If the deployment rate of server virtualization continues at 2010's pace, virtual systems administrators are certainly going to have their hands full this year, many IT industry executives believe. The reason is obvious: As the size and scope of the virtual environment grows, the more complex it becomes and the more difficult to manage and optimize.
This situation has made virtual stall a reality, says Bryan Semple, chief marketing officer at VKernel, an early virtualization management player.
In a recently completed Virtualization Management Index report, VKernel looked at virtual machine (VM)-to-host densities in more than 1,500 environments from 50 to 5,000 VMs. "What we saw surprised us: The larger the environment, the worse the density. Just the opposite should occur - the larger the environment, the greater the ability to balance loads, shift loads and maximize underlying resources," Semple says.
He attributes the problem to the ineffective use of capacity management. "System administrators are dramatically over-allocating resources to ensure performance since there is such uncertainty to the actual resource requirements for an application. Mission-critical applications don't get deployed into uncertain environments. Especially when the ROI is limited due to the need to significantly over-allocate resources to remove the uncertainty."
Few enterprise server teams have invested in the management and automation tools that can help them reduce workloads and improve their abilities to manage the environment more efficiently, agrees Jay Litkey, CEO of Embotics, another early virtualization management company. "Consequently, they already are running flat out with no cycles to spare."
For that reason, Litkey says, enterprise IT will continue moving away from point tools to more all-encompassing virtualization management systems. "VM monitoring, information sharing, lifecycle management, capacity/performance management and self-service will become key components of any virtualization management system, and solutions that integrate and automate these disciplines within an enterprise architecture will be essential," he says.
What's more, he adds, "Management systems will become easier to deploy and use: Administrators simply don't have the time for the traditional complex implementation and high learning curves."
To Bob Quillin, chief marketing officer at Hyper9, another early virtualization management company, the months ahead will see "the emerging dominance of virtualization management over physical management - cracking the code on key issues such as virtual sprawl, dynamic capacity planning and management and automation for the private cloud."
On that latter point, he says, "In 2011, real-time analytics and intelligent decision support will be critical to automating the manual and complex resource allocations" enabled by self-service portals that allow users to grab virtual machines in private cloud environments while "virtualization management automation also will help IT teams gain better visibility and control into public cloud deployments - assessing the cost/benefit/risk of each." This includes the ability to move some VMs back into the data center while moving other VMs out, he adds.
Automation may well dominate the virtualization world in 2011, agrees Andi Mann, vice president of virtualization product marketing at CA. As he wrote in a recent blog, "Virtualization deployments require more skills, more people, faster cycle times, and fewer errors to overcome virtual stall and be successful. Automation is the only way to deal with this conflict, accelerate virtualization, and build an efficient, cost-effective, dynamic data center. It is also essential to cloud computing."
What enterprise IT management pain points do you hope to address in 2011? Drop me a note and let me know.