Kong cascading Hybrid installation

Hi all,

I am looking for help finding a solid, scaleable setup mixing Kong Ingress for Kubernetes with Kong API GW.

Having multiple K8s Clusters with (simplified):

  • 1 customer/tenant per namespace

  • 1 service mesh per namespace

  • 1 Kong IngressController per namespace for ingress traffic (part of mesh)

  • 1 ingress class per tenant

I thought of putting 1…n Kong API GWs (on VMs) in front of all the Kong Ingress instances to route and control traffic, that is coming from the “outside” to K8s applications.
Kong API GW and KIC as dataplane and an additional Kong controlplane to manage all.

Traffic (from outside to K8s) → Kong API GW → Kong IngressController → K8s Pod.

Traffic (inside K8s, between tenantA and tenantB) → Pod tenantA → Ingress tenantB → Pod tenantB.

Is placing a Kong API GW in front of all IngressControllers a good idea?
Are there any alternative setups for this kind of scenario?

Best regards
dmilde


© 2019 Kong Inc.    Terms  •  Privacy  •  FAQ