Skip to content
This repository has been archived by the owner on Nov 16, 2023. It is now read-only.

Commit

Permalink
Update README.md
Browse files Browse the repository at this point in the history
  • Loading branch information
@rtibi
rtibi authored Jun 16, 2020
1 parent 8cfedb3 commit 3b6ece4
Showing 1 changed file with 2 additions and 2 deletions.
4 changes: 2 additions & 2 deletions AKSe-on-AzStackHub/README.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -260,11 +260,11 @@ While the infrastructure of Azure Stack Hub is already resilient to failures and

Therefore it is a good practice to deploy your production Kubernetes cluster as well as the workload to two (or more) clusters hosted in different locations or datacenters and use technologies like Azure Traffic Manager (or comparable solutions) to route users based on cluster response time or based on geography.

![Using Traffic Manager to control traffic flows](/img/aks-azure-traffic-manager.png)
![Using Traffic Manager to control traffic flows](img/aks-azure-traffic-manager.png)

Customers who have a single Kubernetes cluster typically connect to the service IP or DNS name of a given application. In a multi-cluster deployment, customers should connect to a Traffic Manager DNS name that points to the services/ingress on each Kubernetes cluster.

![Using Traffic Manager to route to on-premises Cluster](/img/aks-azure-traffic-manager-on-premises.png)
![Using Traffic Manager to route to on-premises Cluster](img/aks-azure-traffic-manager-on-premises.png)

> This pattern is also a [best practice for (managed) AKS clusters in Azure](https://docs.microsoft.com/azure/aks/operator-best-practices-multi-region#plan-for-multiregion-deployment).
Expand Down

0 comments on commit 3b6ece4

Please sign in to comment.