Energy Efficiency using Intel® Intelligent Power Node Manager

Contents

Executive Summary

Intel® Intelligent Power Node Manager is a software feature that resides on the latest Intel platforms including Intel® Xeon® processor E5 family, providing power, thermal monitoring, and policy-based power management for individual servers.

Neusoft Aclome* is a visualized, automatic and agile cloud management environment to deliver cloud service efficiently. This case study shows how the Neusoft Aclome cloud used Intel Intelligent Power Node Manager in the Intel E5-2680 platform to achieve a power saving of 13% with only 0.5% degradation in primary performance.

Intel Intelligent Power Node Manager (Node Manager)

Intel Intelligent Node Manager provides power and thermal monitoring and policy-based power management for an individual server. It is an out-of-band (OOB) power management policy engine embedded in Intel server chipsets. Processors can regulate their power consumption by manipulating processor internal power states. Node manager works with the server’s hardware and software and the operating system power management (OSPM) to perform this manipulation and dynamically adjust platform power to achieve maximum performance and power for a single node (Figure 1).


Figure 1: Intel® Node Manager Power management Closed Control Loop

Intel Intelligent Power Node Manager is available on the latest Intel platforms including Intel Xeon processor E5 family, Intel Xeon processor E7 family, and Intel Xeon processor E3 family. Node Manager is exposed through a standards based IPMI interface on supported Board Management Controllers (BMCs). It requires an instrumented power supply conforming to the PMBus standard.

Some of the key features of Node Manager include:

  • Dynamic power monitoring: Node Manager measures actual power consumption of a server platform within an acceptable margin of error of +/- 10%. It gathers information from PMBUS instrumented power supplies, provides real-time power consumption data in single-time increments or as a time series, and reports through the IPMI interface.
  • Platform power capping: Node Manager sets platform power to a targeted power budget while maintaining maximum performance for the given power level. It receives a power policy from an external management console through the IPMI interface and maintains power at the targeted level by dynamically adjusting CPU P-states.
  • Power threshold alerting: Node Manager monitors platform power against the targeted power budget. When the target power budget cannot be maintained, it sends out alerts to the management console.

Neusoft Information Service Portal System on Neusoft Aclome

Neusoft is the leading IT Solutions and Services provider in China. Based on IT solution experiences and increasing customer cloud solution demands, with Intel support, Neusoft developed Neusoft Aclome as a complete cloud computing solution. It is the enterprise IT infrastructure, enabling customers to enjoy the benefits of cloud computing without excessive building and validation effort. Neusoft Aclome is an open cloud-based solution management platform for traditional enterprise data centers and Internet data centers (IDCs).

The application used here is Neusoft Information Service Portal System, which is an information portal customized for the State Commission Office for Public Sector Reform (China) (SCOPSR) to provide general Information publishing, query and aggregation for the SCOPSR office. It is a typical 3-tier web application. They use a workload emulator to simulate the workload pressure brought by end user requests to calculate the throughput and response time of the system in a scaled production environment. The main request type is report generation, which is considered a heavy workload request. The performance is measured by the finished requests per second while maintaining <10 second response time for 90% of those requests.

Using Intel Node Manager for Neusoft Aclome Cloud

Neusoft Aclome provides high-level power management features to meet end users’ needs for energy efficiency. The goal of power management is to optimize productivity per watt in order to reduce total cost of ownership. Typical power management requirements include the capability to monitor and cap power in real-time at server, rack, zone, and data center levels. This means the ability to manage aggregated power consumption within a rack, zone, or data center based on available power and cooling resources. Load migration policies based on power consumption and cooling resources bring further benefits. With the help of Intel Node Manager Technology and intelligent policy management, Neusoft Aclome perfectly aligns business solutions with IT resources in an efficient way. Neusoft Aclome not only implements resource on-demand allocation, but also implements power on-demand allocation. Neusoft Aclome can allocate power via real-time monitors and can intelligently allocate power through dynamic power adjustments.


Figure 2: Neusoft Aclome* access power information via Intel® Node Manager

To monitor and control the power consumption of managed servers with Intel Xeon processors, Neusoft Aclome contains a module to perform the communication with Intel Node Manager onboard via standard Ethernet connections based on IPMI protocol. The BMC of the managed server is accessed via a separate IP address from the server itself. Intel Node Manager exposes its management API via the IPMI service the BMC provides through the IP. For each managed server with an Intel Xeon processor, the administrator needs to set the power management IP address correctly, and then the power consumption of the server can be achieved and controlled.

Neusoft Aclome Cloud Power Consumption using Intel Node Manager

The following test is performed to validate the benefits brought by the power management features of Neusoft Aclome based on Intel Node Manager. The test is focused on the web layer of the tested SCOPSR solution.


Figure 3: Topology

Hardware

Servers E5-2680 2.70GHZ x 3, Emulator PC x N
BIOS Settings HT Enabled/NUMA Enabled/Virtualization Enabled
Memory Configuration 8 x 4G DDR3-1333
Disk Configuration 1 SATA Hard Disk
DISK IO is not issue, average disk queue length < 1
Network Configuration Onboard NIC used to connect workload emulator and DB

Software

Workload Neusoft Information Service Portal System for SCOPSR on Neusoft Aclome
Hypervisor Release RedHat* RHEV3.0
Number of VMs Used 8
Number of VCPU per VM 4
Memory per VM 3 G
OS Release RHEL 6.1
Application Release Version 1.0
Database Release Oracle 11.2.0.0.0

Testing Methodology

The test is focused on the web layer of the tested SCOPSR solution. The web layer consists of 8 web server virtual machines to form a cluster, all of which reside on a single server with an Intel Xeon E5-2680 processor and Redhat* RHEV 3.0 hypervisor installed.

