EC2 Pricing and Instance Types for SAA-C03: Which Family, Which Purchase Option
Know this topic? Test it on a full SAA-C03 paper
A complete SAA-C03 practice exam: 65 questions in 130 minutes, an explanation on every item and an Exam Coach report that ranks your weakest domains.
AWS-SAA-FULL
Try a free AWS SAA exam
Already have an account? Log in to redeem
SAA-C03 asks two EC2 questions over and over: which instance family fits the workload, and which purchase option fits the usage pattern. Steady workloads belong on Savings Plans or Reserved Instances. Interruptible work belongs on Spot. Short, unpredictable work stays On-Demand. Dedicated Hosts answer licensing requirements.Checked against AWS's published exam guide on 9 November 2026.
In this guide
Instance families
| Family | Optimised for | Typical workload |
|---|---|---|
| General purpose (M, T) | Balance of CPU and memory | Web servers, small databases |
| Compute optimized (C) | CPU | Batch, high-performance web, gaming |
| Memory optimized (R, X) | RAM | In-memory caches and large databases |
| Storage optimized (I, D) | High local disk I/O | Data warehousing, NoSQL |
| Accelerated (P, G) | GPU and hardware accelerators | Machine learning, graphics |
Purchase options
| Option | Discount | Choose when |
|---|---|---|
| On-Demand | None | Short, spiky or unpredictable work |
| Reserved Instances | Up to about 72% | Steady use, 1 or 3 year commitment |
| Savings Plans | Up to about 72% | Steady spend with more flexibility than RIs |
| Spot | Up to 90% | Fault-tolerant work that can be interrupted |
| Dedicated Hosts | Varies | Per-socket or per-core licensing, compliance |
| Capacity Reservations | None by itself | Guaranteed capacity in a specific AZ |
Discount ceilings as stated by AWS. Spot instances can be reclaimed with a two-minute warning.
Placement groups
| Type | Layout | Use for |
|---|---|---|
| Cluster | Close together in one AZ | Low latency, high throughput (HPC) |
| Spread | Each instance on distinct hardware | A few critical instances |
| Partition | Groups isolated across partitions | HDFS, Cassandra, Kafka |
Auto Scaling in one paragraph
An Auto Scaling group keeps capacity between a minimum and maximum, replacing unhealthy instances. Use target tracking to hold a metric such as CPU at a value, scheduled scaling for known peaks, and predictive scaling for regular patterns. Spread the group across multiple Availability Zones so one AZ failure does not take the service down.
A worked scenario: one workload, three purchase options
Requirement: a retailer has a steady baseline of web servers, large peaks in a sale season, and a nightly report that can restart if interrupted. Cost matters.
- Cover the steady baseline with a Savings Plan, because that usage is predictable.
- Let an Auto Scaling group add On-Demand capacity for peaks, since those hours are short and unpredictable.
- Run the nightly report on Spot Instances, because it tolerates interruption.
- If the sale must have guaranteed capacity in a specific AZ, add a Capacity Reservation.
Real architectures mix purchase options, and exam answers often do too. An answer that puts everything on one option is usually wrong.
EBS volume types
| Volume | Type | Use for |
|---|---|---|
| gp3 | General purpose SSD | Most workloads; IOPS set independently of size |
| io2 | Provisioned IOPS SSD | Critical databases needing sustained IOPS |
| st1 | Throughput optimized HDD | Big data and log processing |
| sc1 | Cold HDD | Infrequently accessed data at the lowest cost |
HDD volumes cannot be boot volumes. Instance store gives very high local I/O, but the data disappears when the instance stops. For long-running stateful workloads, EBS is the safer answer.
Other features the guide names
- Hibernation saves memory to the encrypted EBS root volume, so an instance resumes faster.
- Standard Reserved Instances give the bigger discount. Convertible ones can be exchanged for different configurations.
- Compute Savings Plans apply across instance families and Regions. EC2 Instance Savings Plans commit to a family in a Region for a higher discount.
- Target tracking holds a metric at a value, scheduled scaling covers known peaks, and predictive scaling forecasts regular patterns.
How the exam words it
| If the question says | Think |
|---|---|
| Steady usage for 1 to 3 years, flexibility needed | Compute Savings Plan |
| Can be interrupted, lowest cost | Spot Instances |
| Must keep capacity reserved in an AZ | Capacity Reservation |
| Licensing per core or socket | Dedicated Hosts |
| Resume quickly with memory intact | Hibernation |
| Lowest latency between nodes | Cluster placement group |
EC2 Pricing and Instance Types: practice questions
Five original questions. Each explanation says why the other options fail.
Your free AWS SAA exam code is ready
Sit the full SAA-C03 paper against the real 130-minute clock. AWS SAA is $39 once, lifetime, future versions included.
AWS-SAA-FULL
Redeem the code
Already have an account? Log in to redeem
Frequently asked questions
What is the difference between Savings Plans and Reserved Instances?
Both discount steady usage over 1 or 3 years. Savings Plans commit to a dollar-per-hour spend and are more flexible, while Reserved Instances commit to a specific instance configuration.
When should I use Spot Instances?
For fault-tolerant, flexible workloads that can handle interruption, such as batch jobs, because AWS can reclaim Spot capacity with a two-minute warning.
What is the difference between Dedicated Hosts and Dedicated Instances?
Dedicated Hosts give you a physical server with visibility of sockets and cores, which supports licensing. Dedicated Instances run on dedicated hardware without that visibility.