Inefficient allocation of power and cooling resources in data centers can result in unused power capacity. This increases costs and artificially limits the number of servers that can be deployed. To address the overallocation of power budgets, many server vendors have introduced power management solutions that try to enable the capping of power consumption on a per-machine basis to reduce the unused capacity. However, some of these solutions can have negative or unintended consequences when they overlap, conflict, or otherwise interfere with the power management capabilities that are supported in Windows Server operating systems.
This paper summarizes common approaches to power budgeting in the server marketplace and recommends best practices to ensure that these solutions work as intended and interoperate with Windows Server power management capabilities. The information in this paper is intended for product planners, developers, and system designers.
This information applies for the following operating systems:
Windows Server 2008
Windows Server 2003
Included in this white paper:
| • | Introduction |
| • | Limitations of Current Solutions |
| • | Throttling Mechanisms |
| • | Adjusting Windows Power Policy |
| • | Using ACPI Notifications |
| • | Using CPU Platform Thermal Control Circuits |
| • | Combining Power Throttling Mechanisms |
| • | Windows Processor Throttling Policies |
| • | Dynamic Processor Throttling Policies on Windows Server 2003 |
| • | Processor Performance States on Windows Server 2008 |
| • | Best Practices and Recommendations |