In the SAA-C03 exam, storage questions account for roughly 10 to 15% of all questions. The core question is always: "Which storage service fits this workload?" To pick the right answer, you need to understand the performance characteristics of each service precisely.
EBS Volume Types — Think of Different Filing Cabinets
EBS (Elastic Block Store) is block storage that you attach to an EC2 instance. Think of it as an external hard drive that you plug into a computer.
Let us compare EBS volume types to different kinds of filing cabinets.
SSD Family — The Fast Drawer
SSD volumes are ideal for workloads where IOPS (Input/Output Operations Per Second) matters most.
gp3 (General Purpose SSD 3rd Generation):
The general-purpose SSD for most workloads. It delivers a baseline of 3,000 IOPS that you can independently scale up to 16,000 IOPS without increasing storage capacity. It is 20% cheaper than gp2 while offering more flexible performance tuning. On the exam, if you see "general purpose, no special requirements," choose gp3.
gp2 (General Purpose SSD 2nd Generation):
IOPS scales automatically with volume size (3 IOPS per GB). The larger the volume, the more IOPS it automatically provides. While gp3 is the better choice today, watch for the exam keyword "IOPS proportional to capacity" — that points to gp2.
io2 Block Express / io1 (Provisioned IOPS SSD):
Ultra-high-performance SSD. io2 Block Express supports up to 256,000 IOPS, used for mission-critical databases like Oracle RAC that demand extreme I/O performance. In the filing cabinet analogy, this is the turbo-powered high-speed drawer.
HDD Family — The High-Volume Warehouse
HDD volumes are designed for workloads where throughput (MB/s) matters more than IOPS. One critical rule: HDD volumes cannot be used as EC2 boot volumes.
st1 (Throughput Optimized HDD):
A low-cost HDD optimized for big data, data warehouses, and log processing. Supports up to 500 MB/s throughput. Exam keywords: "big data," "high throughput," "sequential access," "MapReduce," "Kafka."
sc1 (Cold HDD):
The cheapest EBS option. Used for data that is accessed infrequently. Supports up to 250 MB/s throughput. Exam keywords: "cold data," "lowest cost EBS," "infrequently accessed."
Key selection rule: IOPS (transaction processing) is the priority -- SSD (gp3 or io2) Throughput (large sequential reads/writes) is the priority -- HDD (st1) Mission-critical DB, maximum possible performance needed -- io2 Block Express
!SSD versus HDD EBS volume families
S3 Performance Optimization
S3 looks like a simple object store, but the exam also tests how to handle large-scale file operations efficiently.
Prefix-Based Performance Scaling
S3 can handle 3,500 PUT/COPY/POST/DELETE and 5,500 GET/HEAD requests per second per prefix. By spreading data across multiple prefixes, you can scale throughput linearly.
Example: Using 4 different prefixes allows you to process 22,000 GET requests per second.
Multipart Upload
For files larger than 100 MB, multipart upload is recommended. The file is split into multiple parts that upload in parallel, dramatically increasing upload speed. For files over 5 GB, multipart upload is mandatory.
Exam keywords: "fast upload of large files," "files larger than 5 GB."
S3 Transfer Acceleration
S3 Transfer Acceleration speeds up uploads by routing them through CloudFront edge locations. Instead of traveling across the entire internet to reach the S3 bucket, your data only travels over the internet to the nearest AWS edge location, then continues to S3 over AWS's own high-speed internal network.
Exam keywords: "global users uploading quickly to S3," "fast upload from overseas clients."
EFS and FSx — Choosing a Shared File System
When you have multiple EC2 instances that need to access the same files simultaneously, you need a shared file system. While EBS volumes generally connect to a single EC2 instance at a time, EFS and FSx allow multiple instances to connect concurrently.
EFS (Elastic File System)
A fully managed shared file system for Linux using the NFS protocol. Multiple EC2 instances across multiple AZs can access it simultaneously. Capacity scales up and down automatically, and you pay only for what you use.
Exam keywords: "shared file system for multiple EC2 instances," "Linux NFS," "multi-AZ access."
FSx for Windows File Server
A fully managed Windows shared file system using the SMB protocol. It integrates with Active Directory and is fully compatible with Windows-based applications.
Exam keywords: "Windows shared file system," "SMB protocol," "Active Directory integration."
FSx for Lustre — A Dedicated High-Speed Highway
FSx for Lustre is a high-performance parallel file system purpose-built for HPC (High Performance Computing) and machine learning training workloads.
Highway analogy: A regular road (EFS) serves passenger cars and trucks together. But heavy freight trucks running at full speed need a dedicated highway. FSx for Lustre is that dedicated high-speed highway for HPC data.
It integrates directly with S3. The typical pattern is: pull data from S3 into FSx for Lustre, process it at high speed, then write results back to S3.
Exam keywords: "HPC," "machine learning training," "high-performance parallel file system," "high-speed processing with S3 integration."
FSx for NetApp ONTAP
A multi-protocol file system supporting NFS, SMB, and iSCSI simultaneously. If you are currently running NetApp ONTAP on-premises, you can migrate to the cloud without changing your applications.
Exam keywords: "on-premises NetApp migration," "multi-protocol storage."