7 Steps to

Save
Big
on Your AWS Bill
A Practical Guide to Cloud Cost Optimization

www.n2ws.com
Overview
The public cloud offers many advantages over the traditional model of IT – through low-cost
infrastructure, fast provisioning and access to cutting-edge cloud-based technologies.

But only if you adapt to the new on-demand pay-as-you-go model of the cloud and learn how to manage
your resources efficiently.

Without an effective cost-optimization strategy, you can end up wasting money on

unused and underutilized cloud resources.

You can miss out on opportunities for significant discounts on your monthly cloud bills. And you may not
choose the most cost-efficient solution for your applications.

This paper provides you with a strategy for reducing your costs on leading cloud platform AWS. But it also
serves as a practical guide to cloud cost optimization – with diagrams, screenshots and links to additional
resources, so you can start taking cost-saving measures straight away.

Let's start by looking at the optimization tools you'll need for the job.

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 02

Table of Contents

1. Nurture Cost Awareness 09

2. Ditch Unused Resources 11

3. Fine-Tune Your Compute Resources 14

4. Optimize Your Storage 17

5. Use Auto Scaling 23

6. Adopt an RI Strategy 26

7. Purchase Spot instances 29

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 03

Cost-Optimization
Toolkit
Monthly AWS Bill
Your monthly billing report is key to understanding what's driving spending.

Use it to identify the services that are costing you the most, as these are the low-hanging fruit where you
can make the biggest savings.
You can also use it to check whether you're making good use of any Reserved Instances (RIs) – by
comparing the total usage available with the total number of hours you actually consume.

What's more, you can track spending and monitor the progress of your cost-optimization regime.
In the case of consolidated bills, line items are broken down by account, product and region. Bills for
individual accounts are simply organized by product and region.

AWS CloudWatch
An infrastructure monitoring service that provides insights for managing your
AWS resources.

You can use it to visualize log data from your EC2 deployments. You can monitor estimated charges for
individual AWS services and set up email alerts that notify you when costs exceed a certain threshold.
You can also create alarms and take automated actions whenever resource utilization reaches predefined
thresholds.

CloudWatch provides CPU utilization data for EC2 instances by default. But you can also set up custom
metrics to monitor memory utilization.

AWS Cost Explorer

A cost analysis and forecasting tool, which helps you make sense of your costs across
a complex array of resources, providing high-level visualizations of spending patterns
over periods of up to 12 months.

Cost Explorer includes a suite of default reports that help you quickly compare costs between different
products, accounts or regions.

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 04

In addition, a Reserved Instance (RI) reporting feature shows you how well you're utilizing your RIs and
how much of your overall instance usage is covered by them.

You can also dig deeper by filtering and grouping your data as well as creating customized reports.

AWS Trusted Advisor

An AWS online management tool that makes a series of best-practice checks aimed at
helping you reduce cost and improve system performance, reliability and security.

The cost-optimization category includes checks and alerts for:

Underutilized EC2 and RDS instances

EBS volumes with little or no write activity
Idle load balancers
Unused Elastic IP addresses

It also makes RI recommendations based on your last month's instance usage patterns and alerts you to
RIs that have recently expired or due to expire in the next 30 days.

EC2 Right Sizing

An implementation recommended by AWS that provides detailed recommendations for
right-sizing your Amazon EC2 instances base on utilization over a two-week period.

The solution is based on a CloudFormation template, which launches a series of components that form
the application stack. Setup requires technical knowledge of working on the AWS platform.

Amazon S3 Analytics – Storage Class Analysis

An S3 storage optimization tool, which helps you see how much of
your data you access as it ages over time.

It filters your data into a series of predefined age groups, starting from objects less than 15 days old
through to objects 730 days and older, and presents a visualization for each group showing how much
data you stored during that time period and how much of it you retrieved.

Storage Class Analysis is intended as a visual aid for making decisions on when to transition objects
from S3 Standard to the lower-cost S3 Standard – IA tier.

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 05

Amazon QuickSight
QuickSight lets you dig deeper into your S3 access patterns by providing more detailed
graphical insights that you can use as a basis for setting your S3 lifecycle policies.

It includes a number of default visualizations, which provide side-by-side comparisons of storage

consumption and retrieval across each of the different predefined age groups.

