(19)
(11)EP 2 761 453 B1

(12)EUROPEAN PATENT SPECIFICATION

(45)Mention of the grant of the patent:
20.05.2020 Bulletin 2020/21

(21)Application number: 12772621.4

(22)Date of filing:  24.09.2012
(51)International Patent Classification (IPC): 
G06F 9/54(2006.01)
(86)International application number:
PCT/US2012/056945
(87)International publication number:
WO 2013/048970 (04.04.2013 Gazette  2013/14)

(54)

SYSTEM AND METHOD FOR PREVENTING SINGLE-POINT BOTTLENECK IN A TRANSACTIONAL MIDDLEWARE MACHINE ENVIRONMENT

SYSTEM UND VERFAHREN ZUR VERHINDERUNG VON PUNKTUELLEN ENGPÄSSEN IN EINER TRANSAKTIONALEN MIDDLEWARE-MASCHINENUMGEBUNG

SYSTÈME ET PROCÉDÉ DE PRÉVENTION D'UN GOULET D'ÉTRANGLEMENT LOCALISÉ DANS UN ENVIRONNEMENT DE MACHINES ÉQUIPÉES D'UN LOGICIEL MÉDIATEUR DE TRANSACTION


(84)Designated Contracting States:
AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR

(30)Priority: 29.09.2011 US 201161541054 P
08.03.2012 US 201213415670

(43)Date of publication of application:
06.08.2014 Bulletin 2014/32

(73)Proprietor: ORACLE INTERNATIONAL CORPORATION
Redwood Shores, CA 94065 (US)

(72)Inventors:
  • LITTLE, Todd
    Palatine Illinois 60067 (US)
  • JIN, Yongshun
    Beijing (CN)
  • NIU, Erli
    Beijing (CN)

(74)Representative: D Young & Co LLP 
120 Holborn
London EC1N 2DY
London EC1N 2DY (GB)


(56)References cited: : 
EP-A2- 0 278 316
US-B1- 6 766 358
US-A1- 2005 220 128
  
  • FORIN A ET AL: "High-performance distributed objects over system area networks", PROCEEDINGS OF THE 3RD USENIX WINDOWS NT SYMPOSIUM USENIX ASSOC. BERKELEY, CA, USA, 1999, pages 21-30, XP002690702, ISBN: 1-880446-29-4
  • ISHIZAKI T ET AL: "CrispORB: high performance CORBA for system area network", HIGH PERFORMANCE DISTRIBUTED COMPUTING, 1999. PROCEEDINGS. THE EIGHTH INTERNATIONAL SYMPOSIUM ON REDONDO BEACH, CA, USA 3-6 AUG. 1999, LOS ALAMITOS, CA, USA,IEEE COMPUT. SOC, US, 3 August 1999 (1999-08-03), pages 11-18, XP010358697, DOI: 10.1109/HPDC.1999.805277 ISBN: 978-0-7803-5681-8
  • VALENTE ET AL: "Collocation optimizations in an aspect-oriented middleware system", JOURNAL OF SYSTEMS & SOFTWARE, ELSEVIER NORTH HOLLAND, NEW YORK, NY, US, vol. 80, no. 10, 14 August 2007 (2007-08-14), pages 1659-1666, XP022197272, ISSN: 0164-1212, DOI: 10.1016/J.JSS.2007.01.033
  • DAVID CHERITON: "The V distributed system", COMMUNICATIONS OF THE ACM, ASSOCIATION FOR COMPUTING MACHINERY, INC, UNITED STATES, vol. 31, no. 3, 1 March 1988 (1988-03-01), pages 314-333, XP058349448, ISSN: 0001-0782, DOI: 10.1145/42392.42400
  • Hennadiy Pinus: "Middleware: Past and Present a Comparison", , 1 January 2004 (2004-01-01), XP055575922, Retrieved from the Internet: URL:https://userpages.umbc.edu/~dgorin1/45 1/middleware/middleware.pdf [retrieved on 2019-04-01]
  
Note: Within nine months from the publication of the mention of the grant of the European patent, any person may give notice to the European Patent Office of opposition to the European patent granted. Notice of opposition shall be filed in a written reasoned statement. It shall not be deemed to have been filed until the opposition fee has been paid. (Art. 99(1) European Patent Convention).


Description

Copyright Notice:



[0001] A portion of the disclosure of this patent document contains material which is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure, as it appears in the Patent and Trademark Office patent file or records, but otherwise reserves all copyright rights whatsoever.

Field of Invention:



[0002] The present invention is generally related to computer systems and software such as middleware, and, is particularly related to supporting a transactional middleware machine environment.

Background:



[0003] A transactional middleware system, or a transaction oriented middleware, includes enterprise application servers that can process various transactions within an organization. With the developments in new technologies such as high performance network and multiprocessor computers, there is a need to further improve the performance of the transactional middleware. These are the generally areas that embodiments of the invention are intended to address. The paper entitled "High-performance distributed objects over system area networks" by A. Forin et al. from the 3rd USENIX Windows NT Symposium discloses a distributed object system which uses the Virtual Interface Architecture (VIA) and Microsoft's Distributed Component Object Model (DCOM).

