AWS Enterprise Summit Netherlands - Cost Optimisation at Scale

© 2016, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Mario Thomas
Senior Consultant EMEA
AWS Professional Services
21 September 2016
Cost Optimisation at Scale
Reducing the cost of using AWS

What We’ll Cover Today
• A quick overview of TCO
• Does “at scale” apply to you?
• Applying “architecting for cost” principles to your TCO
analysis and your AWS account(s)
• Recap

What is TCO?
Definition: Comparative total cost of ownership analysis (acquisition
and operating costs) for running an infrastructure environment end-to-end
on-premises vs. on AWS.
Used for:
1) Comparing the costs of running an entire infrastructure environment or
specific workload on-premises or in a co-location facility vs. on AWS
2) Budgeting and building the business case for moving to AWS

AWS Includes Everything In It’s Price
Hardware
in a Data
Center

Cost is Rarely the Only Reason to Choose AWS
Pay-as-you
Go Model
Lower Overall
Costs
Stop
Guessing
Capacity
Agility / Speed
/ Innovation
Avoid
Undifferentiated
Heavy Lifting
Go Global in
Minutes
✔ ✔ ✔ ✔ ✔ ✔
On-
Premise/
Co-Location
X X X X X X

Businesses are migrating IT to cloud
I'm not going to sell another aircraft
engine because I run a global data
center operation really well. That's
AWS's differentiator in the
environment.
Jim Fowler, CIO General Electric (2015)
2014 2017
34 Data
Centers
4 Data
Centers

Cloud can be a better solution that build
We work closely with AWS to develop
a security model, which we believe
enables us to operate more securely in
the public cloud than we can in our
own data centers.
Rob Alexander, CIO Capital One (2015)
2014 2017
8 Data
Centers
3 Data
Centers

Trend: Retirement of Legacy Technology Debt
From 6 Data Centers to 2.
Time to market reduced by
6 months.
Productivity up 100% in
front end-processes; up
40% in back-end processes.
Closed 8 Data Centers.
Queries 98% faster, 80%
cheaper than previous data
warehouse solution.
Data Center server footprint
cut 91%.
$ 3.5M saved over 5 years.
75% ops migrated to AWS.
$ 20M annual savings.
Saved 50% in CAPEX by
exiting Data Centers.
From 40 Data Centers to 6.
25% cost savings vs. colo.
30% faster development
Migrated 500+ servers,
closed Data Centers, cut
costs by 40%, grew
productivity over 30%.
(Italy) Reduced CAPEX
30%. PCI-DSS compliant
within 3 months.

So how do we do it?
≠

Traditional Data Centers are costly and inefficient…
Initial Fixed
Capacity
Utilization
Unused Capacity = Wasted $
Downtime, Lost Customers, Lost Revenue (Impossible to measure)
More Wasted $
Time

TCO = acquisition costs + operations costs
Hardware – Server, Rack
Chassis PDUs, ToR
Switches (+Maintenance)
Software - OS,
Virtualization Licenses
(+Maintenance)
Facilities Cost
Hardware – Storage Disks,
SAN/FC Switches
Storage Admin costs
Network Hardware – LAN
Switches, Load Balancer
Bandwidth costs
Network Admin costs
Server Admin / Virtualization Admin4
Diagram doesn’t include every cost item. For example, software costs can include database, management, and middle-tier
software costs. Facilities cost can include costs associated with upgrades, maintenance, building security, taxes, etc. IT labor
costs can include security admin and application admin costs.
Space Power Cooling
Facilities Cost
Space Power Cooling
Facilities Cost
Space Power Cooling
Server Costs
Storage Costs
Network Costs
IT Labor Costs
1
2
3

Efficiency Category Efficiency p/app % of Total
Business Productivity 181,669.00$ 41%
IT Staff Productivity 133,389.00$ 30%
IT Infrastructure Cost Reduction 131,073.00$ 29%
Total 446,131.00$ 100%
Source: 2015 Study on AWS Value
The opportunity cost of on-premise IT
• Only 29% of the benefits of AWS Cloud are from IT cost
reduction
• 41% of the AWS Cloud opportunity is increased productivity
IDC projects that
organizations using AWS
will realize increased
revenue and productivity
benefits worth an average
of $173,003 per application
per year over five years, or
$7.04 million per
organization

…and that these benefits increase over time
Source: IDC Business Value of AWS Accelerates over
time
According to IDC, this relationship between length of time using AWS and return is due to customers leveraging the more optimized
environment to generate more applications along a learning curve.
$1 Investment in AWS
$8.40 in benefits
At 60 Months of using AWS
~8X
$3.50 in benefits
$1 Investment in AWS
At 36 Months of using AWS
~3X

The TCO Window
$
1 2 3 4 50
TCO
Migration Cost
Cost Optimising / BAU
Current / Do Nothing
AWS Environment
Payback
Period
Time (Months)
Cost

Cost Optimisation
$ $
Paying for what
you use
Paying for what
you need

The four pillars of cost optimisation
Right sizing Reserved
Instances
Increase
elasticity
Measure,
monitor, and
improve

