Running Lean and Mean: Designing Cost-efficient Architectures on AWS (ARC313) | AWS re:Invent 2013

Running Lean and Mean:
Designing Cost-Effective Architectures on AWS
Constantin Gonzalez, Solutions Architect, Amazon
Ralph Gootee, CTO and co-founder, PlanGrid.com
November 14, 2013

© 2013 Amazon.com, Inc. and its affiliates. All rights reserved. May not be copied, modified, or distributed in whole or in part without the express consent of Amazon.com, Inc.
Friday, November 15, 13

What You’ll Get out of this Session
• A lower AWS bill
• A more scaleable, robust, dynamic architecture
• More time to innovate
• Real-world customer examples
• All in 7 easy to implement sections


Cloud Computing Benefits
• Pay-as-you-go, no up-front investments
• Low on-going cost
• Flexible capacity
• Speed, agility & innovation
• Focus on your business
• Go global in Minutes


Amazon EC2


#1: Use Auto Scaling


Poll: Who Uses Auto Scaling Already?


AWS CloudFormation: Load Balancer
"ElasticLoadBalancer" : {
"Type" : "AWS::ElasticLoadBalancing::LoadBalancer",
"Metadata" : {"Comment" : "This is my load balancer!"},
"Properties" : {
"AvailabilityZones" : [ "us-east-1a", "us-east-1b", "us-east-1c",],
"LBCookieStickinessPolicy" : [ {
"PolicyName" : "CookieBasedPolicy",
"CookieExpirationPeriod" : "30"
} ],
"Listeners" : [ {
"LoadBalancerPort" : "80",
"InstancePort" : "80",
"Protocol" : "HTTP",
"PolicyNames" : [ "CookieBasedPolicy" ]
} ],
"HealthCheck" : {
"Target" : "HTTP:80/",
"HealthyThreshold" : "2",
"UnhealthyThreshold" : "5",
"Interval" : "10",
"Timeout" : "5"
} } }


AWS CloudFormation: Launch Configuration
"LaunchConfig": {
"Type" : "AWS::AutoScaling::LaunchConfiguration",
"Metadata" : {
"AWS::CloudFormation::Init" : {
"config" : {
… packages, sources, files, services …
}
}
},
"Properties": {
"ImageId" : "ami-149f7863",
"InstanceType"
: "m1.small",
"SecurityGroups" : [ {"Ref" : "WebServerSecurityGroup"} ],
"KeyName"
: "MySSHKey",
"UserData"
: { "Fn::Base64" : { "Fn::Join" : ["", [
"#!/bin/bash -vn",
… your user data script …
]]}}
}
}


AWS CloudFormation: Auto Scaling Group

}

"WebServerGroup" : {
"Type" : "AWS::AutoScaling::AutoScalingGroup",
"Properties" : {
"AvailabilityZones" : [
"us-east-1a",
"us-east-1b",
"us-east-1c",
],
"LaunchConfigurationName" : { "Ref" : "LaunchConfig" },
"MinSize" : "1",
"MaxSize" : "5",
"DesiredCapacity" : "2",
"LoadBalancerNames" : [ { "Ref" : "ElasticLoadBalancer" } ]
}


Auto Scaling Tips
• Scale up like a rocket
• Scale down like a feather
• Find the right instance type
• Leave headroom for peaks
• Bonus: Better fault-tolerance
• Bonus: Better scalability


#2: Turn off Unused Instances


#2: Turn off Unused Instances
• Developer/test/training instances
• Use simple instance start/stop, or
• Tear down/build up altogether
• Instances are disposable
• Automate, automate, automate:
• AWS CloudFormation
• Weekend/off-hours scripts
• Use tags


Customer Dev/Test Example

35% saved

Monday Friday End of vacation season


#3: Use Reserved Instances


#3: Use Reserved Instances

3y RI
Break even

1y RI
Break even

Reserved Instances Just Got More Flexible
• Can now be moved between AZs
• Can be moved between EC2-Classic
and EC2-VPC platforms
• Size can be modified within the same
instance family


#4: Use Spot Instances


Poll: Who Uses Spot Instances Already?


• Price based on supply/demand
• You choose your maximum price/hour
• Your instance is started if Spot Price is lower
• Your instance is terminated if Spot Price is higher
• But: You did plan for fault-tolerance,
didn’t you?



$3.28
(1367%)
On-demand:
$0.24
$0.028 (11.7%)

$0.026 (10,8%)

• Very dynamic pricing
• Opportunity to save 80%-90% cost
• But there are risks

• Different prices per AZ
• Leverage Auto Scaling!
• One group with Spot instances
• One group with On-Demand
• Get the best of both worlds


Spot Instance Use Cases
• Batch processing
• Amazon Elastic MapReduce
• Web crawlers
• Development/Test
• Video/Image rendering
• HPC
• Monte-Carlo analysis

AWS CloudFormation: Using Spot Instances
"LaunchConfig": {
"Type" : "AWS::AutoScaling::LaunchConfiguration",
"Metadata" : {
"AWS::CloudFormation::Init" : {
"config" : {
… packages, sources, files, services …
}
}
},
"SpotPrice": "0.59",
"Properties": {
"ImageId" : "ami-149f7863",
"InstanceType"
: "m1.small",
"SecurityGroups" : [ {"Ref" : "WebServerSecurityGroup"} ],
"KeyName"
: "MySSHKey",
"UserData"
: { "Fn::Base64" : { "Fn::Join" : ["", [
"#!/bin/bash -vn",
… your user data script …
]]}}
}
}


Customer Case Study: tadaa.net
• Photo sharing mobile app
• Built-in social network
• >3 Million users
• Moved from on-demand to Spot
• 57% reduction of Amazon EC2 bill
through Spot Instances


Ralph Gootee
CTO and co-founder
ralph@plangrid.com


PlanGrid’s
Autoscaling
Based
on
SQS
counts
New
EC2
instances
start
fast
with
AMIs
Conﬁgurable
Performance
One
Months
Dev
Time*


Queue Count
Server Count


September

October

$15,000

$9,000

without
autoscaling


autoscaling!