Summary:



[0004] Described herein is a system and method for supporting exchanging messages between a local machine and a remote machine in a transactional middleware machine environment using Remote Direct Memory Access (RDMA) protocol. The transactional middleware machine environment can prevent single-point bottleneck and achieve short latency in a manner like a local message transfer. The transactional middleware machine environment comprises a first message queue and a second message queue. The first message queue is associated with a server in a first transactional machine, which can be accessed using a queue address in the first message queue. The second message queue is associated with a client in a second transactional machine. The client can send a service request message to the server directly using the first message queue, and the server can receive the service request message from the first message queue and send a service response message directly to the client using the second message queue.

Brief Description of the Figures:



[0005] 

Figure 1 shows an illustration of a transactional middleware machine environment that can provide direct message transfer, in accordance with an embodiment of the invention.

Figure 2 illustrates an exemplary flow chart for providing direct message transfer in a transactional middleware machine environment, in accordance with an embodiment of the invention.

Figure 3 shows an illustration of setting up a bypass bridge process feature in a transactional middleware machine environment, in accordance with an embodiment of the invention.

Figure 4 shows a functional configuration in a transactional middleware machine environment, in accordance with an embodiment of the invention.

Figure 5 is a functional block diagram of a transactional server in accordance with an embodiment of the invention.

Figure 6 is a functional block diagram of a client server in accordance with an embodiment of the invention.


Detailed Description:



[0006] Described herein is a system and method for supporting a transactional middleware system, such as Tuxedo, that can take advantage of fast machines with multiple processors, and a high performance network connection. A transactional middleware system can exchange messages between a local machine and a remote machine using Remote Direct Memory Access (RDMA) protocol to achieve short latency in a manner like a local message transfer. The transactional middleware machine environment comprises a first message queue associated with a server in a first transactional machine, wherein the server operates to be accessed using a queue address in the first message queue. The transactional middleware machine environment further comprises a second message queue associated with a client in a second transactional machine. The client operates to send a service request message to the server directly using the first message queue, and the server operates to receive the service request message from the first message queue and send a service response message directly to the client using the second message queue.

[0007] In accordance with an embodiment of the invention, the system comprises a combination of high performance hardware, e.g. 64-bit processor technology, high performance large memory, and redundant InfiniBand and Ethernet networking, together with an application server or middleware environment, such as WebLogic Suite, to provide a complete Java EE application server complex which includes a massively parallel in-memory grid, that can be provisioned quickly, and can scale on demand. In accordance with an embodiment, the system can be deployed as a full, half, or quarter rack, or other configuration, that provides an application server grid, storage area network, and InfiniBand (IB) network. The middleware machine software can provide application server, middleware and other functionality such as, for example, WebLogic server, JRockit or Hotspot JVM, Oracle Linux or Solaris, and Oracle VM. In accordance with an embodiment, the system can include a plurality of compute nodes, IB switch gateway, and storage nodes or units, communicating with one another via an IB network. When implemented as a rack configuration, unused portions of the rack can be left empty or occupied by fillers.

[0008] In accordance with an embodiment of the invention, referred to herein as "Sun Oracle Exalogic" or "Exalogic", the system is an easy-to-deploy solution for hosting middleware or application server software, such as the Oracle Middleware SW suite, or Weblogic. As described herein, in accordance with an embodiment the system is a "grid in a box" that comprises one or more servers, storage units, an IB fabric for storage networking, and all the other components required to host a middleware application. Significant performance can be delivered for all types of middleware applications by leveraging a massively parallel grid architecture using, e.g. Real Application Clusters and Exalogic Open storage. The system delivers improved performance with linear I/O scalability, is simple to use and manage, and delivers mission-critical availability and reliability.

[0009] In accordance with an embodiment of the invention, Tuxedo is a set of software modules that enables the construction, execution, and administration of high performance, distributed business applications and has been used as transactional middleware by a number of multi-tier application development tools. Tuxedo is a middleware platform that can be used to manage distributed transaction processing in distributed computing environments. It is a proven platform for unlocking enterprise legacy applications and extending them to a services oriented architecture, while delivering unlimited scalability and standards-based interoperability.

[0010] In accordance with an embodiment of the invention, a transactional middleware system, such as a Tuxedo system, can take advantage of fast machines with multiple processors, such as an Exalogic middleware machine, and a high performance network connection, such as an Infiniband (IB) network.

[0011] In accordance with an embodiment of the invention, a transactional middleware system can exchange messages between a local machine and a remote machine using Remote Direct Memory Access (RDMA) protocol to achieve short latency in a manner like a local message transfer.

Providing Direct Message Transfer



[0012] In accordance with an embodiment of the invention, a middleware machine environment, such as the Exalogic middleware machine environment, can be a tightly coupled and logically uniform environment for a transactional application, such as a Tuxedo application. There is no single-point bottleneck in transferring messages between machines in the transactional middleware machine environment.

