This document discusses optimizing costs and capacity on Amazon EC2. It provides an overview of AWS global infrastructure including 23 regions and 216 CloudFront points of presence. It then discusses Amazon EC2 instance types and characteristics, and how customers can optimize costs through purchase options like Reserved Instances (RIs), Savings Plans, and Spot Instances. The document also discusses how to configure Auto Scaling groups to use different purchase options together to further optimize costs.

2. © 2020, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Cost and capacity optimization
0 5
Cobus Bernard
Sr Developer Advocate
AWS
@cobusbernard
cobusbernard
cobusbernard

4. AWS global platform
AWS global infrastructure
• 23 Regions with 73 Availability Zones
• 4 Regions coming soon: Indonesia, Italy,
Japan and Spain
216 CloudFront PoPs
• 205 edge locations
• 11 Regional edge caches
• 245 Countries and territories served
AWS global network
• Redundant 100 GbE network
• 100% encrypted between facilities
• Private network capacity between
all AWS Regions except China
SLA of
99.99% availability

5. Amazon EC2 instance characteristics
M5d.xlarge
Instance family
Instance
generation
Instance size
Instance type
CPU
Memory
Storage
Network performance
Additional
capabilities

6. Broadest and deepest platform choice
Workloads Capabilities Options
(AWS, Intel, AMD)
(up to 4.0 GHz)
(up to 24 TiB)
(HDD and NVMe)
(up to 100 Gbps)
(GPUs and FPGA)
(Nano to 32xlarge)
+ + =
270+instance types

7. Continued rapid pace of innovation
Instance growth
270+

8. Customer obsessed
of the roadmap originates with customer requests
and is designed to meet specific needs
90%

9. Optimizing Amazon
EC2 cost and capacity

10. Optimizing Amazon EC2 cost and capacity
We continue to innovate for our customers
Pricing Capacity Guidance

11. Optimizing Amazon EC2 cost and capacity
We continue to innovate for our customers
Pricing Capacity
Capacity management
made easy on the
broadest and deepest
compute platform
Guidance
Cost and capacity
recommendations
enable ease of use
and save time

12. the second
Amazon EC2 purchase options
savings of up to 90%a significant discount
more flexibility

13. To optimize Amazon EC2, combine purchase options
RIs or a Savings Plan
Spot for fault-tolerant,
flexible, stateless workloads
On-Demand

14. Deutsche Börse Group
… covers the entire value chain in securities and derivatives trading.
• Pre-IPO and listing
• Trading
• Clearing
• Settlement
• Custody
• Collateral and liquidity
management
• Market data
• Indices
• Technology

15. Other Instance Types can make the difference
0
0.2
0.4
0.6
0.8
1
r5.2xl r5a.2xl r5d.2xl m5.4xl r4.2xl m4.4xl
InstancePrice$/h
eu-central-1
spot price on-demand price

16. T7 and Cloud Use-Cases – SmokeTest termination rate
1521
2919
4532
3323
2459
20 25 128 29 71
0
1000
2000
3000
4000
5000
2019-Jan 2019-Feb 2019-Mar 2019-Apr 2019-May
Number of SmokeTest Spot Instances and Terminations
SmokeTest all SmokeTest terminated

17. Introducing Savings Plans
Easy to use
Receive discounted rates
automatically in exchange
for a monetary commitment
Flexible
Make a single commitment that
applies across multiple AWS
compute services, even as
your requirements change
Significant discounts
Select from two types of Savings
Plans to receive discounts of up
to 72% on EC2 Instance Savings Plans
and 66% on Compute Savings Plans
Flexible purchase option that offers up to 72% discounts
on Amazon EC2 and AWS Fargate usage

