Benefits include greater visibility into virtual server environment
At a glance, Thomas Fairfield, manager of technology at Elliot Health System, says he can see that the overall CPU utilization for his server clusters is at 20% or less, that the number of CPU-ready virtual machines is down in the single digits – if even there – and that he’s got zero memory ballooning.
This down-and-dirty information on his virtual server environment comes to him via Hyper9, a much-watched virtualization management tool now in its 3.0 release.
Top 5 network management investments
Hyper9 3.0, introduced earlier this month, lets users address the three capacity management phases they encounter as they scale up their virtual environments, says Jonathan Reeve, vice president of product strategy at Hyper9.
For enterprises in Phase 1 deployments, Hyper9 3.0 features a capacity operations dashboard that provides real-time views of resource consumption across CPU, memory and storage. For Phase 2, it offers new advanced analytics aimed at helping with capacity optimization. And for companies reaching into cloud services, or Phase 3 deployments, Hyper9 3.0 provides support for resource purchase and public cloud deployment analysis.
At Elliot, a midsized hospital in Manchester, N.H., with 70 to 75% of its server infrastructure virtualized, Hyper9 has proven a godsend in streamlining the environment and saving management man-hours, Fairfield says.
Using Hyper9, for example, Fairfield says he’s moved VMs among three server clusters to balance out overall CPU and memory utilization. Plus, he’s reduced the number of servers sitting at CPU ready at any one moment from roughly 85 to just one or so, he adds.
Hyper9’s reporting features come in handy, too, Fairfield says. “I can pull up a graph showing overall CPU utilization in a cluster. If I see spikes, I can compare those against the host, the VMs or the data store in that cluster to see if I can see any correlations – for instance, is something on the SAN side causing spikes in the cluster?”
“I’ve got 210-plus virtual machines to manage with a staff of three and a half server people – and that’s hard. We had been doing a lot of reactive work but this lets me be more proactive, he adds. “We’re able to see things instantly that used to take us minutes to hours to get at previously.”
Technically Fairfield and his staff can get the same management information out of VMware vCenter, “but not as cleanly,” he says. “That data is cumbersome to find and took a lot of work to get the visibility we wanted.”
What’s more, after looking into larger-scale end-to-end management systems, Fairfield says he determined those would be too costly for his purposes. “While they’d give us good visibility,” he says, “we decided it’d be better to get good insight in each area – let the network team do the network and the server team the servers.”
Schultz is a longtime IT journalist. You can e-mail her or find her here.