You can switch between different visualization formats, such as multi-line graphs, pie charts and bar
charts, and also create your own visualizations.

AWS Instance Scheduler

An AWS implementation, based on a CloudFormation template, for setting custom
stop-and-start schedules on your EC2 and RDS instances.

The solution helps you reduce your monthly bills by only running instances during the periods you actually
need them. For example, Dev/Test resources are rarely used outside regular 9–5 hours.

So you can define a schedule that closes them down in the evenings, at weekends and during public
holidays. Setup is relatively simple but requires some technical knowledge of working on the AWS platform.

AWS Lambda
A fully managed event-driven serverless service, which you can use to deploy code that
runs in response to CloudWatch events.

This gives you the scope to automate some of your cost-optimization processes. For example, you could set
up a Lambda function to automatically find and delete unused EBS volumes or delete snapshots when they
reach the end of a specified retention period.

Lambda is also a component of the automated AWS Instance Scheduler deployment.

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 06

Spot Instance Advisor
A simple online reference tool for weighing up the likelihood of a Spot Instance
interruption against the savings you can make over the equivalent on-demand instance.

It is particularly useful if you're relatively new to Spot Instances, as it can help you identify low-risk
instance types for applications that don't yet have sufficient flexibility or fault tolerance to cope with
higher levels of interruption.

Tag Editor
Tags facilitate visibility over your infrastructure by helping you quickly identify who is
responsible for each of your resources what they use them for. They also play an
important role in lifecycle management.

You can tag resources as and when you create them. You can also manage tags for an individual resource
within the associated AWS console for that resource.

But if you need to centrally manage a number of tags at any one time, it's easier to use the AWS Tag
Editor. With the Tag Editor, you can search through your inventory of taggable resources and then
change, remove or apply new tags.

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 07

AWS Command Line Interface (CLI)
The AWS CLI gives you the power to directly manage all your AWS infrastructure from
a single unified interface.

You can perform many of your cost-optimization tasks quickly and efficiently by running a single command
or script. For example, you can view Spot Instance pricing history, scale elastic EBS volumes and delete
old EBS snapshots without leaving the same terminal session. Although the AWS CLI is a very powerful
tool, it takes a lot of time and learning to reach a useful level of proficiency.

aws ec2 delete-snapshot --snapshot-id snap-1234567890abcdefg

Example of a delete-snapshot command in the AWS CLI

Third-Party Tools
In addition to Amazon's own set of cost-management solutions, you should also con-
sider third-party tools, which typically offer additional functionality, such as enhanced
best-practice checks, multi-cloud capabilities and higher levels of automation.

These include conventional cost-optimization tools, which provide features such as right-sizing, RI recom-
mendations and budget reporting.

But other types of third-party software can also help you reduce your monthly AWS bills. For example,
data management tools can help you take advantage of the most cost-efficient storage methods by giving
you the flexibility and control to move data around your AWS environment. Moreover, a solid tool will
allow you to stop and start Amazon EC2 instances at predefined times to reduce the unnecessary costs of
idle servers.

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 08

ST EP 01

Nurture Cost
Awareness
"The ability to scale up and down on demand has allowed resource procurement to
transition from being solely owned by the finance team to being the responsibility of
stakeholders across IT, engineering, finance, and beyond. This democratization of resource
procurement has initiated an ever-growing group of cost-conscious stakeholders."

- Amazon Web Services, March 2018

Nurture Cost Awareness
Whatever public cloud provider you use, nurturing cost awareness should be one of you
top cloud priorities.

With the on-demand utility model of the cloud, it's easy to provision computing resources. But, at
the same time, it's just as easy to leave them running when you no longer need them.

Not only that, but you no longer need large upfront investments to fund new infrastructure. As a
result, where IT spending was previously the sole responsibility of central finance, it's now
devolved to a much broader spectrum of departments.

This calls for a change in culture towards a far more cost-aware mindset. It requires governance
over and accountability for cloud usage. But it also requires a cost-aware approach to application
architecture.

Governance and Accountability

You can meet your governance and accountability objectives by:

Enforcing tagging policies so everyone knows what each resource is used for.
Accurately allocating costs to individual business units.
Regularly reviewing your AWS costs and the success of your cloud transformation.

Cost-Aware Architecture
Your application design should take advantage of the most cost-efficient AWS services
wherever possible.

