Luiz - Fotolia
Diamanti Spektra's latest update extends its coverage to Google Cloud, enabling customers to create and manage Kubernetes clusters on all three major public clouds.
Diamanti Spektra 3.2 adds Google Cloud Platform (GCP) support for Ultima, Diamanti's hybrid cloud storage data plane. Ultima is a storage layer with integrated data services such as backup, replication, migration and failover, but it previously worked on only AWS and Azure. Ultima supports a wide range of Kubernetes distributions, including Red Hat OpenShift, VMware Tanzu, Amazon EKS, Google Kubernetes Engine and Azure Kubernetes Service.
The new support allows customers to create GCP-based Kubernetes clusters through Diamanti Spektra's management layer, deploy applications to those clusters and migrate applications between them and any clusters on another public cloud or on-premises storage. Spektra provides visibility into customers' Kubernetes clusters across all environments, allowing them to make application deployment decisions based on geography, cost or performance.
There is rising demand for GCP support among Diamanti's customers, said Brian Waldon, vice president of product. One of the goals of Diamanti Spektra is to be a cloud-vendor-neutral and Kubernetes-distribution-neutral management platform, so on a technical level, it made sense to "complete the story," he said.
"Our future customers are already operating on GCP, and we want to be able to meet them where they are," Waldon added.
The Spektra 3.2 update also added CRI-O support, allowing customers to switch to that as the default runtime instead of Docker. Kubernetes has deprecated Docker support as a container runtime after version 1.20 in favor of runtimes that use its Container Runtime Interface (CRI). Some of Diamanti's customers are already switching off Docker, Waldon said.
Brian WaldonVice president of product, Diamanti
CRI-O was in top demand among Diamanti's customers and the open source community in general, Waldon said. This support addresses that demand, and it allows customers who are making that transition to test their applications against CRI-O before cutting off Docker.
"Docker isn't going to be the tool set that rules all Kubernetes deployments moving forward. This was the open source community's decision, and we're just following the community's lead here," Waldon said.
A different audience
Mainstream storage products such as Catalogic CloudCasa and Hewlett Packard Enterprise SimpliVity use the Container Storage Interface (CSI) driver to provide storage on the back end for Kubernetes applications. This is different from the hyperconverged infrastructure (HCI)-like approach of cloud-native vendors such as Diamanti and Portworx (which is owned by Pure Storage), said Dave Raffo, senior analyst at Evaluator Group. Similar to how HCI works for virtual machines, compute and storage are attached to Kubernetes nodes and managed through Kubernetes such that when applications are moved across nodes, they will always have storage.
HCI allowed virtual server admins to spin up VMs without having to first provision storage through a storage admin. Cloud-native vendors are doing the same thing, but for containers and the people who work most directly with them -- developers. They also provide the data services that would typically be found with storage arrays, such as encryption, deduplication, snapshots and replication.
"The people using Kubernetes are developers and cloud architects. They don't want to deal with storage arrays, and they don't want to deal with the people who manage storage arrays," Raffo said.
Diamanti is targeting this audience, and the latest Spektra release is supporting what they're using. It's still a bit early to identify any Kubernetes deployment trends and best practices, so Diamanti's best bet is to follow the customers' lead, Raffo said. All the Kubernetes management vendors such as Robin Systems, Longhorn and Rancher will eventually want to support CRI-O and the three major public clouds, Raffo added.