18. Types of Savings Plans
Provide the lowest prices, up to 72% off (same
as Standard RIs) on the selected instance family
(e.g., C5 or M5), in a specific AWS Region
Offer the greatest flexibility, up to 66% off
(same prices as Convertible RIs)
Flexible
across
 Instance family: e.g., Move from C5 to M5
 Region: e.g., Change from EU (Ireland) to EU
(London)
 OS: e.g., Windows to Linux
 Tenancy: e.g., Switch Dedicated tenancy to
Default tenancy
 Compute options: e.g., Move from EC2 to
Fargate
Flexible
across
 Size: e.g., Move from m5.xl to m5.4xl
 OS: e.g., Change from m5.xl Windows
to m5.xl Linux
 Tenancy: e.g., Modify m5.xl Dedicated
to m5.xl Default tenancy
Compute
Savings Plans
EC2 Instance
Savings Plans

19. Comparing RIs and Savings Plans
Savings Plans offer all the benefits of RIs as well as improved flexibility and reduced management
Compute
Savings Plans
EC2 Instance
Savings Plans
Convertible RIs* Standard RIs
Savings over On-Demand Up to 66% Up to 72% Up to 66% Up to 72%
Low price in exchange for
monetary commitment    
Pricing automatically applies
to any instance families    
Pricing automatically applies
to any instance size   ** **
Pricing automatically applies
to any tenancy or OS    
Automatically apply to
Fargate usage    
Pricing automatically applies
across any AWS Region
   
1- and 3-year term
length options
   
* Convertible RIs can be changed across instance families, sizes, OS, and tenancy – they require customers to manually perform exchanges
** Regional Convertible RIs and Regional Standard RIs provide instance size flexibility

20. Getting started with Savings Plans
Review your Savings
Plans recommendations
in AWS Cost Explorer
Customize
recommendations
based on your needs
(type of Savings Plan,
payment option,
term length)
Review hourly
commitment (e.g., $10/hr)
and add to cart
Eligible Amazon EC2
and AWS Fargate
usage is charged at a
discounted Savings
Plans rate up to your
commitment level
AWS Cost Explorer guides you through the purchasing process
Just like RIs, you can purchase Savings Plans via the RI operations team

21. Spot, On-Demand capacity reservations, and
Savings Plan together
On-Demand capacity
reservations
Savings Plan
Spot Instances
Cost-effective,
scalable compute

22. Capacity
Interruptions only
happen if OD
needs capacity
Pricing
Smooth, infrequent
changes, more predictable
Instances
Same infrastructure as
On-Demand and RIs
Usage
Choose different instance
types, sizes, and AZs in
a single fleet or EC2 Auto
Scaling group
Pricing is based on long-term supply and demand trends; no bidding!
Save up to 90% using EC2 Spot Instances

23. Low, predictable prices
Up to 90% discount over On-Demand prices
Faster results
Increase throughput up to 10x while staying in budget
Easy to use
Launch through AWS services (e.g., Amazon ECS, Amazon EKS,
AWS Batch, Amazon SageMaker, Amazon EMR) or integrated
third parties
Why Spot Instances?

24. Minimal interruptions
Check for 2-minute interruption notification via instance
metadata or Amazon CloudWatch events, and automate by
 Checkpointing
 Draining from ELB
 Using stop-start and hibernate to restart faster
Interruption handlers for Amazon ECS and Amazon EKS
Amazon Elastic
Kubernetes Service
(Amazon EKS)
 Connection between termination requests from AWS infrastructure to nodes
 Tasks running on Spot Instances will automatically be triggered for shutdown
before the instance terminates, and replacement tasks will be scheduled
elsewhere on the cluster
Amazon Elastic
Container Service
(Amazon ECS)
Handling Spot interruptions
Less than 5% of Spot Instances were interrupted in the last 3 months

25. Flexibility is key to successful Spot usage
Instance flexible Time flexible Region flexible

26. B2B Enterprise TechSports, Media & Entertainment Financial Services
Consumer AppsResearch AdTech & MarTech
Customers across different industries and verticals use Spot

27. Optimizing Amazon EC2 cost and capacity
We continue to innovate for our customers
Pricing
Achieve optimal
price/performance
with different
purchase models
Capacity Guidance
Cost and capacity
recommendations
enable ease of use
and save time

