Main Nav

Combining Data Center Consolidation and Server Virtualization: The Path to Nirvana or the Road to Hell?

Tuesday, March 30, 2010


IT organizations combining data center consolidation and server virtualization must understand how the sequence of consolidation operations (i.e., server virtualization before, during, or after data center consolidation) can impact different aspects of the project. With that understanding, an institution can make the best choice for its particular set of circumstances. A successful combination of server virtualization and data center consolidation yields benefits, including a flexible infrastructure, efficient use of IT resources, reduced costs, and a better posture for adoption of cloud-related services. In Burton Group's opinion, the best path to this “consolidation nirvana” is to perform server virtualization before data center consolidation.

Links to documents within this file might require secure access to restricted websites.

Burton Group( provides technically in-depth research and advisory services for colleges and universities, government agencies, and commercial enterprises. Burton Group's practical and unbiased research and advice helps technologists make smart IT infrastructure decisions in increasingly complex environments. Burton Group covers directories, identity management, application platforms, architecture, and network and telecom infrastructure topics. Like ECAR, Burton Group is an unbiased advocate for the user and more than 80% of Burton Group's clients are user organizations rather than suppliers. EDUCAUSE member institutions can become users of Burton Group research services through EDUCAUSE pricing. Burton Group is an ECAR partner and can be contacted by email at or by telephone (801-523-6023).

Download This Resource

Tags from the Community



..which is that it assumes (without making it explicit) that all readers will be running Microsoft Windows-based servers.

Indeed, the only instance of 'Windows' in the whole piece is in the author bio.  And 'Linux' never appears at all.   This would appear to suggest that the report is OS-neutral or that the 2 OSs are identical in their need for virtualization. IMHO that is not the case.

The instability of Windows servers running more than one application has driven the Server Virtualization (SV) industry and for that reason, it has prospered.  If a server is unable to run more than one application, it will spend most of its life running well below its capabilities and thus waste hardware (HW).  Therefore it makes sense to run entire multiple OSs on the HW, even while wasting the multiple OS memory footprints. If the server OS could stably support more than one application, then the hardware could be better utilized and piling technology layers atop each other to enable even a modest increase in HW utilization utility would be less attractive. For a peek into running a medium-sized, user-centric IT operation on Linux servers (and thin clients)  you could do a lot worse than visit:

Virtualizing Linux/Unix servers is much less compelling for the reason that they are meant to support multiple applications and do quite happily, under very heavy load for months without reboot.   There are reasons for virtualizing Linux/Unix servers as well, but have more to do with failover/high availability.  And since in many situations, there would need to have backup hardware for such services, virtualization itself becomes less of an issue.

While the Windows Server OS has gotten progressively better, it still does not seem to be anywhere near as price:performance competitive, nor as stable as Linux, hence the continuing demand for SV products.  For virtually :) every service that Windows Servers can provide, there exists a competitive Linux technology that will probably do it better, faster, & in the long term, cheaper.  (Transition costs will almost always make a conversion more expensive in the very short term; ie using a bicycle instead of a car to commute a mile to work will initially be more expensive due to the cost of the bicycle, rack, panniers, etc.)

Not to promote a comment into an essay, let me conclude it by agreeing that while SV is a compelling approach for increasing the efficiency of a poor technology, it does not necessarily convert it to a good technology.