A method of managing unit-of-work identifiers in a distributed transaction processing system

Assignee: International Business Machines Corp

•

Filing: 1991-08-13

•

Granted: —

•

Expires: 2011-08-13

📄 Original PDF 🌐 Google Patents 🔄 Sync Now

Abstract

Abstract The system has a plurality of processing nodes (300) which are logically connected to form a tree, Normally, each node (300) automatically increments its present identifier at the end of processing each transaction to derive the next transaction identifier. However, in response to a failure condition which results in a severed tree, the distributed database (304) that now reside in the first and second subtrees that result from the severance are forced to consistent states. The transaction identifier is incremented in each of the nodes (300) of the first subtree. Transaction work in the first subtree then proceeds with the incremented identifier. The second subtree is dismantled. In response to a nonfailure condition which results in a severed tree, the distributed database in the first and second subtrees that result from the nonfailure severance is committed to a permanent state. The transaction identifier in each of the nodes of the first subtree is incremented. Transaction work in the first subtree then proceeds with the incremented identifier. A new transaction identifier is generated and propagated to all nodes of the second subtree. Transaction work in the second subtree then proceeds with the new identifier.

Inventors

👤 Kathryn Heninger Britton
👤 Andrew Paul Citron
👤 Bruce Gilbert Lindsay

Bibliographic Details

Application Number —

Jurisdiction EP

Data Source google_patents

Last Synchronized 2026-10-30 19:21

CPC Classifications

G; G06; G06F; G06F11/00; G06F11/07; G06F11/14; G06F11/1402; Y; Y10; Y10S; Y10S707/00; Y10S707/99951; Y10S707/99952; Y10S707/99953

Claims Analysis

Total Claims: 3

Claim 1

Independent Claim 114 words

A method of managing unit-of-work identifiers in a chained distributed data transaction system having a plurality of processing nodes logically connected to form a tree, in which, under predetermined conditions, each node automatically increments its present identifier at the end of processing a present transaction to derive the next transaction identifier, comprising the steps of : a) in response to a failure condition resulting in a severed tree, forcing the distributed data in the first and second subtrees resulting from the severance to consistent states, b) incrementing the transaction identifier in each of the nodes of the first subtree, c) proceeding with transaction work in the first subtree, and d) dismantling the second subtree.

Claim 2

Independent Claim 143 words

A method of managing unit-of-work identifiers in a chained distributed data transaction system having a plurality of processing nodes logically connected to form a tree, in which, under predetermined conditions, each node automatically increments its present identifier at the end of processing a present transaction to derive the next transaction identifier, comprising the steps of : a) in response to a nonfailure condition resulting in a severed tree, committing to a permanent state any modifications made to the distributed data in both of the first and second subtrees resulting from the severance, b) incrementing the transaction identifier in each of the nodes of the first subtree, c) proceeding with transaction work in the first subtree, d) generating a new transaction identifier, e) propagating the new identifier to all nodes of the second subtree, and f) proceeding with transaction work in the second subtree.

Claim 3

Independent Claim 152 words

A method of managing unit-of-work identifiers under failure conditions in a chained distributed data transaction processing system having a plurality of processing nodes logically connected to form a transaction tree, in which, under nonfailure conditions, each node automatically increments its present identifier at the end of processing a present transaction to derive the next transaction identifier, comprising the steps of : a) in response to a communication failure between two nodes of the system, backing out all data changes which are part of the present transaction in a first one of the subtrees formed by the communication failure, b) incrementing the transaction identifier in each of the nodes of the first subtree, c) proceeding with transaction work in the first subtree, d) backing out all data changes which are part of the present transaction in a second one of the subtrees formed by the communication failure, and e) dismantling the second subtree.

References Cited (Backward Citations)

No citation records indexed.

USPTO Annuity & Maintenance Windows

Maintenance schedule is calculated for granted US utility patents.

Assigned Categories & Tags

No categories assigned to this patent yet.

Assign Category Tag

Select Category

Notes (Optional)