Container technology’s virtualized methods for running distributed software applications are increasingly outshining the work of virtual machines. Many experts agree that containers are more agile and scalable vehicles for workloads compared to VMs. But a lot of existing products — particularly hyperconverged infrastructure appliances — were built with VM-centric principals. Can systems built for VMs transform and adapt to run containers just as effectively as VMs?

Datrium Inc. intends to show that it can with its new Datrium DVX Software 3.0 release. The company’s open converged infrastructure services will now support Red Hat Inc.’s Red Hat Virtualization and Red Hat Enterprise Linux and stateful containers. Plus, the company has tapped Docker Inc. for access to its persistent volumes plug-in.

The new release is a response to customers’ growing desire to run containers alongside VMs, according to Craig Nunes (pictured), vice president of marketing at Datrium. “Our founders are a couple of Diane Greene’s early principal engineers at VMware,” Nunes told Peter Burris (@plburris), host of theCUBE, SiliconANGLE Media’s mobile livestreaming studio. The converged infrastructure the company developed picked up customers of VMware’s vSphere hypervisor and management suite, Nunes added.

Datrium’s converged technology is not to be confused with hyperconverged systems. Differences in Datrium’s design better enable it to branch out from VMs to containers and to scale, according to Nunes.

HCI too much of a good thing for scale

Several years ago, when Hyperconverged Infrastructure was born, “The era of virtualization was hot and heavy,” Nunes said. Vendors could impress customers by managing at the VM level. “As we move into containers, VMs are just not granular enough,” he added.

Part of what makes HCI great for single use cases and edge computing also inhibits its scale-out potential. HCI brings storage into the x86 server, which is a terrific idea in principal, but it has drawbacks. “Because storage is now part of your server, everything is stateful, everything’s a storage node,” Nunes said. This makes it tougher to scale and to service.

Datrium took a step back from total convergence, which resulted in a system with greater flexibility for users, Nunes stated. Datrium separated the persistent storage layer, or what it calls “durable capacity,” from Input/Output processing on the server. Users are free to choose the server of their choice or Datrium’s own compute nodes.

“Servers are stateless like they were with your arrays but have all the benefits that you’re desiring with converged infrastructure,” Nunes said. “It gives folks just a lot more freedom to get the job done.”

That job may include running Red Hat’s virtualization stack or Docker containers on Datrium DVX servers. Datrium can now run containers on both bare metal and in virtualized environments, Nunes pointed out.

Containers’ double-edged sword

Of course, the granularity of containers comes with a host of challenges in addition to its benefits. “It’s kind of a double-edged sword; you can manage everything at that level, but now you’ve got thousands and thousands of them,” Nunes said.

To reign in the swarm, Datrium has built in a management system that puts containers into “Protection Groups.” Naming conventions auto-assign newly spun-up containers to a group with no manual work necessary.

Automatic container cloning across servers and ability to reproduce “pristine” state in development environments are additional features of Datrium’s new release. Built-in services in Datrium’s Data Cloud on top of DVX include analytics, disaster recovery and data deduplication.

“We’re probably consuming 95 percent less capacity than you otherwise would have in your more traditional storage environment,” Nunes said.

The system breaks ground by covering vSphere, Red Hat and containers in one, and should win more customers for the company, Nunes explained. Datrium’s install base grew 50 percent in the last quarter alone, he said.

Watch the complete video interview below.

Photo: SiliconANGLE