Mike Wronski (Middle East)- Preventing VM Sprawl When Adopting Virtualization

Server virtualization promises to reduce costs, simplify management, and enable IT to do more with less. The global growth of virtualization supports this claim. IDC and other analysis' show that hypervisor sales are growing steadily since 2010, with even more growth forecasted specifically for desk top, VDI, use cases.[i]  The cost savings are very attractive to Middle Eastern enterprises, where power and viable space come at a premium. According to Sandeep Saxena, Assistant Vice President at Raqmiyat, a UAE-based systems integrator, a large portion of this EMEA growth will come from the Middle East. With 2010 numbers forecasted at $4.79 billion[ii], we can only assume 2011 virtualization spends will be even larger.

The ROI of using virtualization for consolidation is simple to understand.  If, for example, a datacenter with 2000 physical servers can be consolidated to 100 physical servers using a virtualization technology like VMware's ESX hypervisor, the power and cooling savings are obvious. What is often missed is that consolidation comes with other management challenges. These challenges, if not dealt with properly, can lead to unforeseen costs that would erode the ROI of consolidation.

Virtualization provides simple automated server provisioning. This ability to provision servers at will is a great operational efficiency, but when not well managed it can lead to an unforeseen operational cost.  Groups that have moved beyond simple consolidation have run into a state that the industry has labeled "VM Sprawl".

VM Sprawl is defined as the case where virtual machines (VMs) have been deployed haphazardly, without complete business justification or proper IT management that results in wasted resources (CPU, RAM, and storage).  Waste can come in multiple forms, but common instances are redundant workloads and orphaned VM's whose purpose has been forgotten.  Another, more insidious form of sprawl is categorized by users provisioning VM's without proper authority. As they are deployed outside of any process, these "rogue" VM's are even harder to track down and control.  Sprawl has always existed but it was easier to control in the physical data center as the process to deploy servers was far more complex and time consuming, requiring racking of equipment, cabling and power/cooling budgeting and thus easier to observe its progress.

Now that we understand sprawl and its impact, let's discuss methods to prevent or regain control of VM sprawl. Purpose-built virtualization management tools (i.e. tools beyond the standard provisioning tools provided by the virtualization vendor) that provide detailed visibility and policy controls for the entire virtual infrastructure are integral to preventing sprawl.  Traditional systems management tools may not be well integrated into the virtual environment and thus may not provide the visibility required to even identify sprawl, much less the options required to regain control.

When choosing third party virtualization management tools it is important to look for specific feature sets that can be used to control sprawl. The following is a high level list of the most important features to look for:




Lifecycle Tracking

Ability to track VM Lifecycle (from provisioning to reclamation) including storage, CPU, RAM and software inventory allows reporting on where resources are used and identification of when they can be reclaimed.


Automation and Workflow

By introducing some cloud like concepts to automate and add approvals based workflow to the provisioning process, the likelihood of rogue or redundant VM's goes down. This level of process control allows for hard and soft limits on VM lifetime giving predictability to resource reclamation activities.

Charge Back  / Cost Modeling

Tools that provide an interface for reporting on costs or charge back data provide financial information that is then used as part of justification for new VM's. VM owners are made aware that their usage has real costs associated and can be held accountable.

Inventory Tracking / Search

Comprehensive VM inventory that is backed by the ability to search/filter by line of business, application, or resource utilization will provide the visibility not provided by native management tools

Software Asset Tracking

Software inventory discovery to locate and identify where operating system and application software licenses are used or unused to reduce software licensing costs.

Capacity Management

Provide the ability to see resource utilization and map back to VMs and then to owners. Provides both knowledge of where resources are used and indicates when resources will be exhausted.


The primary take away from this article should be that virtualization is growing and enterprises are adopting this technology at a fast rate, with the promise of cost savings and operational efficiency. What has been shown is that if not addressed, the fast rate of adoption, combined with the dynamic nature of the technology can lead to a VM sprawl condition that will negatively impact the cost savings. To address and prevent sprawl, IT administrators should deploy purpose-built tools that leverage the new context and control aspects of virtualization in ways not previously available in data center management.

Mike Wronski is the VP of Product Management for Reflex Systems. Wronski holds a CISSP and Certified Ethical Hacker certifications, as well as an MBA, and a Bachelor of Science degree in Computer Engineering from Florida International University.







« Russell Rothstein (North America) - Service Performance and Availability Issues in the Cloud - Part 2


Chris Pick (Europe) - Are Enterprises Ignoring the Elephant in the Cloud? »