28. Amazon EC2 Cost Optimisation non-prod
100.0
71.4
35.7
29.8
0
20
40
60
80
100
24 x 7 24 x 5 12 x 5 10 x 5
% Running Time
Up to 70%
savings for non-
production
workloads

29. AWS Instance Scheduler
• AWS-provided solution
• Custom start & stop schedules
• Works with EC2 & RDS instances
• Deploy using CloudFormation
• Selectively tag instances to schedule
• Multiple schedules per instance
• 5-minute granularity
https://aws.amazon.com/answers/infrastructure-
management/instance-scheduler/

30. Using Amazon EC2 Auto Scaling
Automatically scale instances across instance families
and purchase options in a single ASG to optimize cost
Capacity-optimized
Prioritize deploying Spot Instances into greater Spot pool capacity
order to lower the chance of interruptions
Lowest cost
Prioritize cost by selecting a mix of On-Demand and Spot Instances
to launch based on the lowest available price
Prioritized list
Use a prioritized list for On-Demand instance types to scale capacity
during an urgent, unpredictable event to optimize performance
Amazon EC2
Auto Scaling
AZ1 and AZ2

31. Amazon EC2 Spot Instance pools explained
$0.27 $0.29$0.50
1b 1c1a
8XL
$0.30 $0.16$0.214XL
$0.07 $0.08$0.082XL
$0.05 $0.04$0.04XL
$0.01 $0.04$0.01L
C4
$1.76
On
demand
$0.88
$0.44
$0.22
$0.11
Each instance family
Each instance size
Each availability zone
In every Region
Is a separate Spot pool
R5
M4
M5
I3 C5R4
i3en R5a
R5d

32. ASG capacity-optimized allocation strategy
us-east-1a
Desired capacity: 12
SpotAllocationStrategy: capacity-optimized
OnDemandBaseCapacity: 0 OnDemandPercentageAboveCapacity: 0
r5.large
m4.large
m5.large
R5 R5
R5 R5
us-east-1b us-east-1c
Overrides: [“r5.large”, “m4.large”, ”m5.large”]
$$
$
$$$
r5.large
m4.large
m5.large
$$$
$$
$
r5.large
m4.large
m5.large
$
$$$
$$

33. ASG lowest-price allocation strategy
us-east-1a
Desired capacity: 12
SpotAllocationStrategy: lowest-price
OnDemandBaseCapacity: 0 OnDemandPercentageAboveCapacity: 0
r5.large
m4.large
m5.large
R5 R5
R5 R5
us-east-1b us-east-1c
Overrides: [“r5.large”, “m4.large”, ”m5.large”]
$$
$
$$$
r5.large
m4.large
m5.large
$$$
$$
$
r5.large
m4.large
m5.large
$
$$$
$$

34. Instance type overrides and allocation strategies
ASG adjusts to new configuration
as it scales up and down
As ASG scales up
Launch capacity according to the new
configuration
As ASG scales down
Prioritize terminating instances not
matching the new configuration
New termination policy:
AllocationStrategy
Instance type overrides: m4.large, m5.large
m4.large m5.large
Instance type overrides: m5.large, c5.large
m4.large m5.large c5.large
Instance type overrides: m5.large, c5.large
m4.large m5.large c5.large

35. To optimize Amazon EC2, combine purchase options
RIs or a Savings Plan
Spot for fault-tolerant,
flexible, stateless workloads
On-Demand

36. Before: Multiple ASGs to use Spot, On-Demand, and RIs together
m4.large Spot ASG Min: 1 Max: 10
m5.large Spot ASG Min: 1 Max: 10
c4.xlarge O-D ASG Min: 1 Max: 10
Availability
Zone 1
Availability
Zone 2
Availability
Zone 3
Before, with
three ASGs
—one for each
instance type/
purchase option

