Nutanix de-converges by allowing dedicated nodes for compute and storage

This could be the way to get HCI out of its ghetto

Nutanix, which made its name with hyperconverged infrastructure that bundled compute and storage in single nodes, has decided the time is right to offer dedicated storage and compute nodes.

The change is significant because today the nodes running Nutanix’s Hybrid Cloud Infrastructure (HCI) stack ideally aren’t a diverse collection of kit. Nutanix prefers you use largely homogenous hardware.

Allowing dedicated nodes means HCI can be used for more workloads. Databases, for example, need some nodes tuned to compute and others for storage. While HCI is a decent enough database platform, Nutanix thinks it can now improve on it.

Diverse nodes also mean Nutanix can pitch HCI as more flexible, and therefore more attractive for more workloads. Rightly or wrongly, HCI – both Nutanix’s product and the concept of hyperconverge.d infrastructure – have been associated with certain workloads, with desktop virtualization at the fore.

At a more prosaic level, this change also gives Nutanix a better chance to pitch for database business, because today scaling an HCI rig means adding a node that includes storage and compute. Database vendors, however, care not for the fact that part of a Nutanix node runs storage services and charge for all the host CPU’s cores. Customers can be left paying licenses to run databases on cores that run storage, not the DB.

Dedicated compute nodes, and the cores they contain, can now be allocated entirely to databases, in theory keeping database licensing costs lower.

Dedicated storage and compute nodes get even more interesting in the public cloud and Nutanix’s cloudy NC2 clusters – the cut of HCI running on bare metal in AWS and Azure.

At Nutanix’s .NEXT conference in Chicago today, the IT biz's execs told The Register that de-converging will mean it becomes possible to use different cloud instance types in the same virtual cluster. Users could therefore pick an instance type designed to host databases and another storage-centric instance type to store data and manage them all as just another Nutanix cluster.

Competing hybrid cloud frameworks can’t currently work with a mix of instance types: Nutanix might be onto something here by making its cloudy stack able to take advantage of the diversity offered by hyperscale clouds.

Data services

Databases also matter to Nutanix because the other headline news from the .NEXT conference is more “data services.”

Nutanix has two definitions for that term. One can be applied to the managed database-as-service offering called NDS it’s sold for some time. At .NEXT it announced that service is now the flagship for Project Beacon, an effort to create a range of PaaS products that run on any cloud and don’t require the presence of Nutanix’s hybrid cloud infrastructure stack.

Nutanix’s schtick here is that PaaS is usually tied to a particular vendor. But its DBaaS will be available as a managed service on native public cloud infrastructure. Execs haven’t said what other PaaS products Nutanix plans to create, or when they’ll debut, but the direction is clear: services that developers rely on to underpin apps – usually related to data – will become PaaS products that Nutanix will happily run anywhere developers want them.

No timeframe for the delivery of more PaaS products under Project Beacon was discussed, not were specifics of future services detailed at the conference.

The other meaning of “data services” relates to data protection. Nutanix has taken its software-defined storage, snapshots, and disaster recovery tech, and applied them all to Kubernetes.

Another newbie is Multicloud Snapshot Technology (MST), which allows snapshots of Nutanix rigs to be moved into cloud native object stores, starting with the AWS S3. Doing so gives Nutanix users a new data protection and DR option.

Nutanix also likes the idea that you could take a snapshot of an app, move it to a cloudy data store, then start to run the app in that same cloud. Kubernetes is, in theory, supposed to make apps more portable. Nutanix thinks this effort contributes to that story.

Cloudy control

Also confirmed at the conference was Nutanix Central, a cloud management tool that “provides a single console for visibility, monitoring and management across public cloud, on-premises, hosted or edge infrastructure.” Providing that infrastructure runs Nutanix, natch.

Nutanix Central includes observability and other tools that app orchestrators covet. The tool will sit alongside Nutanix’s existing Prism management tool for local clusters.

The thinking behind all of the above is that organizations have stumbled into hybrid cloud without planning how to best do deploy, or with the tools to manage it elegantly. Nutanix hopes to develop and provide those tools.

Even Project Beacon reflects the company’s desire to tame the hybrid cloud. Today, Nutanix asserts, deploying an application to numerous targets and locations is painful. A PaaS approach therefore continues the company’s long-held strategy of making tech easier to deploy and manage.

No vendor ever says otherwise. But few are trying to act as an overlay across the many hyperscale and IT-as-a-service platforms buyers can consider in these cloudy times.

Nutanix has one great rival with similar ambitions and capabilities – VMware – and execs of the upstart told The Register that some of its current efforts are designed to catch the eye of Virtzilla’s customers as they fret about what Broadcom’s acquisition might mean. ®