I’m seeing a lot of this, many industry figures within the industry have a lot of evidence that suggests that the further adoption of x86 Server virtualisation strategy is stalling. After being involved in various original Virtualisation consolidation projects, and more recently projects that involve virtualising at quite later on in the Virtualisation trend I am definitely beginning to see this statement stand true. In this post I will explain my view on VM stall and will provide explanations based on what i’ve since the Virtualisation era began;
A large slow down of adoption of more virtualisation in Infrastructure is the cost to actually run it. Virtualisation technology has certainly increased in overall cost. Compare virtualisation software available today with original incarnations and you’ll find that most vendors have hiked the primary purchase and renewal costs of this software dramatically. You can argue that technology available today compared to original Hypervisor variants have additional functionality or in vendor talk “Bang for Buck”. However those features are also indirectly limited by cost, due to the adoption of those technologies not exactly being a turnkey implementation in large enterprises any more. Also additional features or functions in some cases perform tasks or features that either are done by alternative methods and software to provide “just enough” value to an organisation.
Ok nothing really needs to be said here about this one its old hat, but unfortunately its Stalling Virtualisation strategy. We have been subjected to sprawl since Virtualisation strategy arose. And the reality of deploying yet more management technology to control VM sprawl with a Service Catalog is expensive, reliant on knowing your business processes, reliant on educating your business and requires large amounts of investment buy in amongst many other problems.
Nothing is left except….
Tier One based workloads, and I expect they are not as widely adopted as virtualisation vendors state. There is a certainly a stall in the uptake adoption of Large Database, Middleware and Application based servers. I’ve found in larger enterprises very few are lucky to be gaining full acceptance across there organisation to be able to host Tier One on virtualised platforms. Also important to note is the physical servers hosting Tier 1 are not exactly under utilised and obtain quite good value for money. Also remember costs of designing, transformation, testing and integration of Tier 1 apps before they can move into a Virtualised world.
Lastly another issue with reviewing the potential to migrate Tier 1 workloads onto Virtualised estate is software licensing which is next on my list of stall factors…
I think this is more of an issue for the more “greedier ISV’s” used, with a prime example (amongst many) being a large database company that is used by most shops. Very few actually virtualise platforms to run the software they produce based on there rip off pricing strategy. To counteract this you either have to adjust how you architect the estate with siloed clusters which incur more cost and operational complexity OR pay through the nose, and most struggle to justify to a stakeholder that the costs outweigh technical benefits.
This situation is changing (thank god) but it is a struggle to educate stakeholders that it actually is, with vendors very rarely actually stating they are Virtualisation friendly.
Its ironic that to Virtualise it is now beyond economical reach, we have excellent technologies that allow it to be done yet we are limited by increased licensing costs from Hypervisor vendors and ISV’s who have now completely tailored there licensing plans to ensure they do not fall short by it being Virtualised.
Lets hope the stall trend does not continue, Virtualisation technology has served a good purpose to organisations for many years and it would be a shame to see the further adoption of it fail due to basic greed.