Cost Optimisation
Simple Stimulating Stretching
Consolidated Billing
Permissions
Tagging
Idle Resources
Design for Elasticity
Instance Right Sizing
Storage
Purchasing Options
OS Licencing
Offloading Architecture
Higher Level Services

• Receive a single bill for all charges incurred across all linked accounts
‒ Share RI discounts
‒ Combine tiering benefits
• View & manage linked accounts
• Add additional accounts

Permissions
How are you
controlling the
provisioning of
resources?
Control who can
provision resources

Tagging
• Use tagging to identify resources
• Use tags to audit resources
• Turn off untagged resources

Idle Resources
• Dev/Test (Non-Prod) instances
• Use simple instance start/stop, or
• Tear down/build up altogether
• Instances are disposable

Right sizing
• Selecting the cheapest instance available
while meeting performance requirements
• Looking at CPU, RAM, storage, and network
utilization to identify potential instances that
can be downsized
• Leveraging Amazon CloudWatch metrics and
setting up custom RAM metrics
Rule of thumb: Right size, then reserve.
(But if you’re in a pinch, reserve first).

Use Amazon CloudWatch to
collect and track metrics
CloudWatch
Match resources to
workload

Amazon CloudWatch monitoring
Basic
• 7 metrics for Amazon EC2
including:
• CPU utilization
• Data transfer
• Disk usage and more
• 5-minute frequency
• Metrics for Amazon EBS,
Amazon DynamoDB, Amazon
RDS, etc.
Detailed
• 1-minute frequency
• Aggregation by instance type and
AMI

Instance families: Example
Instance vCPU Mem
(GiB)
Monthly price
(3-yr AURI )
Ideal use case
c4.2xlarge 8 15 $125.44 Best price-compute performance
m4.2xlarge 8 32 $137.25 Balanced
r3.2xlarge 8 61 $179.25 Lowest cost per GiB RAM
r3.xlarge 4 30.5 $89.61 Lowest cost per GiB RAM

t2.medium
4 GiB RAM
Baseline performance: 40%
Bursts beyond this based on CPU
credits
Less than $18 per month
Details: http://amzn.to/1sl2bKa
Web servers, dev, small databases
t2

t2 vs m3.medium or c4.large
Instance vCPU Mem
(GiB)
Monthly price
(3-yr AURI )
Ideal use case
m3.medium 1 3.75 $19.08 Always available, balanced
c4.large 2 3.75 $31.36 Always available, compute
t2.medium 2 4 $16.86 Bursty workloads

Savings potential
If t2 works for you
• 11%: Switch from m3.medium to t2.medium
• 46%: Switch from c3.large to t2.medium
Instance optimization
• 9%: Switch to c4 for compute-intensive apps
(m4.2xl -> c4.2xl)
• 35%: Switch to r3 for memory-intensive apps
(m4.2xl -> r3.xlarge)

Storage
AWS Cloud
Amazon
Glacier
Gateway Appliance/
AWS Storage Gateway
Amazon
S3
Block File
On-premises Data Center
Archive Backup Disaster
Recovery
Amazon
EBS
• Amazon S3 Standard Infrequent Access
‒ Lower cost of storage
• Amazon S3 Reduced Redundancy
‒ 99.99% durability vs. 99.999999999%
‒ Up to 20% savings
‒ Great for files that are easy to reproduce
• Amazon Glacier
‒ Same durability as S3
‒ 3 to 5 hours restore time
‒ Up to 89% savings
‒ Great for archiving, long-term backups and
old data

Instance Purchasing Options
On-Demand
Pay for compute
capacity by the hour
with no long-term
commitments
For spiky workloads,
or to define needs
Reserved
Make a low, one-time
payment and receive
a significant discount
on the hourly charge
For committed
utilization
Spot
Bid for unused
capacity, charged at a
Spot Price which
fluctuates based on
supply and demand
For time-insensitive or
transient workloads

Reserved Instances
Commitment Level
1 Year
3 Year
AWS Services Offering RIs
Amazon EC2
Amazon RDS
Amazon DynamoDB
Amazon Redshift
Amazon ElastiCache
* Dependent on specific AWS service, size/type, and region

Reserved Instances
Step 1: RI Coverage
• Cover always on resources
Step 2: RI Utilization
• Leverage RI flexibility to increase utilization
• Merge and split RIs as needed
Rule of thumb: Target 70-80% always on
coverage and 95% RI utilization rate.