37. Then: Spot, On-Demand, and RIs in a single ASG
m4.large Spot Instances
m5.large Spot Instances
c4.xlarge On-Demand instances
The new way
combines purchase
options, instance
types, and AZs in
a single ASG
Single ASGAvailability
Zone 1
Availability
Zone 2
Availability
Zone 3

38. m4.xlarge Spot
Weight of 1
m4.2xlarge Spot
Weight of 2
m4.4xlarge On-Demand
Weight of 4
Availability
Zone 1
Availability
Zone 2
Availability
Zone 3
Different
instance types
contribute
differently to
total capacity
Now: Spot, On-Demand, and RIs in a single ASG with weights

39. Amazon EC2 Spot Instance pools explained
$0.27 $0.29$0.50
1b 1c1a
8XL
$0.30 $0.16$0.214XL
$0.07 $0.08$0.082XL
$0.05 $0.04$0.04XL
$0.01 $0.04$0.01L
C4
$1.76
On
demand
$0.88
$0.44
$0.22
$0.11
16
Weight
8
4
2
1
$0.11
Weighted
price
$0.11
$0.11
$0.11
$0.11
$0.017 $0.018$0.032
1b 1c1a
$0.038 $0.02$0.032
$0.016 $0.02$0.021
$0.025 $0.04$0.02
$0.01 $0.04$0.01

40. API parameters
"MixedInstancesPolicy": {
"LaunchTemplate": {
"LaunchTemplateSpecification": {
"LaunchTemplateName": "MyLaunchTemplate"
},
"Overrides": [
{
"InstanceType": "m4.xLarge",
"WeightedCapacity": "1"
},
{
"InstanceType": "m4.2xLarge",
"WeightedCapacity": "2"
},
{
"InstanceType": "m4.4xLarge",
"WeightedCapacity": "4"
}
]
},
"InstancesDistribution": {
"OnDemandAllocationStrategy": "prioritized",
"OnDemandBaseCapacity": 10,
"OnDemandPercentageAboveBaseCapacity": 50,
"SpotAllocationStrategy": “capacity-optimized",
"SpotInstancePools": 2
}
}
AZ1 and AZ2
Desired
Min
Max
On-Demand base
50% On-Demand
50% Spot
Minimum On-Demand (10)

41. API parameters
"MixedInstancesPolicy": {
"LaunchTemplate": {
"LaunchTemplateSpecification": {
"LaunchTemplateName": "MyLaunchTemplate"
},
"Overrides": [
{
"InstanceType": "m4.xLarge",
"WeightedCapacity": "1"
},
{
"InstanceType": "m4.2xLarge",
"WeightedCapacity": "2"
},
{
"InstanceType": "m4.4xLarge",
"WeightedCapacity": "4"
}
]
},
"InstancesDistribution": {
"OnDemandAllocationStrategy": "prioritized",
"OnDemandBaseCapacity": 10,
"OnDemandPercentageAboveBaseCapacity": 50,
"SpotAllocationStrategy": “capacity-optimized",
"SpotInstancePools": 2
}
}
AZ1 and AZ2
Desired
Min
Max
On-Demand base
50% On-Demand
50% Spot
Minimum On-Demand (10)

42. API parameters
"MixedInstancesPolicy": {
"LaunchTemplate": {
"LaunchTemplateSpecification": {
"LaunchTemplateName": "MyLaunchTemplate"
},
"Overrides": [
{
"InstanceType": "m4.xLarge",
"WeightedCapacity": "1"
},
{
"InstanceType": "m4.2xLarge",
"WeightedCapacity": "2"
},
{
"InstanceType": "m4.4xLarge",
"WeightedCapacity": "4"
}
]
},
"InstancesDistribution": {
"OnDemandAllocationStrategy": "prioritized",
"OnDemandBaseCapacity": 10,
"OnDemandPercentageAboveBaseCapacity": 50,
"SpotAllocationStrategy": “capacity-optimized",
"SpotInstancePools": 2
}
}
AZ1 and AZ2
Desired
Min
Max
On-Demand base
50% On-Demand
50% Spot
Minimum On-Demand (10)

