High Availability in AWS-Multiple AZ and Regions

We keep hearing about the principle of “Design for Failure” from AWS Technical evangelists in various cloud conferences. On the other side, we have seen from the impact of April 2011 AWS outage that many web companies did not design their systems adhering to this principle. The reasons could be ranging from internal IT teams competency/awareness to build High Availability (HA) architectures in AWS to tradeoffs in the cost of operating a complex global HA setup.

In this article, we have shared some of our prior experience in architecting High availability systems using AWS as sample blueprints. We feel this small gesture will help the strong AWS user community to build better solutions and cope up with future outages.

I have taken a sample architecture involving a LAMP stack with background programs in the below presentation and see how we can build high availability aspects into this architecture in steps.

The sample blueprints are categorized as:
Blue print 1: How to achieve High Availability across AWS Availability Zones (AZ’s)?

HA across AWS AZ’s is the first step that web companies should design in their infrastructure to cope up with AWS outages. HA across AZ blueprint leverages the isolated Multiple AZ (Data center) cloud infrastructure of AWS. Systems built with these principles can withstand failures inside a single AWS AZ.

Blue print 2: How to achieve High Availability across AWS Regions?
HA across AWS regions is a sophisticated Infra architecture that is suitable for companies which demand high level of Scalability, load balancing and Availability across the globe. Systems built with these principles can withstand Data Center failures and even survive Continental Disasters. They are cost effective only for large deployments. On the negative side, they are technically complex and costlier to operate.

Cloud Developer Conference May 2011 SiliconIndia : Design for Failure - High Availability using AWS

View more presentations from Harish Ganesan

References:

The embedded presentation was presented in SiliconIndia Cloud Developer Conference (May 2011). PDF version can be downloaded from the below link:

http://www.siliconindia.com/events/siliconindia_events/Cloud_conference_ppt/8KMiles_High_Availability_in_AWS.pdf

http://www.siliconindia.com/events-speakers/cloud-computing-summit-Bangalore-CloudSummitBANGALORE.html