[0013] Figure 1 shows an illustration of a transactional middleware machine environment that can provide direct message transfer, in accordance with an embodiment of the invention. As shown in Figure 1, the transactional middleware machine environment includes a plurality of transactional middleware machines: machine A 101 and machine B 102. Each transactional middleware machine includes an advertized table 112 or 113, for example a Bulletin Board (BB) in Tuxedo. The advertized table can be shared by different machines.

[0014] As shown in Figure 1, client B on machine B 102 can look up the advertized table for a server that provides a service that client B needs. This target server can be either a local server such as server B on machine B 102, or a remote server such as server A on machine A 101.

[0015] If the service is on a local server, client B can send a service request message to the target server and receive a service response message, via one or more System V Inter-process Communication (IPC) queues, such as an IPC queue B 108 on machine B 102.

[0016] On the other hand, if the service is on a remote server such as server A on machine A 101, the client B can use a RDMA messaging queues to exchange message directly even though they are physically distributed among multiple machines. A RDMA queue is a library which can send and receive messages that is based on the RDMA protocol running on a high performance network such as an IB network.

[0017] As shown in Figure 1, client server B 106 can put a service request message directly into a RDMA queue A 110 on machine A 101 associated with the transactional server A 103. Translational server A 103 can get the service request message from the RDMA queue A 110 and processes the service request. Then, transactional server A 103 can put a service response message directly into a RDMA queue B 111 on machine B 102 associated with the client B, so that client B can get the service response.

[0018] Alternatively, as shown in Figure 1, there can be a bridge process A 114 on machine A 101 and a bridge process B 115 on machine B 102. Using these bridge processes, when client B on machine B 102 targets a message to a remote machine A 101, client B can first send the service request message to the System V IPC queue B 108, which is associated with the local bridge process, bridge process B 115. Then, the bridge process B 115 can forward this message to another bridge process, bridge process A 114 on the remote target machine A 101, via network. Finally, the bridge process A 114 at the remote target machine A 101 can send the message to the System V IPC queue A 107 that is associated with the target transactional server A 103. However, this procedure can be time-consuming, and can introduce a relatively long latency. Also, each of the bridge process can be too busy under heavy load and become a single point bottleneck that can impact the throughput of the system as a whole.

[0019] As shown in Figure 1, transactional server A 103 can listen to both a System V IPC queue A 107 and a RDMA queue A 110 at the same time in different threads. A message sender, such as client B, can choose a faster way to send the message to transactional server A 103. In general, the RDMA queue can be used for remote message transferring, and the System V IPC queue can be used for local message transferring since the RDMA queue may be slower than System V IPC queue for local message transferring.

[0020] Figure 2 illustrates an exemplary flow chart for providing direct message transfer in a transactional middleware machine environment, in accordance with an embodiment of the invention. As shown in Figure 2, at step 201, a first message queue is provided to be associated with a transactional server in a first transactional machine. Then, at step 202, the transactional server can publish the first message queue in an advertized table in the transactional middleware machine environment. Finally, at step 203, a client server in a second transactional machine can find the first message queue in the advertized table, and establish a connection with the transactional server in the first transactional machine using the first message queue.

Bypassing the bridge Process in Tuxedo using the RDMA Protocol



[0021] In accordance with an embodiment of the invention, a transactional middleware system can exchange messages between a local machine and a remote machine using a RDMA queue via the RDMA protocol, such as a MSGQ queue library in Tuxedo. The RDMA protocol can be used for transferring data across machines running on a high performance IB network. Using the RDMA protocol, the message sender and receiver can exchange message directly even when they are physically distributed among multiple machines.

[0022] Tuxedo is a transactional middleware with class distributed client-server structure. Each Tuxedo server can publish its MSGQ queue name in a table advertized throughout the domain. A client server can find the queue name of a target server in the advertized table and establish a connection with the target server if it is the first time for the connection. Then, the client server and the target server can establish a connection according to this queue name, and send a message via the established connection.

[0023] Additionally, Tuxedo can use a bridge process in each machine within a domain, to exchange messages across machines. Using a bridge process, every message targeting a remote machine can be first sent to a System V IPC queue associated with a local bridge process. Then, the local bridge process can forward this message to a remote bridge process at remote target machine via the network. Finally, the remote bridge process at the remote target machine can send the message to the target System V IPC queue.

[0024] In accordance with an embodiment of the invention, a RDMA queue library can be implemented with minimum code change, using a series of message transfer APIs that are implemented using a System V style. These message transfer APIs can provide similar functions that the System V IPC queue provides, for example receiving messages from a queue with priority.

