
AWS Core Services Explained: Compute, Storage, Data, IoT, and Machine Learning
This article provides a structured overview of core AWS services by grouping them according to the problems they solve. It covers IoT and device management, compute, data processing, messaging and workflows, storage and databases, analytics, machine learning, and monitoring, helping readers understand where each service fits and how to choose the right AWS building block for different cloud workloads.

AWS has a large service ecosystem, but it becomes much easier to understand when services are grouped by the problem they solve. Some of the major areas include IoT and device management, compute, data processing, messaging and orchestration, storage and databases, analytics and visualization, machine learning, and monitoring.
- IoT and Device Services
This category handles the connection between physical devices and the AWS cloud. It covers device communication, fleet management, security, edge processing, and digital representations of physical systems.
These services are especially useful when working with sensors, embedded devices, vehicles, industrial equipment, or other connected systems. AWS IoT Core acts as the main connectivity layer, while the surrounding services extend this with device management, security, edge computing, and specialized IoT capabilities.
AWS IoT Core is particularly important because it supports lightweight communication protocols such as MQTT, which is well suited for constrained devices and networks.
AWS Service | What it does |
|---|---|
AWS IoT Core | Securely connects IoT devices to AWS using protocols such as MQTT, HTTPS, WebSockets, and LoRaWAN. |
AWS IoT Device Management | Registers, organizes, monitors, and remotely manages IoT devices at scale. |
AWS IoT Device Defender | Audits IoT security configurations and detects abnormal or potentially unsafe device behaviour. |
AWS IoT Greengrass | Extends AWS services to edge devices for local computation and offline operation. |
AWS IoT FleetWise | Collects and analyzes telemetry and sensor data from vehicles. |
AWS IoT TwinMaker | Creates digital twins that represent physical systems digitally. |
- Compute Services
Compute services provide the processing power required to run applications, APIs, backend logic, analytics, and machine-learning workloads.
Two important compute models are Amazon EC2 and AWS Lambda. With EC2, you provision and manage a virtual server. With Lambda, AWS manages the underlying infrastructure and runs your code only when an event occurs.
AWS Service | What it does |
|---|---|
Amazon EC2 | Provides configurable virtual servers for running long-running applications, websites, databases, analytics, and ML workloads. |
AWS Lambda | Runs code in response to events without requiring developers to provision or manage servers. |
For EC2, you choose an Amazon Machine Image (AMI), select an instance type based on CPU, memory, storage, and networking requirements, configure security and networking, and then launch the instance.
Lambda follows an event-driven model. An event such as an HTTP request, IoT message, or file upload triggers a function, AWS provides the required compute resources, the function executes, and the resources are released afterwards.
A useful distinction is:
Aspect | Amazon EC2 | AWS Lambda |
|---|---|---|
Execution model | Continuously running server | Event-driven function |
Infrastructure management | Managed by user | Managed by AWS |
Scaling | Manual or configured Auto Scaling | Automatic |
Best suited for | Long-running services | Short event-driven tasks |
3. Data Processing Services
Data-processing services take incoming raw data and transform, analyze, or process it before storage or further use.
Different services are suited to different workloads. AWS Lambda handles event-driven processing, Amazon Kinesis handles real-time streaming workloads, AWS Glue performs ETL operations, and Amazon EMR handles large-scale distributed data processing using frameworks such as Hadoop and Spark.
AWS Service | What it does |
|---|---|
AWS Lambda | Executes lightweight processing logic automatically when an event occurs. |
Amazon Kinesis | Ingests and processes continuously arriving real-time data streams. |
AWS Glue | Performs ETL operations to extract, transform, clean, and prepare data. |
Amazon EMR | Processes large datasets using distributed frameworks such as Hadoop and Spark. |
A useful way to remember them is:
- Lambda → event-driven processing
- Kinesis → streaming processing
- Glue → data preparation and ETL
- EMR → large-scale distributed data processing
- Messaging and Workflow Services
Modern cloud systems are often distributed across multiple components. Messaging and workflow services allow these components to communicate without being tightly coupled.
Queues are useful when one service needs to process work independently from another. Notification services are useful when events need to be delivered to one or more subscribers, while workflow orchestration helps coordinate multiple steps in a controlled sequence.
AWS Service | What it does |
|---|---|
Amazon SQS | Provides message queues that allow application components to communicate asynchronously. |
Amazon SNS | Publishes notifications and messages to subscribers. |
AWS Step Functions | Coordinates multi-step workflows across different AWS services. |
These services form an important foundation for event-driven and distributed systems, where components do not always need to execute at the same time or depend directly on one another.
- Storage and Database Services
AWS provides different storage technologies because applications store different kinds of data.
A file does not necessarily belong in a relational database, and high-volume sensor readings may require a different storage model from structured transactional data. AWS therefore provides separate services for object storage, relational databases, NoSQL databases, time-series databases, and data warehouses.
AWS Service | What it does |
|---|---|
Amazon S3 | Stores objects such as files, logs, images, videos, and large volumes of raw data. |
Amazon DynamoDB | Provides a high-speed NoSQL key-value/document database for scalable applications. |
Amazon RDS | Provides managed relational databases for structured and relational application data. |
Amazon Timestream | Stores and queries time-series data such as timestamped sensor readings. |
Amazon Redshift | Acts as a data warehouse for large-scale analytical workloads. |
Amazon S3
- Amazon S3 is primarily used for object storage. It can store raw sensor data, images, videos, logs, documents, backups, and other forms of long-term application data.
