What is required to have HTTP HA load balancing using GCP's load balancers? Should I setup two ore more load balancers pointing to the same backends? I can't find a way to define what zones the load balancers reside in, which makes me think that the load balancers are HA by default, but I haven't seen it explicitly stated in the documentation.

Edit: This is regarding HTTP load balancing.

  • The exact answer is going to depend a bit on what protocol you intend to be using and what capabilities your clients have. If your service is reachable through two different IP addresses and clients are able to automatically fail over between them if one is not responding you can get better reliability than is achievable with only a single IP address. – kasperd Jan 21 '18 at 19:24
  • It's also going to make a difference which kind of GCP load balancing you choose. The HTTP and TCP load balancing are for example quite different. – kasperd Jan 21 '18 at 19:30
  • Edited question to reflect HTTP load balancing. – mark Jan 21 '18 at 20:09

The HTTP Load Balancer is a Global resource, which means they do not reside in a single zone. You don't need to take any specific steps to have it be High Availability, having multiple LB or multiple frontends pointing to the same backends is not necessary.

You will want to have multiple back-ends with good health checks configured(or a single backend with multiple instances). Ideally, you will want to use a Managed Instance Group to allow Autoscalling as well.

The main cause of downtime with HTTP(S) Load Balancers is caused by unhealthy or misconfigured backends.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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