Virtualization pros typically rely on trusty spreadsheets for virtualization management, but some administrators and software vendors have begun to evaluate the value of automated capacity management remediation.
More on capacity management
VMware vBenchmark goes beyond capacity planning
Storage resource management tools simplify capacity management
Virtualization capacity planning strategy guide
While automated remediation tools for capacity management have promise, administrators don’t completely trust them.
“There’s a certain level of control that needs to be had,” said Allen Firouz, CEO of Venturian Hosting, a private cloud hosting provider based in Miami. “Automation with oversight, I think, is the right balance.”
Since Venturian charges its customers by the server and for storage, rather than by CPU or RAM capacity, Firouz trusts Quest Software Inc.’s VKernel tool to automatically adjust for additional CPU or RAM. VKernel vOps Server automatically resizes disks as well, but because customers pay for that, Firouz said the company prefers to manage that process carefully and manually. Other IT pros have yet to select a virtualization capacity management tool, but see automation in their future.
“We do a lot of management based on experience right now – based on knowing the [virtual] environment through the staff we have,” said Rob McShinsky, senior systems engineer for Dartmouth Hitchcock Medical Center in Lebanon, N.H. Having a tool managing the environment automatically makes sense because “if we lost a key number of staff, we’d lose a lot of that knowledge base.”
The hospital continues to evaluate which capacity management tools to use, but for McShinsky, a Microsoft Hyper-V user, automation will ultimately be a part of the equation.
“I think it is the direction we’ll all go, especially as we bring in System Center 2012 [since] there's more automation built into that,” he said. “It’s just a matter of taking that step and learning to trust it.”
There are also third-party tools for specialized virtualization capacity management automation, but for some, the cost may not be worth the benefits.
“Quest has some specialized features for virtualization environments which may be better than Operations Manager, but is it worth the extra licensing at this stage?” he said. “Is that extra piece going to give us much more?”
Virtualization capacity management vendors, meanwhile, can be divided into two camps: those who argue that automated remediation of capacity contention issues with allowance for user control is the only way to keep up with ever-expanding virtual environments, and those who say even the biggest customers benefit from careful advanced planning rather than reliance on reactive automated processes.
Quest, VMware and Microsoft: Automated remediation with user control scales with the cloud
Vendors that offer automation with manual overrides claim that human hands can’t keep up with virtualization management as it evolves into cloud computing. According to these vendors, virtual environments get big and get complicated, and they get that way fast. Virtual machines (VMs) are so easy to deploy that VM sprawl is common and environments are constantly in flux. From this perspective, adding administrators for manual management puts too many cooks in the kitchen.
Vendors in this camp, including Quest, VMware Inc. and Microsoft, have released or are developing features that automate the remediation of capacity management problems such as the rightsizing of VM memory, CPU and disks.
All these vendors also acknowledge automation trust issues with built-in user controls and manual overrides. Quest, which updated its vOps Server software May 22, offers the options to control remediation via a scheduler, a button-click, or complete automation as a background process.
VMware hasn’t delivered automated remediation to its vCenter Operations virtualization management tool yet. Company officials say that when it arrives, it will be like the existing Distributed Resource Scheduling (DRS) feature, which can be set to fully automated mode, set to make a recommendation and let administrators decide what action to take or set to full manual mode.
Similarly, Microsoft System Center 2012 builds in automated processes within and among products in the System Center suite, but still allows users to trigger automated elasticity features.
CiRBA and Xangati: Do it right the first time
Other software suppliers say that IT professionals should trust their gut when it comes to their doubts about automated remediation. “Rightsizing” can be wrong, according to Andrew Hillier, co-founder and chief technology officer of CiRBA Inc., a predictive analytics software company that boasts large customers, including Bank of America, AIG, and Lowe’s.
“With Web servers, for example, you may not want to resize when you hit capacity thresholds, you may want to turn on more servers,” Hillier said. Companies may have change freezes during certain weeks of the month or months of the year, policies which automated remediation systems may ignore.
Instead, CiRBA takes a policy-based approach to capacity management that Hillier compares to a hotel reservation system. This method locks capacity based on future requirements and manages placement of VMs according to a “Tetris” of policy and capacity availability.
Hillier argues that this type of planning is more effective than reactive automation. Ideally, features such as VMware’s DRS should be inactive in an environment managed by CiRBA’s software, because capacity will have been managed effectively from the start.
CiRBA is also uninterested in remediation of problems, whether automated or not. “We think of ourselves as the brain, not the actor,” Hillier said.
On a similar note, Xangati, a virtualization performance management software vendor, added capacity management to its software lineup on May 15.
Xangati claims that storage capacity problems can’t be remediated automatically, either because more physical capacity has to be added to a system, or because “storage issues have to be worked [on] constructively, as a team, between the virtualization admin and the storage admin and cannot simply be done in the real world by ‘automation,’” according to a company spokesperson.