Browse Prior Art Database

Method And System For Ordering Key Components Of An Incomplete Dependency Key

IP.com Disclosure Number: IPCOM000202444D
Publication Date: 2010-Dec-15
Document File: 4 page(s) / 65K

Publishing Venue

The IP.com Prior Art Database

Abstract

A method and system is disclosed for ordering key components of a dependency key which does not have all key components in a data replication environment.

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

Page 01 of 4

Method And System For Ordering Key Components Of An Incomplete Dependency Key

Disclosed is a method and system for ordering key components of a dependency key

which does not have all key components in a data replication environment. The key

components of the dependency key are ordered in a manner that later key components exist only if all prior key components exist. For example, a dependency key may be made up of three components; a database name, a database record key string and a record field name.

Valid dependency keys in this instance may be made up of all three

components, a database name and database record key only or just a database name. The dependency key cannot be made up of the database name and field name only. Due to this restriction, two updates are dependent on each other if the dependency key matches up through the last component of the key with the fewest number of components identified.

To order key components of a dependency key in a data replication environment, dependencies among multiple units of recovery (UOR) are tracked.

A UOR is a

sequence of operations within a unit of work between synchronization points. The UOR is considered dependent upon another UOR if any of its dependency keys are dependent on any of the dependency keys within the other UOR.

Additionally, the

dependencies may update several different resources and therefore have several dependency keys which describe all the resources affected.

The determination of dependencies is performed by comparing each dependency key
in the UOR to all dependency keys of all existing UORs. The comparison is performed by checking for matches of each possible key length with any keys that are only of that length. For example, if the new dependency key has three components, a check is performed to see if any dependency keys with only a single component match the first component of the new dependency key. Thereafter, a check is performed to determine if any dependency keys with only two components match the first two components of the new dependency key. Finally, a check is performed to determine if any dependency keys with three or more components match the three components of the new dependency key.

To check the dependency keys, a set of hash tables (one for each possible level of components) is used to allow for efficient determination of the existences of a prior UOR that has the same dependency key. Each hash table has an entry for the last UOR that had a dependency key of only that length and a chain of entries for any UORs that were processed after that entry but have keys of longer length. The first entry for the last UOR allows determining if any new dependency key is dependent on the first entry. The chain of entries is used to determine which prior UOR a new UOR

with a dependency key of this length onl...