AWS: Route53, DNS Failover, Active-Passive
A couple of weeks ago, AWS announced such a new and very useful functionality in Route53 as DNS Failover. This is good news not only for those using the Amazon cloud, but also for everyone else who wants resiliency at the DNS level. We will understand how this works and make a test example in the Active-Passive configuration, with backup of your site on S3.
The endpoint whose availability you want to check is configured with a health check. Checks come from different locations around the world, and your host is considered not operational if it does not respond within 30 seconds: it is automatically removed from the issuance of IP clients. Checks can be done via http and tcp, https is not supported. Also, work is still underway on integration with ELB. TTL advised to set in the region of 60 seconds.
The easiest way you can use this is to keep Primary recording for your primary server, and to send a Secondary static site to S3 if the first one drops.
')
So, you have a regular A record for your domain.
Add a health check for it.
See that the check has been added.
Go back to Hosted Zones, choose our record and Routing Policy: Failover, Failover Record Type: Primary, and Health Check to Associate:205.251.195.209 : 80 /
Make a Secondary record, select a pre-configured S3 bucket, Failover Record Type: Secondary , and that's it.
Also, checks can be associated with LBR or Weighted policies, do Active-Passive-Passive, Active-Active, etc. In production, we are already successfully using Active-Active for internal balancing.
Some details.
The endpoint whose availability you want to check is configured with a health check. Checks come from different locations around the world, and your host is considered not operational if it does not respond within 30 seconds: it is automatically removed from the issuance of IP clients. Checks can be done via http and tcp, https is not supported. Also, work is still underway on integration with ELB. TTL advised to set in the region of 60 seconds.
A working example with backup on S3.
The easiest way you can use this is to keep Primary recording for your primary server, and to send a Secondary static site to S3 if the first one drops.
')
So, you have a regular A record for your domain.
Add a health check for it.
See that the check has been added.
Go back to Hosted Zones, choose our record and Routing Policy: Failover, Failover Record Type: Primary, and Health Check to Associate:
Make a Secondary record, select a pre-configured S3 bucket, Failover Record Type: Secondary , and that's it.
Also, checks can be associated with LBR or Weighted policies, do Active-Passive-Passive, Active-Active, etc. In production, we are already successfully using Active-Active for internal balancing.
Source: https://habr.com/ru/post/170471/
All Articles