Autoscaling
saved

40%
on
our
Amazon
EC2
bill

addiMonally…


Reserved
instances
saved

42%
on
our
Amazon
EC2
bill

PlanGrid’s
Queuing
Amazon
SQS
Json
messages

One
message
per
job
Long
running
jobs


Amazon
SQS
$1,800

*
Amazon
EC2

per
year

per
year

$1,200

*Amazon
EC2:
Large
instance,
1y
reserved,
heavy
utl.

You
Manage
Amazon
EC2
DownMme
Instability
Maintenance
Manpower


AWS
Manages
Amazon
SQS
Redundant
Scaleable
314
Million
Requests
$157
Dollars


PlanGrid’s
Storage
MulMple
Amazon
S3
Buckets
Data
Archived
to
Amazon
Glacier
Huge
Chunks
of
Binary
Data


Amazon
Glacier
will
save

89%
on
our
storage
bill

Amazon S3


#5: Leverage Amazon S3 Storage Classes
• Reduced Redundancy Storage Class
•
•
•
•

99.99% durability vs. 99.999999999%
Up to 20% savings
Great for everything that is easy to reproduce
Amazon SNS notification for lost objects available

• Amazon Glacier Storage Class
•
•
•
•

Same 99.999999999% durability as S3
3 to 5 hours restore time
Up to 89% savings
Great for archiving, long-term backups and old data


Amazon DynamoDB


#6 Optimize Amazon DynamoDB capacity units
• Read/write capacity units (CUs) determine most of
Amazon DynamoDB cost.
• By optimizing CUs, you can save a lot of money
• But:
• Need to provision enough capacity to not run into capacity errors
• Need to prepare for peaks
• Need to constantly monitor/adjust


• Use caching to save read capacity units
• Local RAM caches at app server instances
• Check out Amazon ElastiCache

• Think of strategies for optimizing CU use
• Use multiple tables to support varied access patterns
• Understand access patterns for time series data
• Compress large attribute values

• Use Amazon SQS to buffer over-capacity
writes

1.

2.

4.
EC2

3.

Customer Case Study: tadaa.net
• Huge growth after Instagram
• Heavy users of Amazon DynamoDB


Amazon DynamoDB CUs @tadaa
Dynamic
DynamoDB:
20% saved

Caching/Optimization:
80% saved

Cache
Flush

Growth +
new features

Dynamic DynamoDB


Overall Architecture


#7 Offload Your Architecture
• The more you can offload, the less infrastructure you
need to maintain, scale, and pay for.
• Three easy ways to offload:
• Use Amazon CloudFront
• Introduce Caching
• Leverage existing AWS services


Offload Popular Traffic to Amazon S3, CloudFront

CDN'for'
Sta6c'

CDN'for'
Sta6c'&'

Content'

No'CDN'

Dynamic'
Content'


Server'
Load'

Response'Time'

Server'
Load'

Response'Time'

Server'Load'

Response'Time'

Oﬄoad'

Scale''
Down'

Setting up Amazon CloudFront in the Console


Offload Databases Through Caching


Setting up ElastiCache in the Console


Leverage Existing Services
• Amazon RDS, Amazon DynamoDB or ElastiCache for
Redis, Amazon Redshift
• instead of running your own database

• Amazon CloudSearch
• instead of running your own search engine

• Amazon Elastic Transcoder
• Amazon Elastic MapReduce
• Amazon SQS, Amazon SNS,
Amazon Simple WorkflowService, Amazon SES

Document)
Server)

Simple, more reliable, lower cost

)

Search)
Server)

)

Results

Tools


Simple Monthly Calculator
calculator.s3.amazonaws.com/calc5.html


TCO Calculator
aws.amazon.com/tco-calculator/


AWS Trusted Advisor
aws.amazon.com/premiumsupport/trustedadvisor/

Free with Business or Enterprise Support

Let’s Recap
1. Use Auto Scaling
2. Turn off unused instances
3. Use Reserved Instances
4. Use Spot Instances
5. Leverage Amazon S3 storage classes
6. Optimize Amazon DynamoDB capacity units
7. Offload your architecture

Startup
Spotlight
Sessions
with
Dr.
Werner
Vogels
Thurs.
Nov
14,
Marcello
Room
4406

• SPOT 203 - Fireside Chats – Startup Founders, 1:30-2:30pm
• Eliot Horowitz, CTO of MongoDB
• Jeff Lawson, CEO of Twilio
• Valentino Volonghi, Chief Architect of AdRoll

• SPOT 204 - Fireside Chats – Startup Influencers, 3:00-4:00pm
• Albert Wegner, Managing Partner at Union Square Ventures
• David Cohen, Founder and CEO of TechStars

• SPOT 101 - Startup Launches, 4:15-5:15pm
• 5 companies powered by AWS launching at AWS re:Invent 2013


Please give us your feedback on this
presentation

ARC313
As a thank you, we will select prize
winners daily for completed surveys!


Thank You

Running Lean and Mean: Designing Cost-efficient Architectures on AWS (ARC313) | AWS re:Invent 2013

Related slideshows

More Related Content

Running Lean and Mean: Designing Cost-efficient Architectures on AWS (ARC313) | AWS re:Invent 2013