[0025] Figure 3 shows an illustration of setting up a bypass bridge process feature in a transactional middleware machine environment, in accordance with an embodiment of the invention. As shown in Figure 3, at step 301, a user 311 can first start a DAEMON process. The DAEMON process 313 can be a separate application from the transactional application such as the Tuxedo application. Also, in the example of Tuxedo, at least one DAEMON process is started for each Tuxedo machine. Then, at steps 302 and 303, the user can configure the application server and the transactional application (Tuxedo) respectively. Finally, at step 304, the user can start the Tuxedo applications in order to use the bypass bridge feature, and at step 305, the user allows the communication for exchanging messages between the processes.

[0026] The following Table 1 shows different options for executing a DAEMON program in Tuxedo.
Table 1
OptionsDescription
-i The Infiniband interface's IP address, the value of which is identical to RDMADAEMONIP.
-p The port that daemon accept request from, the value of which is identical to RDMADAEMONPORT.
-m The shared memory size allocated for daemon, the value of which is greater than the number of queues *(number of IPs connecting using remotename + 1) * qsize
-k The shared memory key to be used by daemon.


[0027] For example, a DAEMON program can be started using the following command: Msgq_daemon -i 192.168.10.4 -p 4040 -m 100000000 -k 9100

[0028] Additionally, different kinds of parameters are configured for Tuxedo applications in order to utilize the message queue. These parameters include at least the parameters of the message queue and the parameters that the processes in the Tuxedo applications use to communicate with the DAEMON process. These parameters can be configured in MACHINES section of a configuration file, such as a Tuxedo UBB file. The following Table 2 shows these environmental variables.
Table 2
UBB ParametersDescription
RDMADAEMONIP The IP address for communicating with DAEMON.
RDMADAEMONPORT The port for communicating with DAEMON.
RDMAQSIZE The maximum size of the message queue, with a default value to be 1Mbytes. This parameter can be configured in MACHINES section or SERVERS section (using server option "-b"). Its value in MACHINES section configures the default value for all queues. Its value in SERVERS section only configures the queue created by the server.
RDMAQENTRIES The maximum entries of the message queue, with a default value to be 1024. This parameter can be configured in MACHINES section or SERVERS section (using server option "-c"). Its value in MACHINES section configures the default value for all queues. Its value in SERVERS section only configures the queue created by the server.


[0029] The following Listing 1 includes a few sections of an exemplary Tuxedo configuration file. *MACHINES slcag1906 LMID=cpu_0 APPDIR="/root/john/mnt/eniu/exalogic/cases/config/testcase" TUXCONFIG="/root/john/mnt/eniu/exalogic/cases/config/testcase/tuxconfig" TUXDIR="/root/ john/mnt/eniu/TUX11gR1PS164rp/LC/bId" UID=507 GID=506 RDMADAEMONIP="192.168.10.4" RDMADAEMONPORT=9800 RDMAQSIZE=1048576 RDMAQENTRIES=1024 *SERVERS simpserv CLOPT="-A -b 1000000 -c 1000 -j 1000000" SRVGRP=GROUP1 SRVID=1 Simpserv1 SRVGRP=GROUP1 SRVID=1 Listing 1

[0030] The following Listing 2 is an example of RESOURCES section in a Tuxedo configuration file. *RESOURCES IPCKEY 213456 #Example: #IPCKEY 123456 DOMAINID simpapp MASTER simple MAXACCESSERS 10 MAXSERVERS 5 MAXSERVICES 10 MODEL SHM LDBAL N OPTIONS NO_XA,EXALOGIC,RDMA Listing 2

[0031] As shown in the above example, EXALOGIC and RDMA can be added into the item OPTIONS in the RESOURCES section of a Tuxedo configuration file. If OPTIONS contains both RDMA and EXALOGIC, the bypass feature is activated and the local bridge processes are bypassed. Otherwise, the bypass feature is turned off. If there is a need to enable the RDMA option, the EXALOGIC option is enabled first. After enable RDMA option in the RESOURCES section. Attribute "TYPE" of MACHINES section may not be set, since by default, any machines in MP mode is an Exalogic machine (with the same type) to support RDMA feature.

[0032] With reference to Figure 4, a system 400 for providing direct message transfer in a transactional middleware machine environment, in accordance with an embodiment of the invention will be described. Figure 4 shows a functional configuration of system 400. System 400 includes a providing unit 410, a publishing unit 420, a memory 430, a message queue finder 440, and a connection establishment unit 450.

[0033] Providing unit is configured to provide a first message queue with associated with a transactional server in a first transactional machine such as machine B 103. Publishing unit 420 is configured to publish, via the transactional server, the first message queue in an advertised table in the transactional middleware machine environment. Memory 430 is configured to store the advertised table. Message queue finder 440 is configured to find the first message queue in the advertised table. Connection establishment unit 450 is configured to establish a connection with the transactional server in the first transactional machine using the first message queue. Message queue finder 440 and connection establishment unit 450 can be realized in a client server.

