Hyperconverged solutions can't live without flash

In the world of hyperconverged virtualization, flash is important. It forms a big part of the hyperconvergence value proposition as vendors create distributed hybrid storage arrays from local resources. But hyperconvergence is moving away from every node in the cluster having an identical storage/compute ratio, and this means …

[Anonymous Coward]

Whoosh.......

I think I understand what you're saying, but it might help me if the editor(s) could annotate all of the acronyms.

[MityDK]

Nodes Dedicated to storage?

"Two nodes are MCS+NVMe "godlike" nodes where the really sexy workloads live. Four are storage-heavy nodes that have heaps of storage and a bunch of really fast PCI-E or NVMe storage."

Oh you mean like an all flash or Hybrid storage array?

Novel concept.

[Trevor_Pott]

Re: Nodes Dedicated to storage?

Except these hold only replicas while primary workloads are served off the systems running those workloads. Big difference.

[ntevanza]

Tail, dog

Can't help thinking that the full virtualization tail is wagging the infrastructure dog here. If we could partition OSes properly, we wouldn't need to virtualize whole overweight, needy, solipsistic OSes. The waste is staggering.

Almost all the I/O in a fully virtualized machine is superfluous to the primary workload. The persistent data don't live in the VM. All the flashy I/O that happens inside the VM gets thrown away.

The only, and, as John Major would say, not inconsiderable, advantage of this arrangement is that it's easy to understand. The obvious disadvantage is that it thrashes the bejeezes out of your infrastructure.

Or let's be honest, is all that inefficiency an advantage, when it creates a need for new shiny shiny?

What is the real world status of the various partitioning, paravirtualization and containerization alternatives? Summary please, I'm too busy fighting snapshots.