Support for multiple data centre

[wallrj]

This blog suggests creating a statefulset per DC.

• https://www.ibm.com/developerworks/library/ba-multi-data-center-cassandra-cluster-kubernetes-platform/index.html

So perhaps we can have a Navigator NodePool per datacentre.

We should consider that.

/feature

[munnerz]

Related issue for elasticsearch with some ideas on how we can generalise attributes of hosts that the DB runs on, and make the database aware of the topology of the system. I'd rather avoid introducing terms like "zone" and "data centre" in favour of arbitrary labels on nodes and pods. This is how kubernetes represents failure domains without assuming anything about the topology. #44

…

On Sat, 27 Jan 2018 at 12:51, Richard Wall ***@***.***> wrote: This blog suggests creating a statefulset per DC. - https://www.ibm.com/developerworks/library/ba-multi-data-center-cassandra-cluster-kubernetes-platform/index.html So perhaps we can have a Navigator NodePool per datacentre. We should consider that. /feature — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#227>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAMbPx8eDCTWzTTmsOLjkaPdZYq58Ff1ks5tOxvngaJpZM4RvSH4> .

[munnerz]

This is pretty complex, as a Kubernetes cluster is usually homed in a single region, and federation doesn't seem like it'll be a major theme in Kubernetes multi-cluster in future.

We would need some way to federate Navigator itself in this instance. We'd also need to think about how we'd manage HA for pilot-controller across clusters, and how that'd work.

[munnerz]

(specifically running across Kubernetes clusters, that is)

If this issue is about running on different nodes within a single cluster, then we already have this now and this issue can be closed 😄