You should also make full use of the cloud's elastic features by breaking your applications up into smaller,
individual, loosely coupled components.

This will help you scale your applications more efficiently, as you can increase and decrease the capacity
allocated to each component independently, providing finer grained matching of discrete application
functions to resource requirements.

Similarly, you can deploy different components to AWS instances in different families. That way, you can
achieve more granular control by matching each component to an instance type with just the right mix of
CPU and memory.

What's more, you're better able to scale your applications horizontally. This is the method whereby you
deploy each component to a cluster of virtual machines (VMs) and scale in or out by removing or adding
machines. It is a cost-efficient way to scale your infrastructure because you can use a network of smaller
instances and adjust it by smaller, more precise increments.

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 10

ST EP 02

Ditch Unused
Resources
Now we're onto the cost-saving measures that can make an immediate impact on
your AWS bill. The best place to start is to clean up all unused resources, as these
eat away at your budgets without serving any purpose.

The following core AWS services should be the first on your agenda:
EC2 Instances
Idle EC2 instances are amongst the biggest causes of inflated AWS bills.

But you can prevent unnecessary waste by using CloudWatch alarms to automatically stop or terminate
instances that have been idle for longer than a specified period. You can only terminate machines backed
by Instance Store. But you can either stop or terminate instances backed by EBS – depending on whether
you're likely to need them again.

For example, you may want to stop instances where you only need them during normal working hours.
However, you may want to terminate others altogether if they remain unused for a longer period of time.

You can set up CloudWatch alarms in the:

EC2 dashboard in the AWS Management console

CloudWatch console
AWS CLI by running the put-metric-alarm command

This gives you the scope to automate some of your cost-optimization processes. For example, you could set
up a Lambda function to automatically find and delete unused EBS volumes or delete snapshots when they
reach the end of a specified retention period. Lambda is also a component of the automated AWS Instance
Scheduler deployment.

EBS Volumes
By default, when you terminate an EBS-backed instance, only the EBS volumes that were
attached at launch are deleted. Any additional volumes you attached after you launched
the instance continue to run.

This can cost you big if you leave unattached EBS volumes running for longer than necessary –
especially where you use SSD-backed volumes, which cost more than twice as much as their
HDD counterparts.

You can manually find and remove unattached EBS volumes in the:

EC2 dashboard in the AWS Management console

AWS CLI by running the describe-volumes and delete-volume commands

Don't forget to take a snapshot of your volume before you terminate it. This gives you the option to
recover it should you need it again in the future. You can take snapshots in the EC2 dashboard from the
same menu you use to delete volumes.

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 12

Orphan Snapshots
Just as any additional EBS volumes remain after you terminate an instance, so do
your snapshots.

Although snapshots are incremental, your first backup captures the entire volume. What's more, if your
retention period is long, you may need to store as much incremental data as the initial snapshot.

In other words, your snapshots may be costing you far more than you think. So, unless you want to use
your orphaned snapshots to create new EBS volumes, you should delete them as part of your
cost-optimization routine. You can manually find and remove orphan snapshots in the:

AWS CLI by running the describe-snapshots and delete-snapshot commands

Idle Elastic Load Balancers (ELBs)

You should periodically check the status of each load balancer to ensure it's actively serving registered
instances. If it isn't then either register instances to it or delete it. You can manually find and remove idle
load balancers in the:

EC2 dashboard in the AWS Management console

AWS CLI by running the describe-load-balancers and delete-load-balancer commands

Elastic IPs
Public Internet addresses are a limited resource. So, to discourage waste, AWS restricts
each account to five Elastic IP addresses per region.

It also levies a small hourly charge on addresses that aren't associated with a running instance. The same
goes for Elastic IPs associated with a stopped instance or unattached network interface. To keep costs
down, you should release Elastic IPs if you no longer need them. You can manually find and release
unattached Elastic IPs in the:

VPC dashboard of your AWS Management console

AWS CLI by running the describe-addresses and release-address commands

Tools to Consider: Where to Learn More:

Trusted Advisor Create Alarms to Stop, Terminate, Reboot or Recover an Instance
CloudWatch Terminating Your EC2 Instance
AWS Lambda Trusted Advisor Cost-Optimization Best Practice Checks
AWS Elastic IP Addresses
Deleting an Amazon EBS Snapshot

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 13

ST EP 03

Fine-Tune Your
Compute Resources
Size matters when it comes to cloud cost efficiency. If your instances provide far
more CPU and memory than your workloads need then you're simply pouring
money down the drain.

