On-prem control planes, dark-site upgrades, and multicloud policies target regulated deployments
Nutanix is taking a pop at VMware – again – as it unwraps features that it says allow customers to run distributed sovereign clouds.
"It's actually not too hard to create a sovereign cloud, if you just block off all access," claimed Lee Caswell, senior veep of product and solutions marketing at Nutanix.
He added "as at least one of our competitors" presumes, "you can live with just one virtual private cloud." But he argued "there's a tailwind for distributing data and applications more in the future than we've even seen to date."
That means having the ability to flex sovereignty across endpoints, and "to be able to set security policies locally and propagate them globally."
Nutanix told us its Cloud Platform now gives "orchestrated lifecycle management of multiple dark-site environments, along with on-premises deployment options for governance and control planes."
So the company is rolling out on-prem support for its previously SaaS-only Nutanix Central console service. "That means you can now do upgrades, for example, in dark sites without having any external connections."
The Data Lens ransomware detection software will also be offered on-prem. "We're basically taking everything that was cloud-delivered and making sure that that's operative in an on-prem or dark-site environment."
Back in the cloud, its GC2 service – government cloud clusters – is being made available on AWS, so federal agencies or "anyone in EMEA" can build out sovereign clouds, Caswell told The Reg.
At the same time, the NC2 cloud cluster service is now available on Google Cloud, which presumably should herald GC2 on Google in time. In Europe, NC2 is available on OVHcloud.
Nutanix is gaining more experience on different cloud vendors' stacks, Caswell said. This is important for the development of "security policies that move and are extensible across hyperscalers."
More granularly, it will offer multicloud snapshotting as part of better resilience for multicloud operations.
"The idea is that you're offloading the snapshots from the primary storage that frees up performance," Caswell said. "And now you've got an S3-compatible and sovereign controlled endpoint where those snapshots can then be restored at some time in the future."
While all the new features were couched in terms of supporting AI - sorry to the AI skeptics among the readership - Nutanix also unwrapped a specific LLM metrics dashboard.
Caswell said the adoption of AI is similar to the cloud, with the emphasis shifting from just getting AI done to "AI smart." Put another way, companies need to understand and control costs. But this becomes trickier than simply managing cloud resources, given the unpredictability of token use and generation.
"How can I go and make sure that I understand how models are being used, how tokens are being generated and used?" he said. That includes the ability to track who is using what models.
This has implications for managing memory and power, he said, as well as planning the resiliency of the underlying architecture.
"Why treat AI differently from what you did with business-critical applications and databases and workloads that were mission-critical?"
This is intended to give customers more visibility as they look to run models locally, whether in virtual private clouds or at the edge.
"What we're able to do is start looking at token usage, and you know, you've got predictable costs, but you can also start looking at what's the efficiency. How are you using these?" ®