Scaling New Heights: Azure Red Hat OpenShift Now Supports 250 Nodes

Azure Red Hat OpenShift (ARO) is a fully managed Red Hat OpenShift service on Azure. We are excited to announce two important enhancements to the capabilities of ARO.

1. The ability to configure multiple IP addresses per cluster load balancer is now generally available.
2. ARO clusters can now scale up to 250 worker nodes.

Previously, ARO clusters were limited to 62 worker nodes because they had only one Internet Protocol (IP) address associated with the cluster’s load balancer. By enabling multiple IP addresses for the load balancer, this bottleneck is eliminated, giving organizations greater flexibility to scale their deployments.

These improvements significantly improve the scalability and adaptability of ARO public clusters, enabling organizations to scale their infrastructure more effectively. Our goal is to support larger clusters, providing a powerful solution for enterprises with a wide range of compute needs. In this blog post, we dive deep into the details of deploying large-scale ARO clusters, explore real-world use cases, and provide essential information to help you get started with this powerful new capability.

Deploying a large-scale ARO cluster

For clusters with more than 101 nodes, we recommend using the following control plane nodes (or similar latest generation instance types):

• Standard_D32s_v3
• Standard_D32s_v4
• Standard_D32s_v5

Below is a sample Azure Command Line Interface (CLI) command to deploy a cluster using Standard_D32s_v5 ​​as the control plane nodes.

Deploying Infrastructure Nodes

For clusters with more than 101 nodes, the infrastructure nodes should isolate cluster workloads (e.g. Prometheus) to minimize contention with other workloads. For redundancy and scalability requirements, we recommend deploying three (3) infrastructure nodes per cluster.

Recommended instance types for infrastructure nodes:

• Standard_E16as_v5
• Standard_E16s_v5

For detailed instructions on configuring infrastructure nodes, see: Deploy infrastructure nodes to an Azure Red Hat OpenShift (ARO) cluster.

For detailed instructions on deploying a large-scale Azure Red Hat OpenShift cluster, see: Deploying Azure Red Hat OpenShift Clusters at Scale – Azure Red Hat OpenShift | Microsoft Learn

Add an IP address to a cluster

You can add up to 20 IP addresses to your load balancer. You will need one (1) IP address per 65 nodes, so a cluster with 250 nodes will need a minimum of four (4) IP addresses.

To add an IP address to a load balancer using the Azure CLI, run the following command.

Alternatively, you can add IP addresses via REST (Representational State Transfer) API (Application Programming Interfaces) calls.

caution: Before deleting a large cluster, ensure that the cluster has less than 120 nodes.

A CLI preview is provided to enable this functionality until an official CLI release is available. Instructions on how to download and install the wheel extension file for this preview CLI are provided here. Documentation

The power of 250 nodes

Traditionally, ARO public clusters were created with a public load balancer that featured a single public IP address for outbound connectivity. While this configuration worked well for many scenarios, it limited the maximum number of nodes to 62. Now, you can assign multiple additional public IP addresses to your load balancer, allowing you to scale your cluster to the maximum supported number of nodes, opening up new possibilities for your applications.

Key Features

1. Expand up to 20 IP addresses per cluster load balancer
2. Automatically adjusted outbound rules and frontend IP configuration
3. Increased the maximum number of nodes per cluster to 250.
4. Improved overall cluster scalability and performance.

Use Case: High-traffic eCommerce platform

Consider MegaShop, an e-commerce company that is experiencing rapid growth. They have been running their platform on an ARO cluster, but they are approaching the 62 node limit. They need to scale up significantly to handle the expected surge in traffic as the holiday season approaches.

MegaShop can implement multiple IP addresses on the ARO cluster load balancer to:

1. Expanding beyond the previous 62 node limit
2. Ensures smooth operation even during peak traffic hours.
3. Maintain high availability and performance for our customers.

MegaShop’s DevOps team can easily update existing clusters to use, for example, 10 IP addresses.

This simple change allows MegaShop to confidently scale its infrastructure to meet holiday demand without worrying about outbound connection bottlenecks.

conclusion

With multi-IP address configuration for ARO cluster load balancers now generally available, organizations have greater flexibility than ever to build and scale powerful, enterprise-grade Kubernetes environments on Azure.

Whether you run a high-traffic e-commerce site, a data-intensive analytics platform, or other scalable applications, these new capabilities will ensure your ARO infrastructure scales to meet your business needs.

Get started

New customers can get started by:

1. Set up your Azure subscription
2. Installing Azure CLI
3. Create a new ARO cluster with any number of IP addresses

For more information, best practices, and troubleshooting guides, visit the official Azure Red Hat OpenShift Documentation and Red Hat OpenShift Documentation.

Embrace the power of scalability and take your ARO deployment to the next level by using multiple IP addresses for your cluster load balancer!

Additional Materials