aws meetup - sydney - february

10
© 2013 Adobe Systems Incorporated. All Rights Reserved. Adobe Confidential. Amazon Web Services & Adobe Experience Manager Kam Radhakrishnan | Senior Consultant | Adobe

Upload: markghiasy

Post on 08-Aug-2015

48 views

Category:

Business


2 download

TRANSCRIPT

© 2013 Adobe Systems Incorporated. All Rights Reserved. Adobe Confidential.

Amazon Web Services & Adobe Experience ManagerKam Radhakrishnan | Senior Consultant | Adobe

© 2013 Adobe Systems Incorporated. All Rights Reserved. Adobe Confidential.

jjjSling

AEM UI

Oak & Plugins

REST

Microkernel (Java)

OSGi

AEM Internal Architecture

3

TAR

Presentation Tier

Middle Tier

Data Tier

Traditional Web Application

© 2013 Adobe Systems Incorporated. All Rights Reserved. Adobe Confidential.

How does it work?

5

A

P P P

D D DCache

InvalidationCache

Invalidation

Author Server

Dispatcher Servers

Publish Servers

Elastic Load Balancer

Internet Users

Marketing Users

© 2013 Adobe Systems Incorporated. All Rights Reserved. Adobe Confidential.

Development Setup

6

Advantages Disadvantages

• Easy to setup • Not fault tolerant.

• Built and re-built in minutes • Cannot scale on Author and runtime.

• Best for learning & practice and ideal developer setup. • Not suitable for production.

AZ#1

A P D

Sydney

© 2013 Adobe Systems Incorporated. All Rights Reserved. Adobe Confidential.

Highly Available Architecture

7

Advantages Disadvantages

• Highly scalable runtime with Auto Scaling Group based on CloudWatch stats.

• Author cannot scale.

• TarMK offers optimal performance. • Manual failover on Author

• Author Servers are fault tolerant • At any given point in time we have few servers that are idle.

AZ#1

AZ#2

A P

P

D

DAc

PG

Auto Scaling Group

Auto Scaling Group

Sydney

PG

© 2013 Adobe Systems Incorporated. All Rights Reserved. Adobe Confidential.

Digital Assets(DAM) Heavy Architecture

8

Ac

PG

PG

A

AZ#2

AZ#1

Sydney

Advantages Disadvantages

• Best for customer who have a large datastore that runs in terabytes.

• Cannot scale on Author.

• Offers failover and DR in a different region. • At any given point in time we have few servers that are idle.

• Supports binary-less replication for storage efficiencies, all author and Publish point to one copy in S3.

P D

P DAuto Scaling Group

Auto Scaling Group

© 2013 Adobe Systems Incorporated. All Rights Reserved. Adobe Confidential.

Distributed Author Architecture

9

A

PGA

AZ#1

AZ#1

Sydney

M

MP D

P DAuto Scaling

Group

Auto Scaling Group

Advantages Disadvantages

• Best when there are large numbers of authors that need to be supported across geographies.

• Required skillset in MongoDB to optimize and administer the servers.

• Architecture offers resilience and scalability and automatic failover with Author clustering and mongo dB replica sets.

• Performance is not as great as Tar Micro Kernel.

• Runtime still runs on Tar Micro Kernel which gives good performance for customers accessing the website.

• More servers to monitor and manage.

MA

PG

Singapore

© 2013 Adobe Systems Incorporated. All Rights Reserved. Adobe Confidential.

D

D

User Generated Content Architecture

10

Ac

P

P

A

AZ#2

AZ#1

Sydney

P D

P D

Auto Scaling Group

Auto Scaling Group

M

M

Advantages Disadvantages

• Ability to handle UGC content on Runtime. • MongoDB skill set required for ongoing maintenance.

• Horizontal scalability on Runtime • Performance is not as good as Tar Micro Kernel.

• Moderation on Publish servers. • More servers to monitor and manage.

MA

© 2013 Adobe Systems Incorporated. All Rights Reserved. Adobe Confidential. 11

Q & A

© 2013 Adobe Systems Incorporated. All Rights Reserved. Adobe Confidential. 12