But if they don't provide enough then your application performance will suffer.
Fine-Tune Your Compute Resources
Size matters when it comes to cloud cost efficiency.

What capacity do you need to meet average consumption? And how much additional power do you need
to cope with peaks in demand?

The answers to these questions will help you determine the optimum utilization level of your instances.
But, as a rule of thumb, you should generally aim for 70–80% utilization if you want to minimize costs
without affecting latency.

But right-sizing your instances shouldn't end there. You can fine-tune your resources by switching to
instances with different performance characteristics.

The following table highlights common examples where a different instance family may be a better fit for
your application:

Use Case Switch to

Compute-intensive workloads requiring a higher

Compute-optimized family of instances (C5 and C4)
proportion of CPU

Memory-intensive workloads requiring a higher Memory-optimized family of instances (R5, R5a, R4,
proportion of memory X1e, X1, High Memory and z1d)

Disk-intensive workloads requiring locally

Storage-optimized family of instances (H1, I3 and D2)
attached storage

Small and moderate workloads requiring occasional Burstable Performance family of instances
additional capacity to meet peaks in demand (T2, T3 and T3a)

You should also collect and analyze usage statistics on a regular basis and review your EC2 capacity over
time. And, where possible, you should deploy your application to a fleet of smaller instances.

So, instead of scaling vertically by changing the size of your instance, you can scale your infrastructure
horizontally a little at a time by adding or removing lower-cost machines.

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 15

 T3 Burstable (General Purpose) Instance Pricing
US-East (N. Virginia)

Baseline
Price / Hour Price / Hour
Name vCPUs Performance Memory
(Linux) (Windows)
/ vCPU

t3.nano 2 5% 0.5 GiB $0.0052 $0.0098

t3.micro 2 10% 1 GiB $0.0104 $0.0196

t3.small 2 20% 2 GiB $0.0209 $0.0393

t3.medium 2 20% 4 GiB $0.0418 $0.0602

t3.large 2 30% 8 GiB $0.0835 $0.1111

t3.xlarge 4 40% 16 GiB $0.1670 $0.2406

t3.2xlarge 8 40% 32 GiB $0.3341 $0.4813

You can access statistics for CPU and memory consumption of a specific instance in the:

CloudWatch console
AWS CLI by running the get-metric-statistics command

If you're an Amazon RDS user, you should be aware that DB instances are decoupled from storage. So,
from time to time, you may need to switch to a different DB instance type to meet changing demand on
your database system.

Tools to Consider: Where to Learn More:

Cost Explorer, Amazon EC2 Instance Types
Trusted Advisor, EC2 Pricing
EC2 Right Sizing, Cost Optimization: EC2 Right Sizing
CloudWatch EC2 Right Sizing: AWS Implementation Guide
Monitoring Memory and Disk Metrics for Amazon EC2 Linux Instances
Tagging Your Amazon EC2 Resources

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 16

ST EP 04

Optimize Your
Storage
“Organizations tend to think of data storage as an ancillary service and do not optimize
storage after data is moved to the cloud. Many also fail to clean up unused storage and
let these services run for days, weeks, and even months at significant cost.”

- Amazon Web Services, March 2018

EBS Volumes
In February 2017, AWS changed the way EBS works, making it possible to adjust the
size of current-generation volumes without detaching them first.

This now makes it much easier to optimize your block storage capacity and manage costs. To monitor
the disk utilization of your EBS volumes, you'll need to publish custom usage metrics to CloudWatch.
You can modify the size of an EBS volume in the:

EC2 dashboard in the AWS Management console

AWS CLI by running the modify-volume command

EBS Snapshots
In addition to removing orphan snapshots, you should maintain a snapshot retention
policy that properly balances your data backup requirements with cost.

Deleting an old snapshot won't always necessarily reduce your storage overhead. This is because it may
contain referenced data that's required to restore the volume. If this is the case then the data is preserved
and continues to rack up charges.

However, if a good proportion of the data on your volumes changes frequently then deleting old snapshots
will remove duplicate files and help to keep your storage costs down.

