1. Migrating Existing Applications to AWS
Matt Tavis | Principal Solutions Architect

2. Planning on moving apps to the cloud?
You have a lot to decide…

3. A Path to the Cloud
• Select apps
Deploy • Embrace AWS
• Test platform • Migrate data services
• Plan migration • Migrate components • Re-factor
architecture
• Cutover
Plan Optimize

4. …knowing is half the battle.
Architecture Center
Security Center
Whitepapers
Resources
Case Studies
Solution Providers

5. Migrating Data onto AWS
GBs TBs
One-time upload w/ UDP Transfer Software
Data Velocity Required
constant delta updates (e.g., Aspera, Tsunami, …)
Hours
Days
Transfer to S3
AWS Import / Export
Over Internet
Data Size*
* relative to internet bandwidth and latency

6. Cutting Over Your Master Data Store
Bulk Export Transfer
Transfer Snapshot to AWS
Delta Freeze Cutover
Export Transfer to
Data and
Updates Deltas AWS
Source Unfreeze

7. Finding your first stop…
Express
Local Optimize for AWS
 Re-design for AWS
 Fully embrace
cloud architecture
Forklift Embrace AWS
 Virtualize app  Optimize resources
 Retain operational  Scale on-demand
approach

8. The Migration Continuum
Forklift Embrace Optimize
Effort Scalability Operational Burden
Forklift Embrace AWS Optimize for AWS
• May be only option for • Minor modifications to • Re-design with AWS
some apps improve cloud usage in mind (high effort)
• Run AWS like a virtual • Automating servers • Embrace scalable
co-lo (low effort) can lower operational services (reduce
• Does not optimize for burden admin)
on-demand (over- • Leveraging more • Closer to fully utilized
provisioned) scalable storage resources at all times

9. Mapping the Application to AWS - Forklift
Matching resources
 Examine existing footprint with an eye to resource
requirements for that app – don’t obsess on HW specs!
Building your AMIs
 Create AMIs to match server roles
Converting appliances
 Create/discover virtual variants for physical appliances
 Consult our Solution Providers site
Deploying supporting components
 NASes, SANs, DNS, Domain Controllers, …

10. Mapping the Application to AWS - Forklift
Secure the application
 VPC and/or Public EC2 with Security Groups
 Data protection in-transit and at-rest
Integrate monitoring and management tools
 Use existing tools and/or new cloud management
platforms

11. Mapping the Application to AWS – Embrace
Everything on the last slide(s) plus…
Rethink storage
 Leverage new storage models offered by AWS
Embrace elasticity
 Bootstrapping your AMI
 Dynamic build out, configuration and discovery
Scale out and in on-demand
 Leverage auto-scaling in both directions

12. Mapping the Application to AWS – Optimize
Everything on the last two slides plus…
Re-re-think storage
 Further optimize storage through the use of non-
traditional storage models (e.g., SimpleDB, NoSQL, …)
Parallelize processing
 Use more smaller resources to improve granularity of
resources
Use Spot where possible
 Build for total transient nature to reduce costs
Embrace scalable on-demand services
 Replace software components with scalable AWS
services (e.g., SQS, SNS, SES, …)

13. ELB
Forklift steps:
A Phased Migration to AWS - Forklift
Match resources and build AMIs
• Thinks about application
needs not server specs
• Build out custom AMI for
application roles
Convert appliances:
• Map appliances to AWS
AMI-1 @ AMI-1 @ services or virtual appliance
C1.Medium C1.Medium
AMIs
Deploy supporting components:
• SAN replacements
• DNS
AMI-2 @ AMI-2 @ • Domain controllers
M2.XLarge M2.XLarge
Secure the application
components:
AMI-6 @
M2.XLarge
AMI-3 @ AMI-5 @
M2.2XLarge
AMI-4 @ • Use layered security groups to
C1.Medium M1.Large
replicate firewalls

14. Steps to Embrace AWS:
A Phased Migration to AWS - Embrace
ELB
Rethink storage:
• Leverage S3 for scalable
Auto-scaling Group
WebWeb WebWeb storage
Server
Server Server
Server • Edge cache with CloudFront
• Consider RDS for HA RDBMS
Web Tier
Parallelize processing:
• Bootstrap AMIs for auto-
Auto-scaling Group
App Server
App Server App Server
App Server discovery
• Pass in bootstrapping
parameters
App Tier
• Leverage configuration
management tools for
Master Config automated build out
Database Management
Server
Scale out and in on-demand:
• Use CloudWatch and Auto-
scaling to auto-provision the
Network Network DNS Domain fleet
Filesystem Filesystem Controller

15. Steps to Optimize for AWS:
A Phased Migration to AWS - Optimize
Re-Rethink storage:
• Break up datasets across
Auto-scaling Group
WebWeb WebWeb storage solutions based on
Server
Server Server
Server best fit and scalability
Web Tier
Parallelize processing:
• Spread load across multiple
Auto-scaling Group
App
Server
App Server App Server resources
SQS • Decouple components for
parallel processing
App Tier
App App
Server Server
EMR
Use Spot where possible to reduce
costs
Embrace scalable on-demand
services
Config Network DNS Domain • Scale out systems with
Route
Management Filesystem Controller minimal effort
Server 53
• Route53
• SES, SQS, SNS
• …

16. Mitigating Factors
Software Architecture Limitations
 Lack of control over design of software components
Licensing
 Software licensing costs need to be factored into the
scale out and in plans as not all software vendors have
cloud-ready pricing plans
Security, Compliance and Regulations
 Where, how and the manner in which data can be
processed based on industry regulations can influence
AWS architectural options.
Skillset / Training
 Finding the right personnel to handle AWS migrations
can be pivotal in the success of any migrations.

17. Additional resources:
 Architecture Center:
http://aws.amazon.com/architecture
 Security Center: http://aws.amazon.com/security
 Whitepapers: http://aws.amazon.com/whitepapers
 Resources: http://aws.amazon.com/resources
 Case Studies: http://aws.amazon.com/solutions/case-
studies
 Solution Providers:
http://aws.amazon.com/solutions/global-solution-
providers/
Thank you!