Data processing: measuring – calibrating – or testing – Measurement system in a specific environment – Electrical signal parameter measurement system
Reexamination Certificate
2001-12-13
2004-10-12
Hoff, Marc S. (Department: 2857)
Data processing: measuring, calibrating, or testing
Measurement system in a specific environment
Electrical signal parameter measurement system
C702S062000
Reexamination Certificate
active
06804616
ABSTRACT:
TECHNICAL FIELD
This disclosure relates generally to power consumption of electronic devices, and in particular, but not exclusively, relates to a method of determining the power consumption of a computer system.
BACKGROUND
Data centers are facilities designed to house large amounts of electronic equipment—including devices such as computers, servers and routers—in a central location. Data centers are vital nodes in information superhighway. Data centers are designed to not only provide all the servers, storage systems and networking infrastructure required to support incoming data requests, but to also provide services such as: authentication, security, and RAS (reliability availability serviceability) management. Most often, data centers are large rooms that are closed off and heavily air-conditioned to maintain a suitable operating temperature for the equipment housed there. The rooms typically have racks where the data communication and processing equipment is installed. While there are many benefits to having all the equipment in one place—such as easier access and maintenance—data centers are costly to build and operate. Because of the high cost involved, space in a data center is at a premium, and data center facility managers try to squeeze as much equipment as possible into every data center. To do so, however, data center managers must have an accurate way of assessing the power consumption and cooling requirement of each component housed in the data center. For a data center, power and cooling are closely related, since a large portion of the electrical power consumed by a device usually ends up converted into thermal energy.
A power specification for a computer system, P
MAX,
usually defines maximum power capability of the power supply in the system. In the process of determining a value of P
MAX
, system designers usually consider the worst-case configuration of a system. Thus, P
MAX
for a system represents power consumption when the system is fully populated with hardware. The determination of P
MAX
also assumes that the system is configured with the most power hungry components capable of being used in that configuration, and that the system is running software that causes it to consume maximum power. For example, a server system may be designed to support four processors that may run at between 1.5 GHz and 2.2 GHz, 12 slots of memory, 8 slots for input/output (I/O) adapters and 5 bays for hard drives. The P
MAX
for such a system assumes the system is populated with all four 2.2 GHz (max power) processors, all twelve memory slots, all eight I/O slots populated with power hungry I/O adapters, and the system is using five hard drives.
The present methods for determining P
MAX
suffer from various disadvantages. First, most systems are populated with less hardware components than their capacity. Second, in the vast majority of cases not all the components used are the most power hungry. For example, many systems may be using processors slower than 2.2 GHz. Slower processors usually consume less power. On average, most of the systems could be consuming power far lower than P
MAX
.
Data center facility managers mistakenly interpret P
MAX
to be the actual power consumption of the system, rather than the maximum power capability of the power supply. As a result, data center managers currently use P
MAX
as a guideline to decide the amount of power and cooling they should provide for each rack, and for the data center as a whole. For systems that consume less power that P
MAX
, power delivery and cooling solution based upon P
MAX
may be over-designed and more expensive than necessary. Moreover, the philosophy of using P
MAX
generates demand on infrastructure, much higher than what would be actually needed. Data centers that use thousands of such systems, this could result in much large demand and expense on infrastructure.
REFERENCES:
patent: 5983357 (1999-11-01), Sun
patent: 6301674 (2001-10-01), Saito et al.
patent: 6396169 (2002-05-01), Voegeli et al.
patent: 6504266 (2003-01-01), Ervin
patent: 6512682 (2003-01-01), Cohen et al.
patent: 6594771 (2003-07-01), Koerber et al.
patent: 6625736 (2003-09-01), Berthaud et al.
patent: 2002/0066045 (2002-05-01), Layton et al.
patent: 2003/0005339 (2003-01-01), Cohen et al.
patent: 2003/0056125 (2003-03-01), O'Conner et al.
McPherson, An Introduction to Electrical Machines and Transformers, 1981, John Wiley & Sons, Inc. ISBN 0-471-05586-7, pp23-28 and 222-224.
Blakely , Sokoloff, Taylor & Zafman LLP
Hoff Marc S.
Intel Corporation
Miller Craig Steven
LandOfFree
Method to estimate power and cooling requirements of systems does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method to estimate power and cooling requirements of systems, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method to estimate power and cooling requirements of systems will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3293516