Browse Prior Art Database

Deriving Unique Names in Distributed Storage Management Environments

IP.com Disclosure Number: IPCOM000104643D
Original Publication Date: 1993-May-01
Included in the Prior Art Database: 2005-Mar-19
Document File: 4 page(s) / 153K

Publishing Venue

IBM

Related People

Bilan, TN: AUTHOR [+4]

Abstract

Network file serving mechanisms present shared file systems as an extension to each client node's local file space. A single file or directory in a served file system may be accessed under a number of differing names from different client nodes in the network. Proper implementation of System Managed Storage (SMS), however, requires that each file and directory object be uniquely identified, so that automated storage management policies can be applied consistently. A method is needed for deriving a unique name for the object, independently of the name that is used to access the object from various nodes in the network.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 40% of the total text.

Deriving Unique Names in Distributed Storage Management Environments

      Network file serving mechanisms present shared file systems as
an extension to each client node's local file space.  A single file
or directory in a served file system may be accessed under a number
of differing names from different client nodes in the network.
Proper implementation of System Managed Storage (SMS), however,
requires that each file and directory object be uniquely identified,
so that automated storage management policies can be applied
consistently.  A method is needed for deriving a unique name for the
object, independently of the name that is used to access the object
from various nodes in the network.

      Disclosed is a method for deriving unique names in a network
file serving environment.  The goal is to consistently derive the
same name for a file or directory object regardless of the "local"
name used to identify it on any client node.  The unique name
derivation ensures that SMS policy will be bound to the object in a
consistent manner, and that backup copies of the object can be
located from any client node having proper access.  The unique name
should also be isolated from file serving access profiles or
"mount-point" changes so that backup or archive copies can be
properly identified and associated with their counterparts in the
live file system over time.

      A key to developing distributed applications in a network
environment is the development of transparent access to resources
made available by server nodes.  With file serving, this implies that
file systems served by file servers appear as extensions to local
file systems of client nodes.  The OS/2 LAN Server product, for
example, makes remote file systems appear as locally attached drive
partitions to LAN Server clients (through the NET USE command).
UNIX* Network File System** (NFS) mechanisms use mount command
constructs to enable client nodes to "mount" remote file systems as
extensions (or sub-trees) of a client's single directory tree
structure.  From the client node's perspective, access to files that
reside on remote nodes appears the same as access to files that
reside on drives that are physically attached to the client node.
Just as virtual memory techniques provide the user's application with
the illusion of executing within a much larger physical memory space,
transparent distributed file systems provide the illusion of having
much more storage space than actually exists on a user's machine.

      For purposes of storage management, however, the transparency
of distributed file serving must be resolved to a consistent name for
each file and directory so that the following conditions can be met,
regardless of the network path from which the object is accessed:

1.  The same storage management policy is associated with the object,
    using the derived consistent name for the object.
2.  All backup or archive copies for an object can be loc...