virtualization.info

Strange!!! It looks like a rigged communication... Maybe VMware roadmap is too hard to follow for some competitors...

Actually, the VMware roadmap is too hard to follow for most companies. The competitors' roadmaps are equally sinusoidal.
Virtualization continues to shift the costs to a different area and the easiest way to justify this technology is to place "business continuity" in the top 5 reasons for adoption.

I personally think that people are starting to realize that it's harder to KISS when adding YALOC (Yet another layer of complexity) and that each layer of complexity causes an exponential increase in costs.

Virtualization on servers is really useful only when you need to dynamically start and stop sandboxable os-dependent processes. Ultimately it's an inefficient kernel scheduler that happens to be able to run multiple operating systems.

I've been noticing this for a while, but I think it's just the natural point in the hype cycle. (See my blog post here.)

And Christopher, I'd say the evidence is that we're witnessing something happening that's a little deeper. Even if you dismiss arguments about transforming the datacenter and business processes, just try to pull VMotion and DRS and HA and SRM out of a VMware admin's hands -- it's transformed their daily activity at a visceral level.

"Virtualization on servers is really useful only when you need to dynamically start and stop sandboxable os-dependent processes."

That sounds like every Windows server-side application. Why do you think there are so many Windows VMs?

I'd like to see the math for that exponential increase in costs. Sure, you need to learn the technology, but it's incredibly powerful as a control and management improvement over single-OS-instance server infrastructure. Even if space/power savings and hardware independence aren't that important to you, there's a ton of utility in snapshots/templates/duplication/virtual appliances, etc.

I have to chime in with noisemaker here; virtualization, when done properly, is about REDUCING complexity by lowering the box count and standardizing your servers into self-contained entities that can migrate freely across your infrastructure.

>Virtualization on servers is really useful >only when you need to dynamically start and >stop sandboxable os-dependent processes

What do you mean by "only"? I would say that that is what x86 deployments are all about.... :-)

It would be like saying that a car is only useful from driving to point A to point B. Yes that is exactly what a car is supposed to do. Isn't that enough?

Plus add on top of that all the things you cannot do in a physical environment (VMotion being one of many).... Or do you think that a static 1Server-1OS-1Appl deployments are the future of the Datacenters?

It add complexity or an additional layer of software.... but is there any other way around the problem?

Massimo.

Frankly, I'm not surprised at all by these articles. If you walked the floor at VMworld, you'd know why. Almost every vendor on the floor was discussing the complexities of virtual environment management from their specific point of view. Whether you were dealing with virtualized storage or trying to manage the performance and availability of a multi-tier hybrid, virtual/physical business service, there was a vendor there to tell you how complex that was and how their solution could help. The main theme of the show floor could have been "Solving the Complexity of Managing Virtual Environments".

What this points out is that the tools for managing these environmnets are just evolving to support production environments. And the most interesting thing is that IT is not going to want virtual-only management solutions. The true winners are going to be vendors who can offer solutions that work across virtual and physical environments. It just doesn't make sense to create another technology silo with a set of specialized tools requiring specialized staff to run them.