restEncrypting data where it is stored, such as on disks, in databases, or in backups, so that it cannot be read without the key. On AWS it usually uses keys in AWS KMS.Glossary entry, so a requirement to compute rolling or windowed results on a stream in real time points to Managed Service for Apache Flink.

Reverse mindmap: from the question to the service

Exam scenarios rarely name the service you need. They describe a requirement. Find the kind of problem the question describes, then follow its clue to the service.

Analytics: which service?

The data arrives as a stream
- Real-time processing, several consumers, replay leads to Kinesis Data Streams
- Deliver to S3, Redshift, or OpenSearch with no code leads to Data Firehose
- Convert streaming JSON to Parquet or ORC leads to Data Firehose
- Existing Apache Kafka applications leads to Amazon MSK
- Windowed counts or aggregations on a stream in real time leads to Managed Service for Apache Flink
The data is queried with SQL
- Ad hoc SQL on files in S3, serverless leads to Athena
- Data warehouse for complex reporting leads to Redshift
- Query S3 from Redshift without loading leads to Redshift Spectrum
The data is prepared or processed
- Discover schemas, central table catalog leads to Glue crawlers and Data Catalog
- Serverless ETL, no clusters leads to AWS Glue
- Spark, Hadoop, Hive, or HBase with cluster control leads to Amazon EMR
The data is searched
- Full-text search, log analytics, monitoring leads to OpenSearch Service
The data is governed, shared, or shown
- Column, row, or cell permissions on a data lake leads to Lake Formation
- Share data lake tables with other accounts leads to Lake Formation
- Subscribe to third-party data sets leads to AWS Data Exchange
- Dashboards for business users, embedded analytics leads to Amazon Quick

Commonly confused pairs

If the choice is between	Choose the first when	Choose the second when
Kinesis Data Streams or Data Firehose	You need custom real-time processing, several consumers, or replay	You only need to deliver data to a destination
Kinesis Data Streams or Amazon MSK	You want an AWS-native stream with no Kafka to run	You need Apache Kafka compatibility
Athena or Redshift	Queries are occasional and the data stays in Amazon S3	Many users run frequent, complex reporting queries
AWS Glue or Amazon EMR	You want serverless ETL	You need control over big data frameworks and clusters
Athena or OpenSearch Service	You query structured data with SQL	You search text or analyze logs
Data Firehose or Managed Service for Apache Flink	You deliver a stream to a destination	You compute results, such as windowed aggregations, continuously on the stream
Amazon Macie or Lake Formation	You need to find sensitive data, such as personal data, in S3 buckets	You need to control who can query which databases, tables, columns, and rows

Key takeaways

Identify the stage the question is about: ingest, store, process, query, or govern.
"Deliver" and "no code" point to Data Firehose; "real-time processing" and "multiple consumers" point to Kinesis Data Streams.
"Serverless SQL on Amazon S3" points to Athena; "data warehouse" points to Redshift.
"Real-time windowed analytics on a stream" points to Managed Service for Apache Flink.
"Fine-grained data lake permissions" points to Lake Formation; "find sensitive data in S3" points to Amazon Macie.