-
公开(公告)号：JPH11272484A
公开(公告)日：1999-10-08
申请号：JP917399
申请日：1999-01-18
Inventor： FREUND THOMAS , HOUSTON IAIN STUART CALDWELL
IPC: G06F15/16 , G06F9/46 , G06F9/52 , G06F15/177
CPC classification number: G06F9/52 , G06F9/465 , Y10S707/99938
Abstract: PROBLEM TO BE SOLVED: To control a server object in parallel by a highly efficient system with the saving of a CPU processing cycle by delaying request dispatch to an execution thread from a buffer holding a request when collision access to server object resource is generated.
SOLUTION: A parallel control unit 34 receives input from a request positioned at the top of a buffer in an objet adapter 32 and decides the meaning of the request. For deciding the meaning of the request which is executed at present on the thread, inputs are received from the execution threads 33a, 33b and 33c. The parallel control unit 34 compares the meaning of the request positioned at the top of the buffer with the meanings of the requests which are executed at present. Only when the dispatch of the request positioned at the top of the buffer does not generate collision access to the server object resource, it is permitted that the request is dispatched to the thread which can be used.
COPYRIGHT: (C)1999,JPOAbstract translation: 要解决的问题：通过高效系统并行地控制服务器对象，通过在产生对服务器对象资源的冲突访问时，从保存请求的缓冲区延迟请求分派到执行线程，从而节省CPU处理周期。 解决方案：并行控制单元34从位于对象适配器32中的缓冲器顶部的请求接收输入，并决定请求的含义。 为了确定当前在线程上执行的请求的含义，从执行线程33a，33b和33c接收输入。 并行控制单元34将位于缓冲器顶部的请求的含义与当前执行的请求的含义进行比较。 只有当位于缓冲区顶部的请求的调度不产生对服务器对象资源的冲突访问时，才允许将请求发送到可以使用的线程。
-
公开(公告)号：JPH11259443A
公开(公告)日：1999-09-24
申请号：JP909999
申请日：1999-01-18
Applicant: IBM
Inventor： FREUND THOMAS , HOUSTON IAIN STUART CALDWELL
IPC: G06F9/54 , G06F9/46 , G06F9/48 , G06F15/177 , G06F15/16
Abstract: PROBLEM TO BE SOLVED: To schedule and dispatch a client request to be executed by a server object by arranging client requests held in a buffer in a priority order based on a priority order decision rule under the consideration of the state of plural executing sleds and each characteristic of the held request by a scheduling means. SOLUTION: An object request broker 31 transfers a request for a specific sever object to an object adapter 32. This server object is provided with plural parallel executing sleds 33a-33c, and the different instances of the sever objects are executed in parallel. A priority order deciding unit 34 decides the priority order of queued requests in the object adapter 32 waiting for the execution by the server object. A request re-arraying unit 35 inspects the priority order assigned to each queued request, and re-arrays the queued requests according to the priority order.
-
公开(公告)号：DE10045437A1
公开(公告)日：2002-04-04
申请号：DE10045437
申请日：2000-09-14
Applicant: IBM
Inventor： BEAVEN JOHN ANTHONY , CHESSELL AMANDA ELISABETH , GRIFFIN CATHERINE , HOUSTON IAIN STUART CALDWELL , MULHOLLAND MARTIN , ROBINSON IAN , VINES DAVID JOHN
Abstract: The device receives an input indicating the result of a user transaction, and an input indicating the result of a sub-transaction. A result for the expanded transaction is determined from the two inputs. The conclusion of one or more sub-transactions is programmably selected based on the latter result, prior to settlement processing of the sub-transactions. The extended transaction is selectively brought to settlement, without settling all sub-transactions. Independent claims are also included for: (a) a method of coordinating an extended transaction (b) a computer program product
-
公开(公告)号：GB2326492B
公开(公告)日：2002-03-20
申请号：GB9712930
申请日：1997-06-20
Applicant: IBM
IPC: G06F9/46
Abstract: A concurrency control mechanism for a client/server computing system analyzes each client request to determine whether it can be given concurrent access to a server's resources, along with other client requests which are presently accessing the server's resources. Each client request triggers the setting up of a lock request mode which contains conflict resolution logic based on the particulars of the corresponding request. When a new request comes in, the new request's newly created lock request mode is compared to the lock request modes of the previous requests which are currently accessing the server's resources and the conflict resolution logic is carried out to determine whether the new request should be also allowed access.
