Browse Prior Art Database

Power Savings Using Predictive Usage Methods

IP.com Disclosure Number: IPCOM000123244D
Original Publication Date: 1998-Jul-01
Included in the Prior Art Database: 2005-Apr-04
Document File: 6 page(s) / 185K

Publishing Venue

IBM

Related People

Smith, GJ: AUTHOR [+2]

Abstract

Energy conservation is critical in computer installations. Large numbers of hard disk drives (HDD) are being used at an increasing rate as demand for information grows. This is particularly true for image and video data. The timely invocation of power saving modes and HDD data backup is very important as energy costs are an overhead expense item. The problem is how to tune the power-down backup cycle to minimize impact to an enterprise and its customers.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 37% of the total text.

Power Savings Using Predictive Usage Methods

   Energy conservation is critical in computer
installations.  Large numbers of hard disk drives (HDD) are being
used at an increasing rate as demand for information grows.  This is
particularly true for image and video data.  The timely invocation of
power saving modes and HDD data backup is very important as energy
costs are an overhead expense item.  The problem is how to tune the
power-down backup cycle to minimize impact to an enterprise and its
customers.

   One solution to this problem is to use predictive methods
to forecast appropriate timing for power-down and backup of computer
systems, particularly HDD.  While prediction and forecasting are
certainly not new in the computer systems, the interplay between
power saving modes and backup has not been adequately addressed for
most enterprises.

   Growth of HDD use into larger sub-systems provides and
opportunity to divide storage into pools.  Size and content of each
storage pool is determined by availability requirements.  Typically,
systems that use storage pools today structure the pools by capacity.
By adding availability requirements, it becomes possible and
advantageous to invoke power saving modes during idle periods.  An
example illustrating the storage pool concept could be e-mail, which
requires high availability during the business day and much lower
availability requirements during non-business hours.  Storage pools
not in use can invoke power saving modes.  As usage is required,
storage pools are returned to normal operation.

   The predictor hardware and software determines, based on
availability of storage pools, the probable window of opportunity
for data backup and power saving modes.  Invoking power saving modes
and data back-up become customized to each enterprise, making for
efficient storage and energy use.

   A power saving mode is defined as any action that reduces
energy.  HDD's have a power saving mode available: the small
computer system interface (SCSI) Start/Stop Unit command.  Potential
power savings using the Start/Stop Unit command is approximately 6
watts per HDD.  Removing all power from an HDD has a potential
savings of up to 15 watts.  A storage pool is defined as a portion of
HDD storage on a system.  See Figure 1.  N storage pools can be
defined.  Six hypothetical application will illustrate power savings
using predictive usage methods (see Figure 2).  The steps used to
optimize power savings are listed below:
  1.  Build or modify storage pools by applications and/or
      functions.
      a.  Divide available system storage into more than one
          storage pool.  Each storage pools will be structured
          to an HDD boundary.  All data stored on an HDD will
          be part of the same storage pool.  Selection criteria
          for what data is stored in a storage pool include
          periods of time wh...