[0034] In accordance with an embodiment of the invention, Figure 5 and Figure 6 shows a functional block diagram of a transactional servers 500 and a client server 600 configured in accordance with the principles of the invention as described above. The functional blocks of the transactional server and the client server may be implemented by hardware, software, or a combination of hardware and software to carry out the principles of the invention. It is understood by persons of skill in the art that the functional blocks described in Figure 5 and Figure 6 may be combined or separated into sub-blocks to implement the principles of the invention as described above. Therefore, the description herein may support any possible combination or separation or further definition of the functional blocks described herein.

[0035] As shown in Figure 5, a transactional server 500 for use in a first transactional machine in a transactional middleware machine environment includes a publishing unit 510 and an establishing connection unit 520. The publishing unit 510 is configured to publish, a first message queue associated with the transactional server 500 in the first transactional machine, in an advertized table in the transactional middleware machine environment. The establishing connection unit 520 is configured to, in response to a client server in a second transactional machine operating to find the first message queue in the advertized table, establish a connection with the client server in the second transactional machine using the first message queue.

[0036] In some embodiments, the client server is associated with a second message queue in the second transactional machine, and the client server operates to receive messages using the second message queue.

[0037] In some embodiments, the client server operates to send a service request message to the transactional server 500 directly using the first message queue.

[0038] In some embodiments, the transactional server 500 further comprises a first communication unit 530. The first communication unit 530 is configured to receive the service request message from the first message queue, and send a service response message directly to the client server using a second message queue that is associated with the client server in the second transactional machine.

[0039] In some embodiments, there is no single-point bottleneck in transferring messages between machines in the transactional middleware machine environment.

[0040] In some embodiments, the first message queue uses a network protocol for transferring data across machines running on a high performance network.

[0041] In some embodiments, the network protocol is a remote direct memory access (RDMA) protocol for an Infiniband (IB) network.

[0042] In some embodiments, the transactional server 500 further comprises a second communication unit 540. The second communication unit 540 is configured to use a second message queue associated with the transactional server 500 in the first transactional machine for communication with a local client in the first transactional machine.

[0043] In some embodiments, the second message queue is an inter-process communication (IPC) queue that is associated with a local bridge process.

[0044] In some embodiments, the transactional server 400 further comprises a listening unit 550. The listening unit 550 is configured to listen to both the first message queue and the second message queue at same time in different threads, and chooses a faster route to send a message.

[0045] As shown in Figure 6, a client server 600 for use in a second transactional machine in a transactional middleware machine environment includes a finding unit 610 and an establishing connection unit 620. The finding unit 610 is configured to find, a first message queue associated with and published by a transactional server in a first transactional machine, in an advertized table in the transactional middleware machine environment. The establishing connection unit 620 is configured to establish a connection with the transactional server in the first transactional machine using the first message queue.

[0046] In some embodiments, the client server 600 further comprises a receiving unit 630 configured to receive messages using a second message queue associated with the client server in the second transactional machine. In some embodiments, the client server 600 further comprises a sending unit 640 configured to send a service request message to the transactional server directly using the first message queue.

[0047] In some embodiments, the transactional server operates to receive the service request message from the first message queue, and send a service response message directly to the client server 600 using a second message queue that is associated with the client server 600 in the second transactional machine.

[0048] In some embodiments, there is no single-point bottleneck in transferring messages between machines in the transactional middleware machine environment.

[0049] In some embodiments, the first message queue uses a network protocol for transferring data across machines running on a high performance network.

[0050] In some embodiments, the network protocol is a remote direct memory access (RDMA) protocol for an Infiniband (IB) network.

[0051] In some embodiments, the transactional server uses a second message queue associated with the transactional server in the first transactional machine for communication with a local client in the first transactional machine.

[0052] In some embodiments, the second message queue is an inter-process communication (IPC) queue that is associated with a local bridge process.

[0053] In some embodiments, the transactional server listens to both the first message queue and the second message queue at same time in different threads, and chooses a faster route to send a message.

[0054] The present invention may be conveniently implemented using one or more conventional general purpose or specialized digital computer, computing device, machine, or microprocessor, including one or more processors, memory and/or computer readable storage media programmed according to the teachings of the present disclosure. Appropriate software coding can readily be prepared by skilled programmers based on the teachings of the present disclosure, as will be apparent to those skilled in the software art.

[0055] In some embodiments, the present invention includes a computer program product which is a storage medium or computer readable medium (media) having instructions stored thereon/in which can be used to program a computer to perform any of the processes of the present invention. The storage medium can include, but is not limited to, any type of disk including floppy disks, optical discs, DVD, CD-ROMs, microdrive, and magneto-optical disks, ROMs, RAMs, EPROMs, EEPROMs, DRAMs, VRAMs, flash memory devices, magnetic or optical cards, nanosystems (including molecular memory ICs), or any type of media or device suitable for storing instructions and/or data.

[0056] The foregoing description of the present invention has been provided for the purposes of illustration and description. It is not intended to be exhaustive or to limit the invention to the precise forms disclosed. Many modifications and variations will be apparent to the practitioner skilled in the art. The variations may include combination of two or more features disclosed herein. The embodiments were chosen and described in order to best explain the principles of the invention and its practical application, thereby enabling others skilled in the art to understand the invention for various embodiments and with various modifications that are suited to the particular use contemplated. It is intended that the scope of the invention be defined by the following claims and their equivalence.


