Node balancer

Hello,

You are correct - that's one of the main advantages of NodeBalancers. NodeBalancers themselves are highly available - if a container fails or the NB fails to respond or some other bad condition is detected, then the NB falls back to another container. The failover threshold is a few seconds, in most cases. This is all baked into the NodeBalancer platform and is one fewer thing you need to worry about :)

-Chris

@Fearghal:

I am considering using Linode for a client's production environment and have some questions regarding the node balancer.

How is the node balancer setup in terms of high availability? There wouldn't be much point deploying Linodes to different regions to ensure high availability if the node balancer fails and stops access to all Linodes behind it. I can't seem to find any detailed information on how exactly it is setup.

Thanks

Note that a NodeBalancer can only balance Linode backends located within the same datacenter as eachother/itself.

@caker:

Hello,

You are correct - that's one of the main advantages of NodeBalancers. NodeBalancers themselves are highly available - if a container fails or the NB fails to respond or some other bad condition is detected, then the NB falls back to another container. The failover threshold is a few seconds, in most cases. This is all baked into the NodeBalancer platform and is one fewer thing you need to worry about :)

-Chris

Ok thanks, I wasn't sure the NB itself had a failover to another NB if something went wrong.

@rmcintosh:

@Fearghal:

I am considering using Linode for a client's production environment and have some questions regarding the node balancer.

How is the node balancer setup in terms of high availability? There wouldn't be much point deploying Linodes to different regions to ensure high availability if the node balancer fails and stops access to all Linodes behind it. I can't seem to find any detailed information on how exactly it is setup.

Thanks

Note that a NodeBalancer can only balance Linode backends located within the same datacenter as eachother/itself.

Hmm did not realize that. I was planning on deploying to both London and Frankfurt. Can you suggest any other reliable way to distribute traffic between two node balancers at each of those locations?

Thanks!

@Fearghal:

Hmm did not realize that. I was planning on deploying to both London and Frankfurt. Can you suggest any other reliable way to distribute traffic between two node balancers at each of those locations?

You can still achieve this sort of functionality, but it has to be done on your own. Round Robin DNS is a simple way to achieve it, but has some set backs. I recommend reading into that and other options to see what would work best for you.