43. API parameters
"MixedInstancesPolicy": {
"LaunchTemplate": {
"LaunchTemplateSpecification": {
"LaunchTemplateName": "MyLaunchTemplate"
},
"Overrides": [
{
"InstanceType": "m4.xLarge",
"WeightedCapacity": "1"
},
{
"InstanceType": "m4.2xLarge",
"WeightedCapacity": "2"
},
{
"InstanceType": "m4.4xLarge",
"WeightedCapacity": "4"
}
]
},
"InstancesDistribution": {
"OnDemandAllocationStrategy": "prioritized",
"OnDemandBaseCapacity": 10,
"OnDemandPercentageAboveBaseCapacity": 50,
"SpotAllocationStrategy": “capacity-optimized",
"SpotInstancePools": 2
}
}
AZ1 and AZ2
Desired
Min
Max
On-Demand base
50% On-Demand
50% Spot
Minimum On-Demand (10)

44. API parameters
"MixedInstancesPolicy": {
"LaunchTemplate": {
"LaunchTemplateSpecification": {
"LaunchTemplateName": "MyLaunchTemplate"
},
"Overrides": [
{
"InstanceType": "m4.xLarge",
"WeightedCapacity": "1"
},
{
"InstanceType": "m4.2xLarge",
"WeightedCapacity": "2"
},
{
"InstanceType": "m4.4xLarge",
"WeightedCapacity": "4"
}
]
},
"InstancesDistribution": {
"OnDemandAllocationStrategy": "prioritized",
"OnDemandBaseCapacity": 10,
"OnDemandPercentageAboveBaseCapacity": 50,
"SpotAllocationStrategy": “capacity-optimized",
"SpotInstancePools": 2
}
}
AZ1 and AZ2
Desired
Min
Max
On-Demand base
50% On-Demand
50% Spot
Minimum On-Demand (10)

45. API parameters
"MixedInstancesPolicy": {
"LaunchTemplate": {
"LaunchTemplateSpecification": {
"LaunchTemplateName": "MyLaunchTemplate"
},
"Overrides": [
{
"InstanceType": "m4.xLarge",
"WeightedCapacity": "1"
},
{
"InstanceType": "m4.2xLarge",
"WeightedCapacity": "2"
},
{
"InstanceType": "m4.4xLarge",
"WeightedCapacity": "4"
}
]
},
"InstancesDistribution": {
"OnDemandAllocationStrategy": "prioritized",
"OnDemandBaseCapacity": 10,
"OnDemandPercentageAboveBaseCapacity": 50,
"SpotAllocationStrategy": “capacity-optimized",
"SpotInstancePools": 2
}
}
AZ1 and AZ2
Desired
Min
Max
On-Demand base
50% On-Demand
50% Spot
Minimum On-Demand (10)

46. AWS and third-party integrations with Spot Instances
and EC2 Auto Scaling
Amazon EC2
Auto Scaling
Amazon EC2
fleet
Amazon EMR AWS
CloudFormation
AWS
Batch
AWS Thinkbox
Amazon Elastic
Container Service
Amazon Elastic
Kubernetes Service
AWS
Fargate
Amazon
SageMaker
AWS Elastic
Beanstalk

47. Optimizing Amazon EC2 cost and capacity
We continue to innovate for our customers
Pricing
Achieve optimal
price/performance
with different
purchase models
Capacity
Capacity management
made easy on the
broadest and deepest
compute platform
Guidance

48. AWS Compute Optimizer
Recommends optimal instances for Amazon EC2 and Amazon EC2 Auto
Scaling groups from 140+ instances from M, C, R, T, and X families
Applies insights Saves timeLower costs
performance

49. Simplifying compute optimization
AWS Compute
Optimizer
Identify optimal
AWS compute resources
for your workloads
Mettle scans your AWS
infrastructure and uses
machine learning to
automatically identify
optimal AWS resources
for your workloads
Identifies workload
characteristics and
profile based on the
data gathered
Matches the resource
requirements of your
workloads to optimal
AWS resources with
recommendations
Amazon
CloudWatch
metrics
EC2 Instance
EC2 Auto
Scaling groups
Helps you visualize
what-if scenarios
based on the
recommended
resources
AWS resources
metadata

