Browse Prior Art Database

Object-Oriented Framework for Redundant Persistent Configuration Data

IP.com Disclosure Number: IPCOM000118574D
Original Publication Date: 1997-Mar-01
Included in the Prior Art Database: 2005-Apr-01
Document File: 4 page(s) / 122K

Publishing Venue

IBM

Related People

Martin, DR: AUTHOR [+2]

Abstract

Disclosed is a program that ensures persistence and accuracy of system configuration data. The Redundant Persistent Data (RPD) program keeps multiple copies of the system configuration data on different devices at different locations such that multiple failures are tolerated. RPD maintains consistency and accuracy of the data by keeping a level indicator that is incremented for each update.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 50% of the total text.

Object-Oriented Framework for Redundant Persistent Configuration Data

      Disclosed is a program that ensures persistence and accuracy of
system configuration data.  The Redundant Persistent Data (RPD)
program keeps multiple copies of the system configuration data on
different devices at different locations such that multiple failures
are tolerated.  RPD maintains consistency and accuracy of the data by
keeping a level indicator that is incremented for each update.

      RPD guarantees that data being managed is written to disk,
ensuring persistence through system power cycles and other
terminations.  To guarantee that a single disk or other component
failure will not cause loss of any persistent data, RPD maintains
multiple copies of the persistent data across different disks.  The
Redundant Persistent Data is not written to all configured disks on
the system, since that could be quite a large number; therefore,
updating the RPD on all disks could take quite a bit of time.
Instead, RPD is written to a subset of configured disk units, chosen
to minimize  the possibility that a single hardware failure would
make all copies of  the RPD inaccessible.  This 'RPD selection set'
changes as disk units are  added or removed, over the life of the
system, to optimize the distribution of the RPD copies.

      Consistency of a persistent copy of Redundant Persistent Data
is ensured by writing a sequence number in the disk header of each
sector that contains RPD.  The sequence number is incremented every
time the RPD is written.  This allows the system to detect the most
recent copy of the Redundant Persistent Data (the most recent copy
has the highest sequence number).

      The primary copy of the RPD is written to the disk from which
the system's initial microcode and operating system code is loaded
when the system is started.  This disk unit is called the load source
or the Initial Microcode Program Load (IMPL) device.  It is important
that the system get RPD information, particularly that which
identifies the disk units configured as part of the system, from the
same disk as the microcode the system is using, to guarantee that the
two both belong to the same logical system configuration.  If the
primary copy of the RPD is not readable, the system uses a duplicate
copy of all RPD kept on the load source.  If both copies of the RPD
on the load source cannot be read, then the system uses the backup
RPD copies on other disk units.  In this case, user confirmation is
required to ensure that the system is using the correct backup copy
(disk units moved from another system or disk units that were
formerly but no longer part of the RPD selection set that contains
current RPD information could contain back-level data).  This
intervention can be confusing and error-prone for the user.  RPD
automates the selection process and attempts to find the most recent
and reliable copy of the system configuration data available, but...