-
公开(公告)号：GB2334116A
公开(公告)日：1999-08-11
申请号：GB9802294
申请日：1998-02-04
Applicant: IBM
Inventor： FREUND THOMAS , HOUSTON IAIN STUART CALDWELL
Abstract: An apparatus for scheduling and dispatching client requests for execution by a server object in a heterogeneous object-oriented client/server computing environment, comprises a request-holding buffer (in 32) having an input connected to a communications channel which channels the client requests to the apparatus, and an output connected to a plurality of parallel execution threads 33a-c; and a scheduling means for distributing client requests stored in the buffer to the plurality of execution threads, the scheduling means placing client requests held in the buffer in priority order based on a priority determining rule which takes into account the state of the plurality of execution threads and the nature of each of the held requests.
-
Patent Agency Ranking
公开(公告)号：GB2326492A
公开(公告)日：1998-12-23
申请号：GB9712930
申请日：1997-06-20
Applicant: IBM
IPC: G06F9/46
Abstract: A concurrency control mechanism for a client/server computing system analyzes each client request to determine whether it can be given concurrent access to a server's resources, along with other client requests which are presently accessing the server's resources. Each client request triggers the setting up of a lock request mode which contains conflict resolution logic based on the particulars of the corresponding request. When a new request comes in, the new request's newly created lock request mode is compared to the lock request modes of the previous requests which are currently accessing the server's resources and the conflict resolution logic is carried out to determine whether the new request should be also allowed access.
-
公开(公告)号：DE69734432D1
公开(公告)日：2005-12-01
申请号：DE69734432
申请日：1997-12-11
Applicant: IBM
Abstract: An apparatus for dispatching client requests within a server computer system, wherein the client requests are issued from at least one client computer system on a different platform from said server computer system and are passed through a communications channel to said server computer system, the apparatus comprising: a buffer having an input connected to said communications channel and an output; a plurality of parallel execution threads connected to said output of said buffer; and a scheduling means for distributing client requests stored in said buffer to said plurality of execution threads in a manner such that related client requests are sent to the same execution thread.
-
公开(公告)号：DE69530252T2
公开(公告)日：2004-01-29
申请号：DE69530252
申请日：1995-08-22
Applicant: IBM
Inventor： COBB EDWARD ELLIS , HOLDSWORTH SIMON ANTONY JAMES , HOUSTON IAIN STUART CALDWELL , SMITH STANLEY ALAN
Abstract: A system and method for reducing message traffic during two phase commitment in a distributed transaction processing system. Transactional objects manage transactions in a distributed system. A coordinator object is responsible for ensuring data consistency by enforcing atomic transaction changes. Any change to one or more resources by a transaction is either made at all or none of the resources. Distributed transactions use interposition to create subordinate coordinator for domains other than the primary domain. Each subordinate coordinator controls transaction commitment of all resources within that domain via a single message sequence per coordinator. Dynamic registration of subordinate coordinator resources ensures that the subordinate coordinator is not added to the commit tree of the root coordinator unless it has recoverable resources. Dynamic registration minimizes the size of the commit tree thereby minimizing the number of messages transmitted during commitment processing.
-
公开(公告)号：DE10045408A1
公开(公告)日：2001-05-31
申请号：DE10045408
申请日：2000-09-14
Applicant: IBM
Inventor： BEAVEN JOHN ANTHONY , CHESSELL AMANDA ELIZABETH , GRIFFIN CATHERINE , HOUSTON IAIN STUART CALDWELL , MULHOLLAND MARTIN , ROBINSON IAN , VINES DAVID JOHN
Abstract: A mapping processor (106) reads imaged values (107) from a table, corresponding to heterogeneous user input components (102,103) for the commitment or reset request. A result processor (108) is programmed to process the input imaged values using different Boolean expressions and to output the result (109) to the users. Independent claims are also included for the following: (a) Batch processing method; (b) Batch processing program stored in memory medium
-
公开(公告)号：GB2320594A
公开(公告)日：1998-06-24
申请号：GB9626602
申请日：1996-12-20
Applicant: IBM
Abstract: In a distributed system, processing requests are sent from client systems to a server system. The server has plural parallel execution threads 22a-22n for servicing the requests. It is desirable that an execution thread 22a-22n handles related client processing requests (e.g. processing requests that are part of the same transaction) and so the object adapter 23 in the server which receives the client processing requests sends related client processing requests to the queuing buffer (e.g. 21a) of the same execution thread (e.g. 22a).
-
-
-
-
-
-
-
-
-
Search Results
15
records
(took 0.018 seconds)
