Amazon EC2 High Availability Database Architecture

Question

I'm currently spec'ing out a solution running on EC2. Setting up web servers, utilizing S3, Cloud Front, Cloud Watch, etc have been straightforward enough. Using Elastic Load Balancers for HA cross Availability Zone for our web servers has also been straightforward.

I would really like to have cross Region Availability in addition to cross Availability Zone. This would help alleviate the potential for an outage due to region outage.

I haven't been able to find much information about folks running two way replication cross region for their dbs. Is this possible? What is the performance like?

Answer 1

Before investing much time and effort, you may want to take into consideration that Amazon is promising a High-Availability offering on top of the Amazon Relational Database Service (RDS).

New Features for Amazon RDS Coming Soon

High Availability Offering — For developers and business who want additional resilience beyond the automated backups provided by Amazon RDS at no additional charge. With the high availability offer, developers and business can easily and cost-effectively provision synchronously replicated DB Instances in multiple availability zones, to protect against failure within a single location.

Answer 2

The answer provided is not really correct. You asked about cross-region availability but the answer provided addresses availability zones.

From Amazon:

Q: What is an Availability Zone?

A: Availability Zones are distinct locations within a Region ...

Q: Will my standby be in the same Region as my primary?

A: Yes. Your standby is automatically provisioned in a different Availability Zone of the same Region as your DB Instance primary.

Availability zone reside within a single region (US East 1a, US East 1b) etc, not across regions (US East (Virginia), US West(N. California)). Replication between zones may provide a level of availability, but, does not necessarily protect against a regional event.

A contact at Amazon did not know if there were any plans to span RDS across regions.

Answer 3

To speak about Cross-Region availability you need to provide some key metrics: What's the volume of changes you have per time period (day/hour) and how frequently do you want to sync your databases.

Secondly you need to think about how your overall architecture will use those DBs: as warm standby or active member ?

Elastic Load Balancers will balance only within a region, if you want to use cross-region services you need other ways of dispatching the traffic.