The workload has two main metrics: the primary performance metric is Requests per Second, measuring the basic performance and the second is Response Time, measuring Quality of Service (QoS). A normal running workload should have the response time S < 10s to maintain the QoS.

  1. Workload running without power control
  2. Workload running with power control

To correctly compare the power efficiency, we will try to maintain the same Request per Second in the two tests.

For both tests, the workload given is not fully stressed. Instead, an average pressure is given to emulate a production environment. For the test with power control, Neusoft Aclome will first try to set a relatively loose power value, and the value is then specified from a larger value to smaller gradually until the response time meets the threshold. With this method, we are able to calculate the maximum power Node Manager can help save without impacting the required QoS.

Test Results

Test Round Without Power Control With Power Control
Primary Perf Metric Requests per Second Requests per Second
Primary Perf 351.3 rps 349.9 rps
Secondary Perf Metric Response Time (should be <=10 s) Response Time (should be <=10 s)
Secondary Perf 7.3 s 9.8 s
CPU% of web servers 62% 61%
Power 212 w 184 w

The tests showed that by adding a customized power policy via Neusoft Aclome based on Node Manager, a 13% power saving can be achieved with less than 0.5% degradation in Primary Performance.

Conclusion

Based on this case study it is clear that Intel Node Manager Technology and intelligent policy management help data center managers reduce the power consumption of their servers without having a significant impact on service performance. The IPMI-based interface provided by Intel Node Manager helps end users integrate this technology into their application environment very easily.

Additional Information

  1. Intel® Intelligent Power Node Manager: http://www.intel.com/technology/intelligentpower/index.htm
  2. Node Manager programmer’s reference kit: /en-us/articles/intel-node-manager-programmers-reference-kit
  3. Neusoft: http://www.neusoft.com/

About the Authors

Xiwei Huang is a software engineer in the Software & Service Group at Intel Corporation. Xiwei holds a Master’s degree in Instrument Science from Tsinghua University, Beijing.

Sree Syamalakumari is a software engineer in the Software & Service Group at Intel Corporation. Sree holds a Master's degree in Computer Engineering from Wright State University, Dayton, Ohio.

Notices

INFORMATION IN THIS DOCUMENT IS PROVIDED IN CONNECTION WITH INTEL PRODUCTS. NO LICENSE, EXPRESS OR IMPLIED, BY ESTOPPEL OR OTHERWISE, TO ANY INTELLECTUAL PROPERTY RIGHTS IS GRANTED BY THIS DOCUMENT. EXCEPT AS PROVIDED IN INTEL'S TERMS AND CONDITIONS OF SALE FOR SUCH PRODUCTS, INTEL ASSUMES NO LIABILITY WHATSOEVER AND INTEL DISCLAIMS ANY EXPRESS OR IMPLIED WARRANTY, RELATING TO SALE AND/OR USE OF INTEL PRODUCTS INCLUDING LIABILITY OR WARRANTIES RELATING TO FITNESS FOR A PARTICULAR PURPOSE, MERCHANTABILITY, OR INFRINGEMENT OF ANY PATENT, COPYRIGHT OR OTHER INTELLECTUAL PROPERTY RIGHT.

UNLESS OTHERWISE AGREED IN WRITING BY INTEL, THE INTEL PRODUCTS ARE NOT DESIGNED NOR INTENDED FOR ANY APPLICATION IN WHICH THE FAILURE OF THE INTEL PRODUCT COULD CREATE A SITUATION WHERE PERSONAL INJURY OR DEATH MAY OCCUR.

Intel may make changes to specifications and product descriptions at any time, without notice. Designers must not rely on the absence or characteristics of any features or instructions marked "reserved" or "undefined." Intel reserves these for future definition and shall have no responsibility whatsoever for conflicts or incompatibilities arising from future changes to them. The information here is subject to change without notice. Do not finalize a design with this information.

The products described in this document may contain design defects or errors known as errata which may cause the product to deviate from published specifications. Current characterized errata are available on request.

Contact your local Intel sales office or your distributor to obtain the latest specifications and before placing your product order.

Copies of documents which have an order number and are referenced in this document, or other Intel literature, may be obtained by calling 1-800-548-4725, or go to: http://www.intel.com/design/literature.htm

Software and workloads used in performance tests may have been optimized for performance only on Intel microprocessors. Performance tests, such as SYSmark and MobileMark, are measured using specific computer systems, components, software, operations, and functions. Any change to any of those factors may cause the results to vary. You should consult other information and performance tests to assist you in fully evaluating your contemplated purchases, including the performance of that product when combined with other products.

Any software source code reprinted in this document is furnished under a software license and may only be used or copied in accordance with the terms of that license.

Intel, Xeon, and the Intel logo are trademarks of Intel Corporation in the US and/or other countries.

Copyright © 2012 Intel Corporation. All rights reserved.

*Other names and brands may be claimed as the property of others.

Para obter mais informações sobre otimizações de compiladores, consulte Aviso sobre otimizações.