Earlier this month, a survey commissioned by IT management software provider CA Inc. found that 92 percent of U.S. IT budgets already include energy-efficient software solutions, with virtualization of servers (55 percent) and storage (56 percent) dominating U.S. efforts. That got me thinking — if energy efficiency is so critical, we should show more appreciation for power management tools that can make next-generation systems (virtual and physical) more efficient by making them more responsive.
The on-demand crowd (of which I consider myself a part) loves to talk about cloud computing and dynamic data centers, about policy-based provisioning and resource allocation, about workloads moving from machine to machine or cloud to cloud. But just because these on-demand lovefests always end before someone brings up power management, that doesn’t mean providers are ignoring the issue. Here are four tools that raise the profile of power management using performance-based policies:
Appistry EnergySaver: An add-on to Appistry’s Enterprise Application Fabric, EnergySaver lets customers define performance-based policies for power management. When resources are no longer necessary, EnergySaver puts them to sleep, only bringing them back up as needed. If target resource utilization for a system is set at 50 percent, EnergySaver can power on or power off machines to keep aggregate usage at the predefined level.
Cassatt Active Power Management: Available across Cassatt’s line of Active Response solutions or as a standalone product, Active Power Management takes into account a variety of factors — from system performance to your electric utility’s peak and off-peak pricing schedules — to determine when to turn servers on and off. If machines are needed for failover or to maintain application service levels, Active Power Management can bring them back up automatically.
Virtual Iron LivePower: A standard but still “experimental” feature of Virtual Iron v4.4, LivePower lets users set pre-determined utilization levels for physical machines. (Virtual Iron calls the feature experimental because “it has not yet been widely tested in production environments.”) When utilization falls below that level, LivePower leverages Virtual Iron LiveCapacity to move VMs running on that machine elsewhere. The physical server is shut down, rebooting and reentering the pool as demand picks up.
VMware Distributed Power Management: Another “experimental” feature, Distributed Power Management (DPM) is part of VMware Distributed Resource Scheduler (DRS). DPM monitors power consumption in DRS pools and uses vMotion to consolidate workloads onto fewer physical servers automatically. Unneeded physical machines go into standby mode and come back online as predefined utilization policies dictate.
Finding the right utilization levels for your data center is key to getting the most from the latter two tools. Even where consolidation is a driving force behind the decision to virtualize resources, experts suggest running physical machines at less than full capacity keep native performance high and to safeguard against unforeseen problems.