In this blog post we’ll highlight some of the improvements the team has delivered and discuss some of the advanced use cases that opens up along with highlights from solutions where we’re seeing an increase in demand of application focus for containerized workloads.
As a complement to this blog post. I joined host Calvin Zito in this Around The IT Block podcast.
Read on to learn more about the new features and enhancements in this release.
Flexibility for diversely connected clusters
Kubernetes allows administrators to deploy nodes with different capabilities in the same cluster. There are many common patterns that we’ve observed in the installed base.
- Separating infrastructure software from important business applications
- Restricted connectivity to corporate networks from user generated network traffic
- Compute nodes with specialized hardware, i.e. GPUs and/or FC adapters
From a CSI driver for a SAN storage array perspective, some of these scenarios are very relevant. A CSI driver transact on secure networks, sometimes require specialized hardware (FC) to provide storage connectivity, and in all of the above, exchange storage system credentials for authentication and authorization.
In this new release, Kubernetes administrators may choose which nodes in the clusters the HPE CSI Driver components gets deployed on. Components are divided into the three categories, the controller driver (the control plane), the node driver (the data plane) and the CSPs (API gateways to the backend storage system). Four new fields per controller are available in either the Helm chart or Operator.
- Tolerations: Deploy on nodes which are tainted, i.e. control-plane nodes.
- Affinity: Prefer, require, and/or avoid certain nodes and/or Pods
- Node selector: Pick nodes with a certain label (built-in or user defined)
It’s now also possible to add your own custom labels to the components for easier lookup or create affinity between the components themselves. Such an example would be to always co-locate the CSPs with the controller driver for consistent API response times.
The parameters are further documented in the Helm chart hosted on ArtifactHub.io.
Major improvements
The rise in popularity of the NFS Server Provisioner has led to investigations to cut down the automatic recovery times of a NFS-based PVC. Users can expect roughly a third of the recovery time compared to previous releases.
The end-to-end volume encryption feature available for all HPE CSI Driver supported storage backends has been reworked to support seamless volume expansion. For pre-existing volumes, data needs to be migrated to new PersistentVolumeClaims and such a workflow has been documented on HPE Storage Container Orchestrator Documentation (SCOD) portal: Migrate Encrypted Volumes.
Backend storage volumes exposed to Kubernetes nodes over either iSCSI or FC are subject to sophisticated host configuration. The multipathing component has been improved significantly for this release to ensure nodes are conformed to the respective storage backend best practices to reliably and as performant as possible provide storage to Kubernetes.
Ecosystem updates
As HPE continue to partner in the cloud native computing sphere the HPE CSI Driver continue its momentum to now support Kubernetes 1.24 with notable stable storage APIs for CSI volume expansion and CSI storage capacity tracking.
Red Hat OpenShift 4.10 Extended Update Support (EUS) release is fully certified by HPE and Red Hat to continue expansion of our joint capabilities with the recently announced HPE GreenLake integration.
Updates will follow for as integration tests complete for partners such as SUSE Rancher, Mirantis Kubernetes Engine and VMware Tanzu.
Remember to bookmark SCOD for the latest updates to the support matrix.
Solutions focus
Demand for higher level content has increased recently with more partner and application centric approaches to implement Kubernetes and cloud native operational patterns. As HPE build momentum and excitement for these new deliverables we want to take the opportunity to highlight recent and relevant assets.
- Technical Paper: Containerizing MongoDB with HPE Persistent Storage
- Technical Paper: Data Protection for Kubernetes using HPE Nimble dHCI, Kasten K10 by Veeam, and Scality RING
- Light board: Lift and Transform Apps and Data with HPE Storage
- Light board: Manage multitenancy at scale with HPE Alletra 6000 and Nimble Storage
Be on the lookout right here on Around The Storage Block for future content featuring technical deep-dives, end to end solutions and application focus!
Get your Kubernetes clusters ready!
In order to get customers and partners prepared to get the HPE CSI Driver for Kubernetes installed, the team has prepared a set of useful links.
- Is your environment supported? Check Compatibility and Support on SCOD.
- Are you using Helm to manage the HPE CSI Driver? Head over to ArtifactHub.io.
- Red Hat OpenShift users should check the certified combinations on SCOD.
- Do you know about the HPE CSI Info Metrics Provider for Prometheus? Check out the previous release blog of the HPE CSI Driver right here on Around The Storage Block.
If you have any questions, comments, or concerns, feel free to hop on the HPE Developer Community Slack and interact with the teams in their respective areas, such as #kubernetes #alletra #nimblestorage and #primera. Sign up here and login at hpedev.slack.com.
MichaelMattsson
Data & Storage Nerd, Containers, DevOps, IT Automation