This is a great video Mark Question for you…given the updates to Openshift do you recommend going this type of direction (F5 to pod directly) vs using something like the loadbalancer service in K8s to access the pods via a nodeport proxy?
Any service is fine. Could be NGINX Service for HTTP and POD for TCP. I am working with RedHat to document/demo BIG-IP to HAProxy across multiple clusters. Nodeport or ClusterIP its your preference. However i think you better off using ClusterIP to NGINX and then give the Dev the flexibility todo what they want in OpenShift
One question, why would External BIGIP needs to know Kubes routes in the routing table of BIGIP, assume since not directly connected to any of cluster, it can just follow the default route and upstream can handle routing part. ?
@@MarkDittmer I saw the endpoints from 2 clusters are put into the single one LTM-pool, how can add policy CRD to do weighted LB for the different cluster. For example, I want do weighted traffic for 2 pools (pool1 represent the cluster 1, pool2 represent the cluster2).
using ServiceType LB? We are working on service type multicluster design this week and code available soon. Istio uses a HELM chart and has no CRD. Theirfore all needs to be specified in the service. Message me to get test image and docs.
This is a great video Mark
Question for you…given the updates to Openshift do you recommend going this type of direction (F5 to pod directly) vs using something like the loadbalancer service in K8s to access the pods via a nodeport proxy?
Any service is fine. Could be NGINX Service for HTTP and POD for TCP. I am working with RedHat to document/demo BIG-IP to HAProxy across multiple clusters. Nodeport or ClusterIP its your preference. However i think you better off using ClusterIP to NGINX and then give the Dev the flexibility todo what they want in OpenShift
One question, why would External BIGIP needs to know Kubes routes in the routing table of BIGIP, assume since not directly connected to any of cluster, it can just follow the default route and upstream can handle routing part. ?
CIS can create the routes for BIG-IP. Then BIG-IP simple route to the POD via the next hop "Node IP"
Do we need to configure routing table on BIG-IP manually to direct traffic from BIG-IP to pods via Nodes?
CIS will automate the routing table if you using OpenShift and some other CNIs.
will LB the service with policy in the feature, e.g., LB with multi pools with LTM policy/iRules.
You can assign a policy CRD to the extended ConfigMap which includes any LTM policy/iRules
@@MarkDittmer I saw the endpoints from 2 clusters are put into the single one LTM-pool, how can add policy CRD to do weighted LB for the different cluster. For example, I want do weighted traffic for 2 pools (pool1 represent the cluster 1, pool2 represent the cluster2).
How about OpenShift Ingress in a multi cluster with Istio ingress as a gateway. For those who are heavily using Istio.
Is it possible ?
using ServiceType LB? We are working on service type multicluster design this week and code available soon. Istio uses a HELM chart and has no CRD. Theirfore all needs to be specified in the service. Message me to get test image and docs.