Hi All,
I’m trying to create High Availability Istio cluster setup. I have tired Multicluster example it is working fine, but this approach is single point contact,which mean request will get master cluster only. in case master cluster is not accessible, how to request will handle what is failover?
Request latency will take more time? based on Multicluster Service Mesh demo? due to different region ? ex: asia-zone to america-zone ?
How to handle rate limit policies, if same application got deploy in different cluster?
And i’m looking for Federation type support.
Regards,
Sathish.
1 Like
There are a couple different ways that multicluster can be installed. It looks like you installed the single control plane variant. With that variant if the master cluster is totally lost then the mesh state for the remaining clusters won’t be updated. Your choices are to deploy that master cluster in a manner that ensures the entire cluster is highly available. Alternative you could deploy the model described here: https://istio.io/docs/examples/multicluster/gateways/
With that deployment losing a single cluster is less impactful
Hi John,
Thanks for your quick reply.
Even i have tried gateway connectivity approach,even though entry point will be cluster A ingress gateway right?
How to handle load balance for Cluster A & Cluster B ingress gateway?
How can i maintain tractability at one place? because each cluster has own control plane?
Please suggest me right approach for above use cases.
Regards,
Sathish
I also prefer Muliticluster service mesh with Cluster-Aware Service Routing.But how to handle when primary cluster is down? Is there any solution for the scenario?
Primary may run in multiple zones and clusters, using dns to handle the different IPs.
Control plane can be down for short periods, fails open and workloada keep working.