AWS compresses your snapshots and stores them in S3. So your snapshot costs may seem relatively
inexpensive. But the vendor uses a different pricing system from its regular S3 tiers. This is more than
twice the Standard S3 rate.

Not only that, but the buckets AWS uses to store snapshots are not publicly visible. So you cannot retrieve
them using the standard S3 methods.

Currently, the only way to retain snapshots at a lower cost is to use a third-party cloud data management
tool that's able to copy them to a standard S3 bucket.

Amazon S3
Amazon's object storage offering S3 is the cloud provider's most cost-efficient way to
store data. So, to keep your bills down, you should make use of it wherever possible.

This means limiting your use of more expensive services, such as EBS and EFS, to only those use cases
where they're strictly necessary. S3 is not only the least expensive storage solution, but it also
automatically scales with you. So you don't have to worry about wasting unused provisioned storage.

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 18

The service is available in three main storage classes aimed at different use cases:

Standard
Low-latency storage option with 99.999999999% durability and 99.99% availability.
Designed for frequently accessed data.

Standard – IA
The same low latency and durability as S3 Standard but availability slightly lower at 99.9%. Standard
– IA is designed for infrequently accessed data, as storage charges are lower but data retrieval
charges higher.

Objects can exist in the same bucket as those set as S3 Standard. This means you can transition data
between classes without making application changes.

Glacier
The most economical storage option. However, data retrieval is slow and expensive, making it
suitable only for data archiving and long-term backups.

Example Storage Pricing by Class

US-East (N. Virginia)

S3 Standard S3 Standard - IA Glacier

First 50 TB/month $0.023 per GB $0.0125 per GB $0.004 per GB

First 450 TB/month $0.022 per GB $0.0125 per GB $0.004 per GB

First 500 TB/month $0.021 per GB $0.0125 per GB $0.004 per GB

Charges for data/requests/retrieval also apply

Efficient lifecycle management of your S3 objects is key to lowering your storage costs. So AWS provides a
number of tools to help you manage your buckets and take advantage of its different storage classes.

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 19

Object Tagging
Tags help you better understand your S3 bucket usage. They can be used to filter objects
both as part of your lifecycle analysis and for configuring your lifecycle policies.

You can tag an S3 object in the:

S3 dashboard in the AWS Management console

Tag Editor
AWS CLI by running the put-object-tagging command

Setting up a lifecycle rule in the S3 console

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 20

Amazon S3 Analytics – Storage Class Analysis
Before you configure your lifecycle policies, you'll need to understand the patterns of
access to your data over time.

Storage Class Analysis is a simple data visualization tool that can play a useful role in your S3 optimization
workflow. Although it provides a basic overview of retrieval patterns, it may be all you need to decide at
what age your data is ready for transition between tiers.

You can configure Storage Class Analysis in the:

S3 dashboard in the AWS Management console

Typical dashboard output from Storage Class Analysis

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 21

Amazon QuickSight
Amazon QuickSight is a business intelligence (BI) and data visualization service with a
pre-built integration for S3 Storage Analytics.

It provides more detailed views of your storage retrieval patterns for fine-tuning your S3 lifecycle policies.
If you have Storage Class Analysis enabled, you can access Amazon QuickSight via the:

QuickSight link in the S3 dashboard

The QuickSight dashboard

The QuickSight dashboard

Tools to Consider: Where to Learn More:

Tag Editor Amazon EBS Update – New Elastic Volumes Change Everything
Amazon S3 Analytics – Storage Class Analysis Amazon EBS Pricing – Snapshots to Amazon S3
Amazon QuickSight Amazon S3 Pricing
Amazon S3 Storage Classes
S3 Lifecycle: Supported Transitions and Related Constraints
How to Set Up and Use AWS Storage Class Analysis
Amazon QuickSight
Setting Up Amazon QuickSight

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 22

ST EP 05

Use Auto Scaling

EC2 Auto Scaling is a horizontal scaling feature, which automatically spins up
new instances as the load on your application increases and terminates them
when demand goes down.
Use Auto Scaling
You can adjust the size of your cluster manually, based on a schedule or dynamically based on a target
value for a specific metric, such as CPU utilization.

Dynamic scaling is a particularly effective cost-optimization technique, as it can take almost immediate
action to changes in demand. By contrast, it can take time to manually respond to scaling events – while, in
the meantime, your application can suffer performance issues or rack up unnecessary costs.

