Unified Disaster Recovery as Part of GitOps

OpenShift now supports synchronous data replication and automated failover between geographically close locations for containers and VMs. Red Hat promises that this new capability covers the vast majority of common VM setups, but admins should still verify the proper replication and failover for their specific workloads. 

OpenShift 4.16 also brings a tech preview of the policy-driven definition and management of disaster recovery via declarative code statements. This enables DevOps teams to define the desired state of the entire cluster of VMs and containers to ensure automated recovery and consistency across regions.

Performance and Scalability

OpenShift Virtualization now allows the addition of virtual CPUs to VMs during runtime (without the need for a reboot). In combination with the tech preview of memory overcommit without negative performance impact, this enables great VM density and more efficient hardware utilization. 

VM live migration now adheres to affinity rules to enable DevOps teams to control which hosts VMs can move while maintaining the desired performance and isolation. Certain workloads may perform better when running on the same hosts, while others may need physical separation for performance or security reasons, or they may require the presence of GPUs on their host. 

The new ability to run latency-sensitive applications that require consistent performance opens use cases in the areas of telecommunications networking, financial trading, and industry control systems. This is achieved by OpenShift’s ability to talk directly to the Linux kernel to prioritize CPU cycles based on workload requirements. 

Unified Multi-Cluster VM-Monitoring

Collecting VM metrics and health data from different clusters in different locations and then aggregating them is a key capability required for consistent, and therefore scalable, VM management. This enables the IT team to monitor all VMs across multiple OpenShift clusters from a single dashboard and to centrally resolve issues across all VMs, from one central interface. For example, an admin may be responsible for 50 OpenShift clusters across different regions but will receive a unified dashboard to look at all VMs by health status, resource utilization, or performance, and define alerts depending on these metrics and the type of VMs in question. 

Server, Storage, and Network Integration

OpenShift 4.16 adds several key integrations with hardware platforms by vendors like Dell, HPE, Lenovo, and IBM. These integrations enable VMs running on OpenShift to take advantage of these platforms’ specific capabilities like snapshots, cloning, network policies, quality of service controls, and GPU-based AI training and inference. Therefore, these integrations enable enterprises to leverage past investments in specialized hardware and infrastructure, while modernizing applications on a modern Kubernetes-based platform.

One Fish, Two Fish, Redfish

The new Redfish integration is exciting to see as it demonstrates the significant role this standard can finally play when it comes to enabling scalability of cloud-native apps. OpenShift can now evacuate workloads based on Redfish alerting of problems of the underlying server host. It can also update the BIOS, GPUs, network interface cards, and DPUs of hosts to specific versions, attach “data images” to allow OpenShift to update device drivers and manage the BIOS. This simplifies the deployment and management of OpenShift deployments on bare metal (at the edge).

Simplified Migration

Ansible Migration Factory is a reference implementation of how to leverage Ansible Automation Platform for determining the requirements of virtualized workloads and for the subsequent automated VM migration to OpenShift Virtualization. This provides DevOps teams with the ability to consistently onboard VM workloads, based on predefined criteria and with minimal manual effort.