Diamanti today announced it is adding support for both volume encryption and self-encrypting drives along with support for asynchronous replication to its bare-metal platform for hosting containers.
In addition, Diamanti’s D20 appliance now can be configured with Intel Cascade Lake processors alongside graphical processor units (GPUs) from NVIDIA. Diamanti claims the Intel Cascade Lake processors should boost performance on average by 30% when employed in a D20 appliance.
Jenny Fong, vice president of marketing for Diamanti, says version 2.4 of Spektra, the software Diamanti provides to create a hyperconverged infrastructure (HCI) platform using a bare-metal server, will make it easier for IT teams to secure containerized applications and ensure availability by replicating encrypted data offsite. The goal is to provide IT teams with the tools they expect to be able to manage Kubernetes environments at scale across a fleet of Kubernetes clusters, she says.
For the past several years Diamanti has been at the center of a debate over the merits of deploying containers on bare-metal servers versus virtual machines. While the company makes open source virtual machines available for customers who require it, Diamanti has long contended virtual machines add a layer of unnecessary overhead. In the case of VMware specifically, Diamanti notes that overhead not only impacts application performance but also results in additional fees for commercial instances of virtual machine software and the tools required to manage them.
Naturally, there’s a lot of debate over the impact virtual machines have on container application performance. With the latest version of its hypervisor platform, VMware argues application performance overhead has been minimized to the point of being immaterial. As such, VMware says IT teams can continue to use the platforms and tools they already have and know to manage Kubernetes. At the same time, however, VMware is also making available a cloud service through which IT teams can manage any distribution of Kubernetes.
Fong, however, says there are emerging applications involving analytics infused with machine and deep learning algorithms running on Kubernetes where it’s already been shown that performance matters as Kubernetes is now the de facto standard for running these types of applications. There are also many greenfield IT environments that have not previously invested in IT infrastructure, she notes, adding that even in cases in which organizations still require virtual machines to isolate workloads, lighter-weight open source virtual machines will provide a viable alternative.
Regardless of the hardware path pursued, Diamanti is trying to usurp incumbents such as Dell Technologies and Hewlett Packard Enterprise (HPE) that have long dominated local data centers. Of course, it’s still not clear what percentage of Kubernetes workloads are being deployed in on-premises IT environment versus public clouds. However, it is already apparent servers optimized for Kubernetes deployed in local data centers will in time be a multi-billion market. The only issue unresolved now is when IT organizations might have the budget available to fund the acquisition of additional IT infrastructure.