Auto Scaling Groups

A cluster of EC2 instances is known as an Auto Scaling group, where each group is defined by the following
three key components:

Minimum size
The baseline number of instances below which the size of your group should never fall.

Desired capacity
The number of instances you want to run based on your scaling policy. Your desired capacity moves up
and down dynamically as the load on your application changes. Desired capacity adjustments are
triggered by CloudWatch alarms.

Maximum size
The ceiling figure above which the number of instances in your group should never exceed.

Minimum Size Scale out as needed

Desired capacity

Maximum size

A simple Auto Scaling group

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 24

Before you create an EC2 Auto Scaling group, you'll first need a launch configuration, which is a template
for the instances your group will use.

You can set up an EC2 Auto Scaling group in the:

EC2 dashboard in the AWS Management console

AWS CLI by running the create-auto-scaling-group command

Top Tip
Auto Scaling is also available for other AWS services, including ECS, EC2 Spot Fleets,
Aurora, DynamoDB and EMR.

You can configure all of these resources using AWS Auto Scaling – a single unified interface that simplifies
Auto Scaling management across your services without having to navigate to other consoles.

Tools to Consider: Where to Learn More:

CloudWatch What Is AWS Auto Scaling?
ELB Benefits of Auto Scaling
Creating an Auto Scaling Group Using a Launch Configuration
New AWS Auto Scaling – Unified Scaling for Your Cloud Applications
Using Elastic Load Balancing Health Checks with Auto Scaling

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 25

ST EP 06

Adopt an
RI Strategy
Reserved Instances are an alternative billing mechanism to the on-demand
pricing model, where you purchase a discounted usage credit for a specific type
of instance over a fixed term.
Adopt an RI Strategy
They can deliver savings of up to 75% compared with standard on-demand charges, where the maximum
discount you can expect depends on the following:

RI specifications
The instance type, operating system (OS), tenancy and Availability Zone (AZ).

RI term
Either one or three years

Payment option
All upfront, partial upfront or no upfront payment at all.

Whether you purchase a Standard RI or Convertible RI

Standard RIs offer higher discounts but limited flexibility to modify your instance specifications.
Convertible RIs promise lower discounts but give you the option to exchange your RI for another in
a different instance family or with a different OS, tenancy or payment option.

For maximum return on your RI purchases, your running instances should consume your usage credits
as much of the time as possible. This is because, during any period where you have no active matching
instances, your credit will remain unused and go to waste. This is why RIs tend to suit steady predict-
able workloads.

RIs are a more effective cost-optimization strategy when you use consolidated billing. This is because
an RI credit can apply to a matching instance in any of a company's linked accounts. However, the
account that buys the reservation is always first in line to the discount.

Another strategy for maximizing RI savings is to use them to cover the baseline capacity of an Auto
Scaling group.

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 27

 Reserved Reserved On-demand On-demand On-demand
Instance Instance Instance Instance Instance

Minimum Size Scale out as needed

Desired capacity

Maximum size

Use RIs to cover the capacity required for your minimum group size

As your Auto Scaling group will always run at or above your minimum number of instances, you can effec-
tively guarantee full utilization of your matching reservations.

Other AWS services with reservation models include RDS, Redshift, ElastiCache and DynamoDB.

You can purchase EC2 Reserved Instances in the:

EC2 dashboard in the AWS Management console

AWS CLI by running the purchase-reserved-instances-offering command

Top Tip
Don't forget to monitor your RIs so you can allocate other workloads to them if
they're unused or underutilized. And keep track of their expiry date, as AWS doesn't
renew them automatically.

Tools to Consider: Where to Learn More:

Tag Editor Amazon EC2 Reserved Instance Pricing
Cost Explorer AWS Reserved Instance (RI) Reporting
Trusted Advisor Modifying Reserved Instances
Understanding Consolidated Bills
AWS Consolidated Billing: How to Allocate EC2 Reserved Instance Costs

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 28

ST EP 07

Purchase
Spot Instances
Spot Instances are a disposable class of instances, which are priced based on
supply of and demand for EC2 spare capacity.
Purchase Spot instances
Using Spot Instances can potentially bring the biggest savings to your AWS monthly bill, with discounts
of up to 90% compared with standard on-demand machines.