Amazon RDS
- Amazon RDS provides managed relational databases while AWS handles many administrative tasks such as: *
- hardware provisioning
- database installation
- backups
- software patching
- high availability
- scaling
- RDS supports database engines such as: *
- MySQL
- PostgreSQL
- MariaDB
- Oracle Database
- Microsoft SQL Server
Amazon DynamoDB
- DynamoDB is a NoSQL key-value/document database designed for scalable, low-latency workloads. It differs from RDS because it does not rely on the traditional relational-table model.
Feature | Amazon RDS | Amazon DynamoDB |
|---|---|---|
Data model | Relational | Key-value / Document |
Query approach | SQL | DynamoDB API |
Relationships | Supports joins | No traditional joins |
Typical use | Structured business data | High-scale web and IoT workloads |
Strength | Transactions and relationships | Throughput and low latency |
- Analytics and Visualization Services
Analytics services help turn stored data into useful information.
Some services allow SQL queries over stored data, some provide dashboards, while others focus on search, monitoring, or predictive analytics.
AWS Service | What it does |
|---|---|
Amazon Athena | Runs SQL queries directly against data stored in Amazon S3. |
Amazon QuickSight | Creates interactive dashboards and visualizations. |
Amazon OpenSearch Service | Supports search, log analytics, monitoring, and visualization. |
Amazon Forecast | Performs predictive analytics for patterns such as demand or sensor trends. |
Amazon Athena is useful when data is already stored in Amazon S3 and needs to be queried using SQL without first moving it into a traditional database.
Amazon QuickSight provides the visualization layer by turning processed data into interactive dashboards that users can inspect and analyze.
- Machine Learning Services
Machine-learning services provide infrastructure and tooling for preparing data, building models, training them, deploying them, and monitoring their performance.
A typical Amazon SageMaker workflow follows the ML lifecycle.
AWS Service | What it does |
|---|---|
Amazon SageMaker | Provides a managed environment to build, train, deploy, and monitor machine-learning models at scale. |
The SageMaker workflow - ML lifecycle.
- Data Preparation
Data can be imported from:
- Amazon S3
- AWS IoT Analytics
- databases
- data lakes
and then cleaned and transformed before training.
- Model Development
SageMaker supports popular ML frameworks including:
- TensorFlow
- PyTorch
- Scikit-learn
- XGBoost
- Training
Models can be trained using scalable CPU or GPU resources, with hyperparameter tuning used to improve performance.
- Deployment and Monitoring
Trained models can be exposed through scalable REST API endpoints for both real-time and batch inference. SageMaker can also monitor prediction accuracy, data drift, and performance degradation.
- Monitoring and Observability
Once an application is running, developers need visibility into how the system behaves.
Monitoring and observability help answer questions such as:
- Is the system healthy?
- Is CPU utilization increasing?
- Are Lambda functions failing?
- Is the database receiving too many connections?
- Did an application produce an error?
- Should an alert be triggered ?
Amazon CloudWatch is one of the main AWS services used for monitoring and observability.
AWS Service | What it does |
|---|---|
Amazon CloudWatch | Collects metrics, logs, alarms, dashboards, and operational information from AWS resources and applications. |
CloudWatch includes several important components:
CloudWatch Component | Purpose |
|---|---|
Metrics | Measures values such as CPU utilization, request count, errors, and duration. |
Logs | Stores and analyzes application and system logs. |
Alarms | Triggers actions when metric thresholds are crossed. |
Dashboards | Displays operational metrics and logs visually. |
EventBridge Integration | Responds automatically to AWS service events. |
CloudWatch Agent | Collects additional system metrics such as memory, disk, and processes. |
Examples of monitored data include EC2 CPU utilization and network traffic, RDS database connections, and Lambda invocation counts, errors, and execution duration.
- AWS IoT Analytics
AWS IoT Analytics is designed for receiving, processing, storing, and analyzing IoT data.
Its workflow is organized around four major components:
Processed IoT data can then be used for analytics or machine-learning workflows, while services such as AWS Lambda can interact with deployed SageMaker endpoints for inference.
Component | Role |
|---|---|
Channel | Receives and stores incoming raw IoT data. |
Pipeline | Cleans, filters, transforms, and enriches incoming data. |
Data Store | Stores processed IoT data. |
Data Set | Runs SQL-like queries over processed data and produces analytical outputs. |
Quick Reference
A practical way to remember these services is by responsibility:
Category | Main AWS Services |
|---|---|
IoT & Devices | IoT Core, Device Management, Device Defender, Greengrass, FleetWise, TwinMaker |
Compute | EC2, Lambda |
Data Processing | Lambda, Kinesis, Glue, EMR |
Messaging & Workflows | SQS, SNS, Step Functions |
Storage & Databases | S3, DynamoDB, RDS, Timestream, Redshift |
Analytics & Visualization | Athena, QuickSight, OpenSearch, Forecast |
Machine Learning | SageMaker |
Monitoring | CloudWatch |
IoT Analytics | AWS IoT Analytics |

Final Takeaway
A practical way to understand these AWS services is to think in terms of responsibilities:
connect → compute → process → communicate → store → analyze → learn → monitor
Each service solves a specific problem, and understanding that role is much more useful than memorizing service names individually. This also creates a strong base for moving into more advanced AWS topics such as distributed systems, event-driven architectures, real-time streaming, container orchestration, and AI workloads.
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article