MEYHOD AND APPARATUS FOR INTER-NODE DEADLOCK AVOIDANCE ON A PARALLEL PROCESSING SYSTEM

    公开(公告)号:CA2220612A1

    公开(公告)日:1999-05-03

    申请号:CA2220612

    申请日:1997-11-03

    Applicant: IBM CANADA

    Abstract: The invention provides a method and apparatus including a programming product fo r a parallel data processing system having a plurality of nodes interconnected by communication paths, in which one or more sets of data present at one or more so urce nodes of the system is required to be sent on a plurality of communication paths having communication buffers to selected nodes of the database system for processing, a system of communicating the data over the paths between the nodes to avoid deadlocking the processing system, including: accessing data from the one or more source nodes i n subsets of the data; sending the subsets of data on a plurality of paths to the selected nodes of the processing system; if a path of the paths intended for the sending of one or more of the subsets of data becomes blocked, suspending the sending of subsets o f the data on the path for a predetermined time period; if the blocked path becomes ca pable of accepting additional data before the predetermined time period has expired, resu ming the sending of subsets of data on the path; however, if the predetermined time perio d expires without the path becoming available, placing one or more subsequent subsets of d ata intended for the path into an overflow buffer; when the blocked path again becom es free, sending the subsets of data from the overflow buffer on the path to the intended recipient node for processing.

    DIRECT CALL THREADED CODE
    2.
    发明专利

    公开(公告)号:CA2279222A1

    公开(公告)日:2001-01-30

    申请号:CA2279222

    申请日:1999-07-30

    Applicant: IBM CANADA

    Abstract: A pre-pass and direct call mechanism which replaces the interpreter. The mechanism moves static decisions out of the repeated runtime path, info a pre-pass operation. Advantageously, the mechanism reduces runtime overhead, and improves overall performance of the DBMS during runtime, while maintaining the considerable investment in opcode generation and plan representation, already invested in the implementation of the DBMS.

    MECHANISM FOR INVOCATION OF USER-DEFINED ROUTINES IN A MULTI-THREADED DATABASE ENVIRONMENT

    公开(公告)号:CA2365687A1

    公开(公告)日:2003-06-19

    申请号:CA2365687

    申请日:2001-12-19

    Applicant: IBM CANADA

    Abstract: A database management system provides for the execution of invokable user- defined routines. Applications specify user-defined routines as being threadsafe or not threadsafe and specify the language of user-defined routines. Execution of routines will be carried out dependent on the threadsafe characteristic of the routines and on the language of the routines. Routines are assigned to processes based on their language to have any given process execute routines of the same language. Routines that are specified to be threadsafe are executed in threa ds in a potentially multi- threaded process. Each routine that is specified to be not threadsafe is executed in a fenced-mode processes in which no other routine is being executed.

    MANAGEMENT OF USER-DEFINED ROUTINE LIBRARIES IN DATABASE ENVIRONMENTS

    公开(公告)号:CA2366338A1

    公开(公告)日:2003-06-21

    申请号:CA2366338

    申请日:2001-12-21

    Applicant: IBM CANADA

    Abstract: Library management for libraries of user-defined routines in a database system. Libraries are stored in a long-term storage device and loaded into working memory in t he database system when routines in the libraries are to be executed. An unload thread is defin ed to carry out the process of unloading libraries from the working memory when such libraries are inactive. The unload thread becomes active on a defined timing interval. The unload thread accesses a ha sh table and a most recently used list to determine which of the loaded libraries is inactive an d which of the libraries ought to be unloaded. The hash table maintains information as to which libraries have been accessed in the most recent past defined timing interval. The size of the most recent ly used list can be defined such that all libraries associated with the list are retained in working memory.

    METHOD AND APPARATUS FOR INTER-NODE DEADLOCK AVOIDANCE ON A PARALLEL PROCESSING SYSTEM

    公开(公告)号:CA2220612C

    公开(公告)日:2001-04-24

    申请号:CA2220612

    申请日:1997-11-03

    Applicant: IBM CANADA

    Abstract: The invention provides a method and apparatus including a programming produc t for a parallel data processing system having a plurality of nodes interconnected b y communication paths, in which one or more sets of data present at one or mor e source nodes of the system is required to be sent on a plurality of communication p aths having communication buffers to selected nodes of the database system for processin g, a system of communicating the data over the paths between the nodes to avoid deadlock ing the processing system, including: accessing data from the one or more source nod es in subsets of the data; sending the subsets of data on a plurality of paths to the selected nodes of the processing system; if a path of the paths intended for the send ing of one or more of the subsets of data becomes blocked, suspending the sending of subse ts of the data on the path for a predetermined time period; if the blocked path become s capable of accepting additional data before the predetermined time period has expired, resuming the sending of subsets of data on the path; however, if the predetermined time p eriod expires without the path becoming available, placing one or more subsequent subsets of data intended for the path into an overflow buffer; when the blocked path again b ecomes free, sending the subsets of data fromthe overflow buffer on the path to the inten ded recipient node for processing.

    MANAGEMENT OF USER-DEFINED ROUTINE LIBRARIES IN DATABASE ENVIRONMENTS

    公开(公告)号:CA2366338C

    公开(公告)日:2006-08-08

    申请号:CA2366338

    申请日:2001-12-21

    Applicant: IBM CANADA

    Abstract: Library management for libraries of user-defined routines in a database system. Libraries are stored in a long-term storage device and loaded into working memory in t he database system when routines in the libraries are to be executed. An unload thread is defin ed to carry out the process of unloading libraries from the working memory when such libraries are inactive. The unload thread becomes active on a defined timing interval. The unload thread accesses a ha sh table and a most recently used list to determine which of the loaded libraries is inactive an d which of the libraries ought to be unloaded. The hash table maintains information as to which libraries have been accessed in the most recent past defined timing interval. The size of the most recent ly used list can be defined such that all libraries associated with the list are retained in working memory.

    DETECTION AND PREVENTION OF WRITING CONFLICTS WITHIN NESTED QUERY STATEMENTS

    公开(公告)号:CA2382712A1

    公开(公告)日:2003-10-19

    申请号:CA2382712

    申请日:2002-04-19

    Applicant: IBM CANADA

    Abstract: Conflict checking for writing operations in an SQL database system is carrie d out for nested statements invoked by a parent statement. Each statement has an associated table access data structure defined at compile-time indicating the tables in the database that the statement accesses and the mode of access. At execution-time the table access data structure for ea ch nested statement is accessed and the information is compared with a list of tables maintained fo r the parent statement to determine if there is a read/write or write/write conflict possible if th e nested statement is executed. Where there is no conflict detected, the information from the tabl e access data structure is copied to the list of tables.

    DIRECT CALL THREADED CODE
    8.
    发明专利

    公开(公告)号:CA2279222C

    公开(公告)日:2002-07-16

    申请号:CA2279222

    申请日:1999-07-30

    Applicant: IBM CANADA

    Abstract: A pre-pass and direct call mechanism which replaces the interpreter. The mechanism moves static decisions out of the repeated runtime path, info a pre-pass operation. Advantageously, the mechanism reduces runtime overhead, and improves overall performance of the DBMS during runtime, while maintaining the considerable investment in opcode generation and plan representation, already invested in the implementation of the DBMS.

Patent Agency Ranking