When you purchase a Spot Instance, you're simply charged the current market price for the machine type
you need – subject to an optional maximum price you're willing to pay.

However, AWS can interrupt your Spot Instance if it doesn't have enough capacity to meet demand. It
can also do so if the current Spot price is greater than your maximum price or your request includes a
constraint, such as a launch group or an AZ group, it cannot meet.

Although interruptions are relatively infrequent, they can still happen at any time. So Spot Instances
aren't suitable for mission-critical applications that rely on continuous availability.

Interruption Behavior
You can specify how EC2 interrupts your Spot Instances.

The default interruption behavior is to terminate them. But you also have the option to either stop or
hibernate them.

When the Spot service stops or hibernates your instances, it preserves your EBS volumes. However,
hibernation also preserves instance memory on the root volume. This means your workloads can resume
automatically when your Spot capacity becomes available again.

Currently, hibernation is supported by a limited number of instance types. It also requires installation of
the EC2 Hibernation Agent on your guest OS.

You can make a Spot Instance request in the:

EC2 dashboard in the AWS Management console

AWS CLI by running the request-spot-instances command

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 30

More Predictable Spot Instance Pricing
In 2017, AWS launched a new Spot Instance pricing model that replaced the previous bidding
mechanism with the current simpler system. This reduced both fluctuations in prices and the number
of interruptions.

Spot Instance prices are less volatile under the new pricing model

Usage Pattern Cost-Saving Measure Maximum Potential Savings

Steady state Reserved Instances 75%

Variable but predictable workload Auto Scaling Subject to use case

Dev/Test environment Instance Scheduler 70%

Temporary deployment Spot Instances 90%

At-a-glance optimization measures based on usage patterns

Tools to Consider: Where to Learn More:

Spot Instance Advisor Amazon EC2 Spot Instances Spot Instance Requests
Amazon's New EC2 Spot Pricing Model Managing Spot Interruptions
Spot Instance Advisor How Spot Fleet Works
How to View Spot Instance Pricing History

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 31

"Over a period of six years, the average total cost
per minute of an unplanned outage has risen by
58% – from $5,617 per minute in 2010 to
$8,851 per minute in 2016."

Ponemon Institute Cost of Data Center

Outages report, January 2016

Avoid the Rising Cost of Downtime

A carefully coordinated cost-optimization strategy is essential to the efficient running
of your AWS resources. But it's important you never lose sight of the bigger picture.

Downtime can end up being a far greater cost to your business – through reputational damage, loss of
productivity and loss of revenue. So it's equally essential to have backup and recovery measures in place to
help you maintain business continuity and keep downtime costs down.

Third-party data management tools are designed to simplify the backup and recovery process, so you're
not scrambling for the AWS documentation whenever you need to restore your cloud infrastructure.

But data management tools don't just help in potential disaster situations. They also help you reduce your
AWS bills by providing continuous cloud storage optimization.

AWS Cost Optimization - 7 Steps to Save Big on Your AWS Bill 32

N2WS Backup
& Recovery
Built for scale. Built for simplicity. Built for AWS.
As more companies move to the public cloud, they need a way to ensure their critical data is always

instant recovery button to protect your data from any outage any time.

Top Rated Backup & Recovery solution in AWS Marketplace

Automated Backup + Instant Recovery = No Downtime + No Worries

• Automate backups and get 1-click recovery for Amazon EC2 instances and EBS volumes
•

Your Database Protection Plan: Always Available, Application-Consistent

• Capture the most complete database picture with application-consistent backups for Amazon
RDS databases engines: Amazon Aurora, PostgreSQL, MySQL, MariaDB, Oracle, MSQL
• Enjoy complete support for NoSQL databases: Cassandra, DynamoDB, Mongo DB

Savvy Data Lifecycle Management to Save Your AWS (and Your Budget)
• Save up to 60% on long-term retention costs by decoupling and transferring your EBS
snapshots to the N2WS S3 repository, with the ability to recover to any region or account
• Start, stop and hibernate groups of Amazon EC2 or Amazon RDS instances and save on
computing costs with N2WS LightSwitch

About N2WS

environment, providing backup and disaster recovery functionality across regions and accounts.

Get Free AWS Backup

+1 888 426 4329 • [email protected] • www.n2ws.com