The AWS Service Map
Find any service by how it fits the map
AWS is a huge family of services. Learning the map — compute, storage, network, database, security, analytics — tells you where any service fits and what it's for.
What you will learn
- Group AWS services into the core functional families
- Map a typical app to concrete AWS services
- Navigate documentation and pricing pages without drowning
AWS organizes around families: compute, storage, database, networking, security&identity, analytics, integration and management. Every new service is an answer to a question somewhere in one of those families.
What is this?
Amazon Web Services is the largest public cloud with more than 200 services. The ones you'll meet first: EC2 and Lambda (compute), S3 and EBS (storage), RDS and DynamoDB (databases), VPC (network), IAM (security), Route 53 (DNS), CloudFront (CDN) and CloudWatch (observability). Roughly: EC2 is a VM, S3 is object storage, VPC is your private network, IAM is identities and access, Lambda is functions, RDS is managed relational databases.
Why do you need it?
- Reading an AWS architecture diagram without the map is reading a foreign alphabet.
- Each family has a service that answers a specific workload question — knowing the family narrows the search.
- Interviewers ask 'how would you build X' — the map gives you a syntax for the answer.
- AWS evolves fast, but the map is stable; specific features change monthly.
Mental model
A hardware store arranged by department
The AWS service catalog
You don't wander for the right wrench; you know plumbing is aisle 4. AWS is arranged the same way — find the family, then the exact tool that fits the job.
How it works
- Compute family: EC2 (servers), ECS/EKS (containers), Lambda (functions), plus Auto Scaling and machine images.
- Storage family: S3 (objects), EBS (block volumes), EFS/FSx (file systems), Glacier (archive).
- Database family: RDS/Aurora (relational), DynamoDB (NoSQL key-value), Redshift (warehouse), ElastiCache (cache).
- Networking family: VPC (network), Route 53 (DNS), CloudFront (CDN), ELB (load balancing), Transit Gateway (connectivity).
- Cross-cutting: IAM (access), CloudWatch (monitoring), CloudTrail (audit), KMS (encryption), SQS/SNS (integration).
The AWS families and their flagship services.
In production
- A typical web app: EC2 or ECS in a VPC, RDS for the database, S3 for uploads, Route 53 for DNS, CloudFront in front, CloudWatch for monitoring, IAM for access.
- A serverless pipeline: S3 events trigger Lambda handlers that write to DynamoDB, orchestrated with Step Functions.
- A big-data stack: Kinesis streams events, S3 stores raw, Glue/EMR transforms, Redshift serves analytics.
Best practices
- Learn the flagship of each family before exploring variants (EC2, S3, VPC, RDS, IAM, Lambda).
- Use the AWS documentation service list and pricing pages as the map, not blogs.
- Try the console for exploration but move to the CLI/API for anything repeatable.
- Keep an eye on announcements — every re:Invent renames parts of your mental model.
Common mistakes
- Using EC2 for everything because it's the best-known service.
- Confusing S3 (object storage) with EBS (block storage) when picking a storage model.
- Ignoring IAM until after resources are public.
- Assuming one region means you're 'in AWS' — region choice matters for latency and residency.
Common issues
Can't reach an EC2 instance over SSHSecurity group or NACL blocking, or no key pair configured
Check the instance's security group inbound rule and that your key pair matches the instance.
Lambda times out on every callFunction placed in a VPC with no NAT egress, or timeout too low
Attach a NAT gateway/endpoint or place Lambdas outside the VPC; raise timeouts deliberately.
S3 object is 'not found' but existsBucket policy denies the requester, or a lifecycle rule expired it
Test with the right identity; check versions, object ACL and lifecycle events in CloudTrail.
Debugging tips
- Start diagnosis at the family: if storage, check bucket policy + permissions first.
- CloudWatch metrics and CloudTrail event history belong in every 'what changed' hunt.
- For integration errors, check SQS/SNS dead-letter queues before the consumer code.
- Read the exact service error code — AccessDenied, Throttling and Timeout have different causes.
Interview checkpoint
Quick quiz
Which AWS service is the managed object storage service?
EC2 instances automatically run on a public network with no VPC.