Claims

1. A system for providing direct message transfer in a transactional middleware machine environment including a first transactional machine (101) and a second transactional machine (102), comprising:

a first message queue (110) associated with a transactional server (103), wherein the first message queue and the transactional server are in the first transactional machine, and wherein the first message queue is a remote direct memory access, RDMA, queue;

an advertized table (112, 113) in the transactional middleware machine environment, wherein the transactional server operates to publish the first message queue in the advertized table;

a client server (106) in the second transactional machine which is operable to find the first message queue in the advertized table and establish a connection with the transactional server in the first transactional machine by putting a service request message directly into the first message queue; and

a second message queue (107) in the first transactional machine and associated with the transactional server;

a third message queue (108) in the second transactional machine and associated with the client server,

wherein the second message queue and third message queue are inter-process communication, IPC, queues, the second message queue is associated with a first local bridge process (114) on the first transactional machine, and the third message queue is associated with a second local bridge process (115) on the second transactional machine,

wherein the client server is operable to send a service request message to the third message queue, the second local bridge process is operable to forward the service request message to the first local bridge process, and the first local bridge process is operable to forward the service request message to the second message queue,

wherein the transactional server uses the second message queue for communication with a local client server (105) in the first transactional machine, and wherein the transactional server listens to both the first message queue and the second message queue at same time in different threads such that the client server is operable to choose to send a service request message to the transactional server using either the first message queue or the second message queue.


 
2. The system of claim 1, wherein the client server in the second transactional machine is associated with a fourth message queue (111) in the second transactional machine, wherein the client server in the second transactional machine operates to receive messages using the fourth message queue in the second transactional machine.
 
3. The system of claim 2, wherein the transactional server operates to receive the service request message from the first message queue, and send a service response message directly to the client server in the second transactional machine using a fourth message queue in the second transactional machine.
 
4. The system of any one of claims 1 to 3, wherein there is no single-point bottleneck in transferring messages between machines in the transactional middleware machine environment.
 
5. The system of any one of claims 1 to 4, wherein the first message queue uses an RDMA protocol for transferring data between the first and second transactional machines over an Infiniband network.
 
6. A method for providing direct message transfer in a transactional middleware machine environment including a first transactional machine (101) and a second transactional machine (102), comprising:

providing a first message queue (110) associated with a transactional server (103), wherein the first message queue and the transactional server are in the first transactional machine, and wherein the first message queue is a remote direct memory access, RDMA, queue;

the transactional server publishing the first message queue in an advertized table in the transactional middleware machine environment;

a client server (106) in the second transactional machine finding the first message queue in the advertized table and establishing a connection with the transactional server in the first transactional machine by putting a service request message directly into the first message queue; and

the transactional server using a second message queue (107) in the first transactional machine and associated with the transactional server for communication with a local client server (105) in the first transactional machine,

a third message queue (108) in the second transactional machine and associated with the client server,

wherein the second message queue and third message queue are inter-process communication, IPC, queues, the second message queue is associated with a first local bridge process (114) on the first transactional machine, and the third message queue is associated with a second local bridge process (115) on the second transactional machine,

wherein the client server sends a service request message to the third message queue, the second local bridge process forwards the service request message to the first local bridge process, and the first local bridge process forwards the service request message to the second message queue,

and wherein the transactional server listens to both the first message queue and the second message queue at same time in different threads such that the client server is operable to choose to send a service request message to the transactional server using either the first message queue or the second message queue.


 
7. The method of claim 6, further comprising associating the client server in the second transactional machine with a fourth message queue in the second transactional machine, wherein the client server in the second transactional machine operates to receive messages using the fourth message queue in the second transactional machine.
 
8. The method of claim 7, further comprising receiving, via the transactional server, the service request message from the first message queue, and sending, via the transactional server, a service response message directly to the client server in the second transactional machine using the third message queue in the second transactional machine
 
9. A computer program comprising program instructions for execution by one or more computer systems to perform the method of any of claims 6 to 8.
 
10. A computer program product comprising the computer program of claim 9 provided on a computer-readable storage medium.
 


