Converged infrastructure -- virtualization hardware and software bundles preconfigured for rapid deployment -- has become a major focus for vendors.
There are several converged infrastructure products and initiatives on the market today, and undoubtedly more are to come. The prime example is Cisco Systems' Unified Computing System, a virtualization hardware offering that combines Cisco servers and networking with VMware vSphere and EMC storage.
Those three vendors also have the Virtual Computing Environment coalition, which makes preconfigured virtualization hardware and software bundles called vBlock Infrastructure Packages. And Cisco's main rival, Hewlett-Packard, is also placing its data center chips on converged infrastructure with its BladeSystem Matrix and other offerings.
The converged infrastructure push comes from vendors, so the question remains: What about users? Members of our Server Virtualization Advisory Board weighed in on the issue as they answered these questions:
Does converged infrastructure really address a customer need? When might these virtualization hardware and software bundles be a good a fit, and when should you avoid them?
Jack Kaiser, GreenPages Technology Solutions
In general, these reference architectures are a very good thing for our end users. They are integrated, tested and validated technologies that have been purpose-built to maximize virtualization density ratios. The other benefit is there is just one support organization to call in the event of a failure. This is important to customers, because there will be no finger pointing from one manufacturer to another about whose technology caused the issue.
They are best suited for end users who are designing a new data center or undergoing a major technology refresh. They should be avoided if the end user has recently made a significant investment in a technology or manufacturer that is not part of the bundled package.
Rob McShinsky, Dartmouth Hitchcock Medical Center
Vendors within the medical industry often certify supported hardware and software for their applications -- for reasons of ease of deployment and financial gain. These vendors see a common set of hardware, running a standardized virtual server platform, as the new norm for certain implementations.
From my organization's perspective, however, I do not see how these bundles would be accepted in pre-existing environments. Investments in hardware, software and administrative skills would be diluted, potentially making management less efficient.
I have always been a proponent of selecting the best of breed, which does not seem to be the latest mantra for virtualization vendors these days. Hopefully these types of alliances will not paralyze the selection of effective technology.
Greg Shields, Concentrated Technology
Preconfigured bundles of hardware and software bring modularity to infrastructure. They're an inevitable next step in virtualization, showing how the industry is maturing. And yes, your business will find itself using them more and more over time.
Think of the problem in this way: In the "old" days of just a few years ago, creating a virtual infrastructure involved a substantial amount of performance engineering and/or guesswork to right-size the hardware for the software.
Now, with converged infrastructure, you can purchase what I'll call a "unit of capacity." This bundle is engineered by the manufacturers' experts, who have the ability to assert (or even guarantee) its capacity. No more guesswork. No more over-buying or under-buying.
Our industry will be seeing more and more of these bundles as time goes on, because they make your job easier. Now your job is to figure out how best to integrate these units of capacity into your existing infrastructure.
Rick Vanover, Alliance Data
The concept of "infrastructure in a box" is appealing under the right circumstances. The biggest issue I see with an all-in-one solution is that there may be "throwaway" from existing investments.
Many organizations have disparate product lifecycles that are not on the same expiration schedule. Unless all key components are due to be replaced at once, there is the risk of not maximizing previous investments.
But as a new data center is commissioned, or major infrastructure components are due for replacement, the all-in-one solution becomes attractive. Looking into the future, I do see all-in-one solutions having a place -- either as a very well-defined unit of incremental growth, or as a simple solution for new implementations.
Have a question for the Server Virtualization Advisory Board? Email Colin Steele, Site Editor.