50. Easy to choose with AWS Compute Optimizer
New services that recommend optimal AWS compute resources to reduce costs up to 25%
Recommends optimal EC2 instances
Optimizes performance and reduces costs by
making recommendations to help you
right-size compute to your workloads
Analyzes Amazon CloudWatch metrics and
considers Auto Scaling group configuration for
intuitive and actionable recommendations
Up to three recommendations per workload
Available at no additional charge

51. Workloads
on AWS
Analytics and big data
Databases
DevOps—CI/CD
Enterprise applications
IoT
Machine learning
Storage
Websites and web applications

52. Machine learning
DevOps—CI/CD
Websites and web applications
Workloads
on AWS

53. Machine learning (ML)
Get ML solutions to market faster with access to built-in algorithms,
ML frameworks, and custom models
Save up to 90% in training costs
with managed spot training
Automatically manages Spot
capacity on your behalf
All instance types, training models,
and configurations
Amazon SageMaker
managed spot training

54. Machine learning
DevOps—CI/CD
Websites and web applications
Workloads
on AWS

55. DevOps—CI/CD
Or, run Jenkins build jobs inside your Kubernetes
clusters and cost-optimize with Spot node groups
https://ec2spotworkshops.com/amazon-ec2-spot-cicd-workshop.html

56. CI/CD reference architecture
https://github.com/awslabs/ec2-spot-jenkins-plugin/
Users Application
Load Balancer
Jenkins master
(OD or RI)
Jenkins master
(OD or RI)
Jenkins agent
(Spot)
Jenkins agent
(Spot)
Spot fleet
Availability Zone B
Availability Zone A
VPC
Jenkins agent
(Spot)
Jenkins agent
(Spot)

57. Machine learning
DevOps—CI/CD
Websites and web applications
Workloads
on AWS

58. Websites and web applications
Run web services ranging from ad servers to real-time bidding servers
Deploy web applications or services on containers and scale clusters at a fraction of the cost
Use Auto Scaling with Amazon ECS or Amazon EKS to run any containerized workload,
including a web application
Amazon EC2
Auto Scaling
Amazon Elastic
Container Service
Amazon Elastic
Kubernetes Service
Scale in real time, pay per second, save up to 90%
AWS
Fargate

59. AWS Fargate with EC2 Spot
Up to 70% off over regular
Fargate tasks
Only pay for the resources you use
by automatically scaling based on
tasks, vCPUs, and memory
VM-level isolation by design
Run containers without managing servers or clusters
AWS Fargate

60. To tie it all together…

61. Key takeaways
Get technical
guidance in an AWS
Immersion Day
4
Experiment and test
at a lower cost to
innovate faster
1
How to automate
cost and capacity
optimization
2
Optimize your
workloads by using
best practices
3
CI/CD, analytics,
big data, machine
learning, and
web services
Spot Instances Auto Scaling and
Savings Plans
AWS Compute
Optimizer

62. Schedule an AWS
Immersion Day
AWS experts are here to help,
and it’s free!

63. Learn compute with AWS Training and Certification
20+ free digital courses cover topics related to cloud compute,
including introduction to the following services:
Resources created by the experts at AWS to help you build cloud compute skills
Compute is also covered in the classroom offering, Architecting
on AWS, which features AWS expert instructors and hands-on
activities
• Amazon Elastic Compute Cloud
(Amazon EC2)
• Amazon EC2 Auto Scaling
• AWS Systems Manager
• AWS Inferentia and Amazon EC2
Inf1 instances
Visit the learning library at https://aws.training

64. Thank you!
© 2020, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Cobus Bernard
Sr Developer Advocate
Amazon Web Services
@cobusbernard
cobusbernard
cobusbernard