Browse Prior Art Database

DASD Command Reordering Algorithm That Accounts for Tangential Position

IP.com Disclosure Number: IPCOM000123337D
Original Publication Date: 1998-Sep-01
Included in the Prior Art Database: 2005-Apr-04
Document File: 4 page(s) / 156K

Publishing Venue

IBM

Related People

Anderson, CA: AUTHOR [+5]

Abstract

Disclosed is a Direct Access Storage Device (DASD), or disk drive, command reordering algorithm that reduces average service times. The algorithm does this by minimizing access time of all three spatial access dimensions. It executes faster and can make better choices by utilizing an initial presort that minimizes the access time of only two dimensions. The algorithm also lends itself to self optimization to account for access times, for a given movement, that can vary over time.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

DASD Command Reordering Algorithm That Accounts for Tangential Position

   Disclosed is a Direct Access Storage Device (DASD), or
disk drive, command reordering algorithm that reduces average
service times.  The algorithm does this by minimizing access time of
all three spatial access dimensions.  It executes faster and can make
better choices by utilizing an initial presort that minimizes the
access time of only two dimensions.  The algorithm also lends itself
to self optimization to account for access times, for a given
movement, that can vary over time.

   At some time prior to when a host system starts
requesting a disk drive do reads and writes, parameters used to
determine command beginning and ending spatial positions and access
times between those positions are calculated and saved.

   For example, head switch and cylinder switch formatted
skew times are needed to determine the tangential position of
requested blocks.  This information could be calculated and saved at
manufacturing, format, and/or power up times in order to reduce
overhead during read/write command processing.

   Seek and settle times for all potential accesses need to
be determined also.  This information can be hard-coded (fixed) or be
made changeable so that the data can be updated at a later time with
more accurate information.  Thus if for any reason servo performance
changes from drive to drive, or from time to time for the same drive,
the drives can self-optimize the reordering algorithm dynamically.

   As the commands are received by the drive, the beginning
and ending spatial positions are determined and saved for later
reordering use.  Those algorithms use the values saved at
initialization.  The algorithms themselves are a function of the
format architecture of a particular disk drive.

   Typical current reordering designs do the reordering
as the commands arrive and are enqueued, or placed in the queue.
Those reordering algorithms do not, nor do they need to, take into
account tangential position.  To save processor bandwidth, thus
command overhead, those reordering algorithms can be removed since
another reordering algorithm will eventually execute that does take
all three positions into account.

   But executing a relatively quick reordering algorithm
when commands are enqueued that groups commands together that are
radially close together can make a subsequent re-reordering
algorithm do two things...
  1.  execute faster
  2.  make better choices.

   For example, when re-reordering taking tangential position
into account, it can be advantageous to limit the number of
potential candidates to select from to some number less than the
total number that may be in the command queue...