Method for restarting a long-running, fault-tolerant operation in a transaction-oriented data base system without burdening the system log

Claim 1

A method for restarting a long-running, fault tolerant operation in a transaction-oriented data base system with minimum recourse to any event recording portion of such system such as a log, said system comprising a store having a data base, said data base including a log, a sequential data set, and a data base formatted object; a computer managing transaction, data base, and recovery processing; and means responsive to the computer for accessing the store, said data base formatted object occupying a specified range of addresses within the store, all transaction-oriented changes to the data base being written to the log in support of recovery in the event of interruption, each transaction utilizing BEGIN, COMMIT, or ROLLBACK primitives to bound said transaction, REDOs ensuring transaction return to the most recent COMMIT point, while UNDOs ensuring return to the transaction BEGIN point, comprising the computer-implemented steps of: (a) reiteratively moving a predetermined quantity of the sequential data set from a first position in the store and appending said predetermined quantity to the end of the specified range of addresses of said data base formatted object at a second position within said store, recording in the log concurrent with such reiterative movement only of the current position within the data set and the newly established end position of the data base formatted object, and establishing another end position within the data base formatted object, said step continuing either until the data set becomes exhausted of a transaction in progress becomes interrupted (FIGS. 3, 5); and (b) in the event of transaction interruption, restarting the long-running operation from the last COMMIT point by referencing only the most recently recorded position within the data set and the most recently recorded end position of the data base formatted object, performing REDO and UNDO processing on that portion of the moved data set of step (a) measured from the referenced recorded positions, and repeating step (a) processing (FIGS. 4, 6).

Claim 2

A method for restarting a long-running, fault tolerant operation in a transaction-oriented data base system with minimum recourse to any event recording portion of such system such as a log, said system comprising a store having a data base, said data base including a log, a sequential data set, and a data base formatted object; a computer managing transaction, data base, and recovery processing; and means responsive to the computer for accessing the store, said data base formatted object occupying a specified range of addresses within the store, all transaction-oriented changes to the data base being written to the log in support of recovery in the event of interruption, each transaction utilizing BEGIN, COMMIT, or ROLLBACK primitives to bound said transaction, REDOs ensuring transaction return to the most recent COMMIT point, while UNDOs ensuring return to the transaction BEGIN point, comprising the computer-implemented steps of: (a) reiteratively moving a predetermined quantity of the sequential data set from a first position in the store and appending said predetermined quantity to the end of the specified range of addresses of said data base formatted object at a second position within said store, recording in the log concurrent with such reiterative movement only of the current position within the data set and the newly established end position of the data base formatted object, and establishing another end position within the data base formatted object, said step continuing either until the data set becomes exhausted of a transaction in progress becomes interrupted (FIGS. 3, 5); and (b) in the event of transaction interruption, restarting the long-running operation from the last COMMIT point by referencing only the most recently recorded position within the data set and the most recently recorded end position of the data base formatted object, performing REDO and UNDO processing on that portion of the moved data set of step (a) measured from the referenced recorded positions, and repeating step (a) processing (FIGS. 4, 6).

Claim 5

A method for executing a restartable LOAD operation in a transaction-oriented relational data base system with minimum recourse to any event recording portion of such system such as a log. said system comprising a store having a data base, said data base including a log, a sequential data set, and a relational tablespace; a computer managing transaction, data base, and recovery processing; and means responsive to the computer for accessing the store, said relational tablespace occupying a specified range of addresses within the store, all transaction-oriented changes to the data base being written to the log in support of recovery in the event of interruption, each transaction utilizing BEGIN, COMMIT, or ROLLBACK primitives to bound said transaction, REDOs ensuring transaction return to the most recent COMMIT point, while UNDOs ensuring return to the transaction BEGIN point, comprising the computer-implemented steps of: (a) reiteratively moving a predetermined quantity of the sequential data set from a first position in the store and appending said predetermined quantity to the end of the specified range of addresses of said relational tablespace at a second position within said store in a series of nonoverlapping suboperations, recording in the log concurrent with such reiterative movement only of the current position within the data set and the newly established end position of the relational tablespace, and establishing another end position within the relational tablespace, said step continuing either until the data set becomes exhausted or a transaction in progress becomes interrupted (FIGS. 3, 5); and (b) in the event of transaction interruption, restarting the long-running operation from the last COMMIT point by referencing only the most recently recorded position within the data set and the most recently recorded end position of the relational tablespace, performing REDO and UNDO processing on that portion of the moved data set of step (a) measured from the referenced recorded positions, and repeating step (a) processing (FIGS. 4, 6).

Claim 6

A method for executing a restartable LOAD operation in a transaction-oriented relational data base system with minimum recourse to any event recording portion of such system such as a log. said system comprising a store having a data base, said data base including a log, a sequential data set, and a relational tablespace; a computer managing transaction, data base, and recovery processing; and means responsive to the computer for accessing the store, said relational tablespace occupying a specified range of addresses within the store, all transaction-oriented changes to the data base being written to the log in support of recovery in the event of interruption, each transaction utilizing BEGIN, COMMIT, or ROLLBACK primitives to bound said transaction, REDOs ensuring transaction return to the most recent COMMIT point, while UNDOs ensuring return to the transaction BEGIN point, comprising the computer-implemented steps of: (a) reiteratively moving a predetermined quantity of the sequential data set from a first position in the store and appending said predetermined quantity to the end of the specified range of addresses of said relational tablespace at a second position within said store in a series of nonoverlapping suboperations, recording in the log concurrent with such reiterative movement only of the current position within the data set and the newly established end position of the relational tablespace, and establishing another end position within the relational tablespace, said step continuing either until the data set becomes exhausted or a transaction in progress becomes interrupted (FIGS. 3, 5); and (b) in the event of transaction interruption, restarting the long-running operation from the last COMMIT point by referencing only the most recently recorded position within the data set and the most recently recorded end position of the relational tablespace, performing REDO and UNDO processing on that portion of the moved data set of step (a) measured from the referenced recorded positions, and repeating step (a) processing (FIGS. 4, 6).