Ansprüche

1. System zum Bereitstellen einer direkten Nachrichtenübertragung in einer Transaktions-"Middleware"-Maschinenumgebung, die eine erste Transaktionsmaschine (101) und eine zweite Transaktionsmaschine (102) enthält, das Folgendes umfasst:

eine erste Nachrichtenwarteschlange (110), die einem Transaktions-Server (103) zugeordnet ist, wobei die erste Nachrichtenwarteschlange und der Transaktions-Server in der ersten Transaktionsmaschine sind und wobei die erste Nachrichtenwarteschlange eine Ferndirektspeicherzugriffs-Warteschlange, RDMA-Warteschlange, ist;

eine angezeigte Tabelle (112, 113) in der Transaktions-"Middleware"-Maschinenumgebung, wobei der Transaktions-Server bewirkt, dass die erste Nachrichtenwarteschlange in der angezeigten Tabelle veröffentlicht wird;

einen Client-Server (106) in der zweiten Transaktionsmaschine, der betreibbar ist, die erste Nachrichtenwarteschlange in der angezeigten Tabelle zu finden und eine Verbindung mit dem Transaktions-Server in der ersten Transaktionsmaschine durch Setzen einer Dienstanforderungsnachricht unmittelbar in die erste Nachrichtenwarteschlange einzurichten; und

eine zweite Nachrichtenwarteschlange (107), die sich in der ersten Transaktionsmaschine befindet und dem Transaktions-Server zugeordnet ist;

eine dritte Nachrichtenwarteschlange (108), die in der zweiten Transaktionsmaschine ist und dem Client-Server zugeordnet ist,

wobei die zweite Nachrichtenwarteschlange und die dritte Nachrichtenwarteschlange Zwischenprozesskommunikations-Warteschlangen, IPC-Warteschlangen, sind, wobei die zweite Nachrichtenwarteschlange einem ersten lokalen Brückenprozess (114) auf der ersten Transaktionsmaschine zugeordnet ist und die dritte Nachrichtenwarteschlange einem zweiten lokalen Brückenprozess (115) auf der zweiten Transaktionsmaschine zugeordnet ist,

wobei der Client-Server betreibbar ist, eine Dienstanforderungsnachricht an die dritte Nachrichtenwarteschlange zu senden, der zweite lokale Brückenprozess betreibbar ist, die Dienstanforderungsnachricht an den ersten lokalen Brückenprozess weiterzuleiten und der erste lokale Brückendienst betreibbar ist, die Dienstanforderungsnachricht an die zweite Nachrichtenwarteschlange weiterzuleiten,

wobei der Transaktions-Server die zweite Nachrichtenwarteschlange für Kommunikationen mit einem lokalen Client-Server (105) in der ersten Transaktionsmaschine verwendet und wobei der Transaktions-Server sowohl die erste Nachrichtenwarteschlange als auch die zweite Nachrichtenwarteschlange zur selben Zeit in verschiedenen "Threads" abhört, so dass der Client-Server betreibbar ist, zu wählen, eine Dienstanforderungsnachricht an den Transaktions-Server unter Verwendung entweder der ersten Nachrichtenwarteschlange oder der zweiten Nachrichtenwarteschlange zu senden.


 
2. System nach Anspruch 1, wobei der Client-Server in der zweiten Transaktionsmaschine einer vierten Nachrichtenwarteschlange (111) in der zweiten Transaktionsmaschine zugeordnet ist, wobei der Client-Server in der zweiten Transaktionswarteschlange bewirkt, dass Nachrichten unter Verwendung der vierten Nachrichtenwarteschlange in der zweiten Transaktionsmaschine empfangen werden.
 
3. System nach Anspruch 2, wobei der Transaktions-Server bewirkt, dass die Dienstanforderungsnachricht von der ersten Nachrichtenwarteschlange empfangen wird und eine Dienstantwortnachricht unmittelbar an den Client-Server in der zweiten Transaktionsmaschine unter Verwendung einer vierten Nachrichtenwarteschlange in der zweiten Transaktionsmaschine gesendet wird.
 
4. System nach einem der Ansprüche 1 bis 3, wobei es keinen Einzelpunktengpass beim Übertragen von Nachrichten zwischen Maschinen in der Transaktions-"Middleware"-Maschinenumgebung gibt.
 
5. System nach einem der Ansprüche 1 bis 4, wobei die erste Nachrichtenwarteschlange ein RDMA-Protokoll zum Übertragen von Daten zwischen der ersten und der zweiten Transaktionsmaschine über ein Infiniband-Netz enthält.
 
6. Verfahren zum Bereitstellen einer direkten Nachrichtenübertragung in einer Transaktions-"Middleware"-Maschinenumgebung, die eine erste Transaktionsmaschine (101) und eine zweite Transaktionsmaschine (102) enthält, das Folgendes umfasst:

Bereitstellen einer ersten Nachrichtenwarteschlange (110), die einem Transaktions-Server (103) zugeordnet ist, wobei die erste Nachrichtenwarteschlange und der Transaktions-Server in der ersten Transaktionsmaschine sind und wobei die erste Nachrichtenwarteschlange eine Ferndirektspeicherzugriffs-Warteschlange, RDMA-Warteschlange, ist;

dass der Transaktions-Server die erste Nachrichtenwarteschlange in einer angezeigten Tabelle in der Transaktions-"Middleware"-Maschinenumgebung veröffentlicht;