• Be Fault Tolerant
• Workloads should be Stateless
• Loosely Coupled workloads preferred
• If possible, deploy to Multiple AZs
• Instant Flexibility is king
• Take advantage of the 2 minutes warning
• There is always Spot capacity available¢
Save up to 90% compared to On-Demand¢
What could your team do with a 10,000 core data
center that costs $100 per hour, with one click?
Spot Instances

Spot Instances
Pricing
• Up to 90% discount
Elastic
• Minimum Commitment
• Commit to 1 hour
• Tradeoff
• Potential for interruption
Picking the right Spot Bid Price - Tolerance for interruptions, % likelihood of
termination

0
2
4
6
8
10
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
/Spot
Mixed Use Instances
Reserved Instances enable you to reserve capacity for one or three years by
paying a low, one-time fee for the capacity reservation and receiving a
significant discount on the hourly charge for your instances
AWS Services Offering RIs
• Amazon EC2
• Amazon RDS
• Amazon DynamoDB
• Amazon Redshift
• Amazon ElastiCache
On Demand
Spot
RI

OS Licensing
• Utilise Amazon Linux
• Supported by Amazon
• No licence fees
• Regularly updated
• Savings of up to 30% over other options*

Offload Your Architecture
Standard Setup
• 4 x Medium Instances
$485
• AWS Data Transfer 1 TB
$194
• Total = $679

Offload Your Architecture
+
Standard Setup
• 4 x Medium Instances
$485
• AWS Data Transfer 1 TB
$194
• Total = $679
Optimized
• 1 x Medium Instance
$121
• CloudFront Data 1 TB
$168
• CloudFront Requests
$1.89
• Total = $291
57%
6X
Cheaper
Faster

Turn off non-production instances
• Look for dev/test, non-prod instances that are
running always-on and turn off
Autoscale production
• Use Auto Scaling to scale up and down based
on demand and usage (e.g., spikes)
Rule of thumb: Shoot for 20-30% of EC2
instances running on demand to be able to
handle elasticity needs.

Use Auto Scaling or queue-based
approaches to add resources when
needed, and turn them off when not
Amazon SQS Queue
Auto Scaling

Automatic resizing of compute clusters
based on demand
Trigger autoscaling policy
Feature Details
Control Define minimum and maximum instance pool
sizes and when scaling and cool down
occurs.
Integrated to
Amazon
CloudWatch
Use metrics gathered by CloudWatch to drive
scaling.
Instance types Run Auto Scaling for On-Demand and Spot
Instances. Compatible with VPC.
AWS autoscaling create-autoscaling-group
— Auto Scaling-group-name MyGroup
— Launch-configuration-name MyConfig
— Min size 4
— Max size 200
— Availability Zones us-west-2c
Use Auto Scaling
Amazon
CloudWatch

Utilization and Auto-Scaling: Granularity
More small instances vs. less large instances
29 Large @ $0.32/hr
= $9.28
59 Small @ $0.08/hr
= $4.72

Utilization and Auto-Scaling: Lost $ / Lost Customers
Initial Fixed
Capacity
Utilization
Unused Capacity = Wasted $
Downtime, Lost Customers, Lost Revenue (Impossible to measure)
More Wasted $
Time

AWS
Lambda
Elastic Load
Balancing
Amazon ECR Amazon
CloudFront
Amazon EFS Amazon
Glacier
Amazon
S3
Amazon
DynamoDB
Amazon
ElastiCache
Amazon
RDS
Amazon
Redshift
Amazon
VPC
Amazon
Route 53
AWS KMS
AWS WAF Amazon
Elasticsearch Service
Amazon
EMR
Amazon
Kinesis
Amazon Machine
Learning
AWS IoT Amazon
WorkSpaces

Amazon RDS Multi-AZ
High availability
with a mouse click
Set up primary and standby instances
Set up identical volumes
Create synchronous replication
Create and manage DNS entries
Detect instance failure conditions
Detect network failure conditions
Detect storage failure conditions
…
...
...
Decide when to fail over
Re-establish primary secondary connections
Do-it-yourself MySQL replication
Potentially ~100+ manual steps

Reviewing & Reporting

Detailed billing reports
and
cost insights

Detailed Billing Reports
http://amzn.to/1swNwLV

Cost Explorer

Forecasting

Budgeting

AWS Well Architected
• Security
• Reliability
• Performance
• Cost Optimisation
https://d0.awsstatic.com/whitepapers/architecture/AWS_Well-
Architected_Framework.pdf

Creating a culture of cost transparency
Targets and Metrics Cloud Competency
Center
AWS Enterprise
Support

Metrics & targets
% Instances turned off daily
% of Instances Right-Sized
% Always on Resources
Covered by RIs
% RI Utilization
✔
✔
✔
✔
✘
✘
✘
✘
Set up metrics to define success and track progress

Recap
Simple Stimulating Stretching
Permissions
Tagging
Idle Resources
Storage
Purchasing Options
OS Licencing
Offloading Architecture

If you need extra help

AWS Professional Services
• A global team of consultants across EMEA,
APAC and the Americas
• Work with Enterprise and mid-market
customers across many industries
• Specialists in IT Transformation, Security,
DevOps, Big Data and Infrastructure
• Medium to long-term engagements of one
month or more
• Provide guidance, advice, best practice and
accelerates adoption

Resources to get you started
AWS TCO Calculator
https://awstcocalculator.com
AWS Economics Centre
http://aws.amazon.com/economics/
Case Studies & Research
http://aws.amazon.com/economics/

Email me if you have any questions about today’s presentation:
mario@amazon.com
Thank You!

AWS Enterprise Summit Netherlands - Cost Optimisation at Scale

More Related Content

AWS Enterprise Summit Netherlands - Cost Optimisation at Scale