Virtualization Benefits: Beyond Server Consolidation
Virtualization has provided economic benefits, but there are more benefits to be had.
by Lilac Berniker
The initial drive to virtualize was simple: server consolidation. The savings in floor space and the associated utilities was more than enough to justify the cost of the platform. Organizations flocked to consolidate systems and tidy up their data center and continue to be attracted to the technology. However, in these economic times, IT executives are being asked, “What more can you do to save money?” Server consolidation was tremendous, but will this be a one-hit wonder?
The truth is, although virtualization cut down on some costs, such as ongoing operating costs of electricity and looming capital outlays for new data centers, it did nothing to reduce other elements of the IT budget. In most companies, new virtual environments are equally as over-allocated as their physical predecessors and management costs are not lower. The field is ripe for picking up additional costs savings, but the excess costs are not as tangible as floor space.
There are four key considerations for continuing down the path of cost-cutting success, all of which work hand in hand with your growing virtual environment.
1. Intelligent Resource Allocation: Virtualization has been so successful in enterprise IT that requests for new virtual machines are ongoing. Of course, these VMs are considerably cheaper than physical boxes, so these requests are typically granted.
What are people asking for? Typically, physical boxes were over-purchased (and thus, over-provisioned) to their business owners, in case they needed that extra horsepower or experienced a spike. The beauty of the virtual environment is that it is flexible and sufficiently dynamic to account for a slowly growing or even somewhat unexpected resource need on-demand, without requiring the entire amount to be allocated at the outset. Still, many companies continue to over-provision VMs rather than modify the habit to suit the environment.
Of course, you can only allocate resources tightly if you have the right alerts and controls in place to ensure customers don’t run out, which brings us to the second consideration.
2. Increasing Automation: The growth of virtual environments has been so rapid that we rarely stop to consider just how big the infrastructure is getting. There are millions of dollars in physical resources under management, serving up hundreds (if not thousands) of VMs to many business units. It is a fair bet that every administrator has a lengthy to-do list of settings to change, options to check, and reminders to acknowledge for the next week or next month. How long before that list becomes unmanageable?
Most IT groups can’t afford to add staff due to cost pressures, meaning these lists will only get longer for the same hardworking administrators. The solution is automation: let software handle the repetitive tasks of monitoring machines, checking changes, and even performing adjustments to the infrastructure. Then you can allocate just-in-time resources when a machine reaches capacity, or unplug a VM when its time is up, reclaiming those resources.
3. Resource Reclamation: Environments have grown around both the conversion of physical servers to virtual servers and the establishment of straight-to-virtual new machines. These new machines were often intended to be temporary, but years later few have been decommissioned. Having no physical presence, they are easy to forget. Each of these idle VMs represents resources, from CPU to storage, which could be put to better use. Some estimate that as many as 30 percent of VMs are idle. Ironically, you could gain significant server consolidation benefits if you removed some of these machines.
The more subtle form of resource reclamation comes from identifying underutilized resources. If a machine is allocated 20G of storage space and has only consumed 1G for the past six months, it probably will not need the other 19G immediately. Automatically identifying machines that have been over-provisioned, then taking action, can defer investments in new resources.
4. Recasting CPU and Memory: Physical machines were rarely shared between business owners, and most were purchased by the business itself. Virtual machines, on the other hand, are purchased by IT to serve the business. Thus, a server is a means to provide a service to the business. Capacity, then, can be thought of as “the number of businesses I can serve.”
From the business’s perspective, though, they no longer have “one whole machine,” so naturally they can be expected to question what exactly they are getting. The tendency will be to overprovision, as was always done with physical servers -- but that runs counter to good infrastructure management. In truth, you aren’t giving the business less -- you’re giving them less initially, with the assurance that they can grow as they need to. In the checklist of requirements, virtualization has added a few new boxes, including flexibility, high availability, and even disaster recovery.
By recasting the value being provided away from simply CPU and disk to a complete service, the business and IT can work together to find a solution that addresses the needs without wasting resources.
In these economic times, everyone is feeling the pinch of budgetary cuts. Virtualization provided dramatic improvements in the past few years. In the future, it can continue to fuel more-optimal IT infrastructure and processes if we look beyond consolidation and focus on operations.
Lilac Berniker leads business development and marketing for Fortisphere, a virtualization management software company. She has 10 years of experience in the virtualization and grid computing space. Part of the grid and virtualization community since 1997, Lilac has held roles within the grid open-standards organization, The Globus Alliance, and the Grid division at IBM. You can contact the author at firstname.lastname@example.org.