Browse Prior Art Database

Storage Management Mechanism for Managing Snapshot Pages

IP.com Disclosure Number: IPCOM000109665D
Original Publication Date: 1992-Sep-01
Included in the Prior Art Database: 2005-Mar-24
Document File: 4 page(s) / 154K

Publishing Venue

IBM

Related People

Heyrman, PJ: AUTHOR [+4]

Abstract

Disclosed is a storage management method used to efficiently maintain a point in time (or "snapshot") image of data so that the snapshot image can be saved to backup media while at the same time system users are allowed to change the "live" image (hereafter referred to as a "save while active" operation). Like other snapshot algorithms, this method only copies data that is being changed in order to reduce the temporary storage required to maintain the snapshot image. However, this method further reduces the temporary storage needed to maintain the snapshot image if the data being saved does not actually change during the save, while at the same time still guarantees that the integrity of the snapshot image is always maintained.

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

Storage Management Mechanism for Managing Snapshot Pages

       Disclosed is a storage management method used to
efficiently maintain a point in time (or "snapshot") image of data so
that the snapshot image can be saved to backup media while at the
same time system users are allowed to change the "live" image
(hereafter referred to as a "save while active" operation).  Like
other snapshot algorithms, this method only copies data that is being
changed in order to reduce the temporary storage required to maintain
the snapshot image.  However, this method further reduces the
temporary storage needed to maintain the snapshot image if the data
being saved does not actually change during the save, while at the
same time still guarantees that the integrity of the snapshot image
is always maintained.

      The following terms are used throughout this article:
o    Page - the minimum amount of virtual storage that is transferred
between auxiliary and main memory (512 bytes)
o    Segment - a virtual address range consisting of 1 to 32768
virtual pages
o    Object - a logical group of 1 or more related segments

      Two basic data structures are used to maintain the snapshot
view of object data during a save while active (SWA) operation:
o    A directory that contains an entry for each address range
(segment) in the object being saved.  Each directory entry will map
from 1 to 32768 pages based on the amount of data in the segment that
is going to be saved.
o    A control block for each segment that has changed while a
snapshot view of the segment is being maintained.

      The main components of this control block are:
      -    A bit map that contains a bit for every page being tracked
in this segment.  A bit value of '0'B indicates the page has not been
changed; a value of '1'B indicates that the page has been changed and
a snapshot copy of the page exists.
      -    A disk address table which contains the disk unit and
sector address of the snapshot copy version of each page being
tracked in the segment.  The disk address table entry for a page
contains a value only if the bit map setting for the page is '1'B.
      The address of this control block (once it becomes created) is
stored in the directory entry for the segment.

      To show how these data structures are used to maintain a
snapshot view of object data, we can follow an object through
different stages of a SWA operation and see how these operations
interact with the data structures:
o    Snapshot start - when an object reaches a state that is valid
for saving, storage management is allowed to "capture" this state.
The state is captured by adding an entry into the SWA directory for
each segment which cont...