Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I am running an Elastic Load Balancer backup by 2 instances in different Availability Zones in US East. I am using Multi-AZ RDS as well. Ideally this should ensure that if one AZ goes down, it should not effect the app because everything is spread across multiple AZs.

But the recent AWS outage took the app down for a long time. I am not sure how this can happen. It would be great if someone can point out what went wrong.

Major question here I have is how can I avoid this in future? I can setup app servers across different regions or even providers and use DNS for load balancing but what do I do with MySQL? Read Replicas will introduce some lag which I would want to avoid.

share|improve this question
    
You'll need to provide more information about exactly how the application went down in order to get an answer. –  mgorven Oct 26 '12 at 6:15
    
app went unreachable. It started returning a 502 which I think was due to the nonavailability of RDS. Later it just stopped working and we got timeouts on the browser. –  Aditya Patawari Oct 26 '12 at 12:32

2 Answers 2

You would need to have AWS instances in different regions to counter the last two big AWS outages.

As elastic load balance is per AZ, you would also need to add route53 to the mix to give you failover to the other region

share|improve this answer
    
That does not answers the question. I have already written that I can use multiple regions or providers. I want to know how mysql replication will work in this case. –  Aditya Patawari Nov 2 '12 at 6:29
    
You said you were in US East, that is one region, you would need to be in US East and for example US West to be in two regions. I was pointing out what went wrong... –  shouldbeq931 Nov 3 '12 at 9:29

You'll have to setup two working copies of your application in several AZ, so you'll have to use replication (master-master), so when one AZ goes down you can switch traffic to another. This is hard way by both financial and human resources.

share|improve this answer
    
Finance are HR are not the issue. Issue would be latency. Would MySQL be able to handle latency of 150ms? –  Aditya Patawari Nov 3 '12 at 15:43

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.