dass ein Client-Server (106) in der zweiten Transaktionsmaschine die erste Nachrichtenwarteschlange in der angezeigten Tabelle findet und eine Verbindung mit dem Transaktions-Server in der ersten Transaktionsmaschine durch Setzen einer Dienstanforderungsnachricht unmittelbar in die erste Nachrichtenwarteschlange einrichtet; und

dass der Transaktions-Server eine zweite Nachrichtenwarteschlange (107), die in der ersten Transaktionsmaschine ist und dem Transaktions-Server für eine Kommunikation mit einem lokalen Client-Server (105) in der ersten Transaktionsmaschine zugeordnet ist, verwendet,

eine dritte Nachrichtenwarteschlange (108), die in der zweiten Transaktionsmaschine ist und dem Client-Server zugeordnet ist,

wobei die zweite Nachrichtenwarteschlange und die dritte Nachrichtenwarteschlange Zwischenprozesskommunikations-Warteschlangen, IPC-Warteschlangen, sind, wobei die zweite Nachrichtenwarteschlange einem ersten lokalen Brückenprozess (114) auf der ersten Transaktionsmaschine zugeordnet ist und die dritte Nachrichtenwarteschlange einem zweiten lokalen Brückenprozess (115) auf der zweiten Transaktionsmaschine zugeordnet ist,

wobei der Client-Server eine Dienstanforderungsnachricht an die dritte Nachrichtenwarteschlange sendet, der zweite lokale Brückenprozess die Dienstanforderungsnachricht an den ersten lokalen Brückenprozess weiterleitet und der erste lokale Brückendienst die Dienstanforderungsnachricht an die zweite Nachrichtenwarteschlange weiterleitet,

und wobei der Transaktions-Server sowohl die erste Nachrichtenwarteschlange als auch die zweite Nachrichtenwarteschlange zur selben Zeit in verschiedenen "Threads" abhört, so dass der Client-Server betreibbar ist, zu wählen, eine Dienstanforderungsnachricht an den Transaktions-Server unter Verwendung entweder der ersten Nachrichtenwarteschlange oder der zweiten Nachrichtenwarteschlange zu senden.


 
7. Verfahren nach Anspruch 6, das ferner umfasst, den Client-Server in der zweiten Transaktionsmaschine einer vierten Nachrichtenwarteschlange in der zweiten Transaktionsmaschine zuzuordnen, wobei der Client-Server in der zweiten Transaktionsmaschine bewirkt, dass Nachrichten unter Verwendung der vierten Nachrichtenwarteschlange in der zweiten Transaktionsmaschine empfangen werden.
 
8. Verfahren nach Anspruch 7, das ferner umfasst, über den Transaktions-Server die Dienstanforderungsnachricht von der ersten Nachrichtenwarteschlange zu empfangen und über den Transaktions-Server eine Dienstantwortnachricht unmittelbar an den Client-Server in der zweiten Transaktionsmaschine unter Verwendung der dritten Nachrichtenwarteschlange in der zweiten Transaktionsmaschine zu senden.
 
9. Computerprogramm, das Programmanweisungen für die Ausführung durch ein oder mehrere Computersysteme umfasst, um das Verfahren nach einem der Ansprüche 6 bis 8 auszuführen.
 
10. Computerprogrammprodukt, das das Computerprogramm nach Anspruch 9 umfasst, das auf einem computerlesbaren Speichermedium vorgesehen ist.
 


Revendications

1. Système permettant d'assurer un transfert direct de messages dans un environnement de machines logicielles intermédiaires transactionnelles comportant une première machine transactionnelle (101) et une deuxième machine transactionnelle (102), comprenant :

une première file d'attente de messages (110) associée à un serveur transactionnel (103), la première file d'attente de messages et le serveur transactionnel résidant dans la première machine transactionnelle et la première file d'attente de messages consistant en une file d'attente d'accès direct en mémoire à distance, RDMA ;

une table annoncée (112, 113) dans l'environnement de machines logicielles intermédiaires transactionnelles, le serveur transactionnel fonctionnant pour publier la première file d'attente de messages dans la table annoncée ;

un serveur client (106) dans la deuxième machine transactionnelle apte à localiser la première file d'attente de messages dans la table annoncée et établir une connexion avec le serveur transactionnel dans la première machine transactionnelle en plaçant un message de requête de service directement dans la première file d'attente de messages ; et

une deuxième file d'attente de messages (107) dans la première machine transactionnelle et associée au serveur transactionnel ;

une troisième file d'attente de messages (108) dans la deuxième machine transactionnelle et associée au serveur client,

la deuxième file d'attente de messages et la troisième file d'attente de messages consistant en des files d'attente de communication interprocessus, IPC, la deuxième file d'attente de messages étant associée à un premier processus pont local (114) sur la première machine transactionnelle et la troisième file d'attente de messages étant associée à un deuxième processus pont local (115) sur la deuxième machine transactionnelle,

le serveur client étant apte à envoyer un message de requête de service à la troisième file d'attente de messages, le deuxième processus pont local étant apte à retransmettre le message de requête de service au premier processus pont local et le premier service pont local étant apte à retransmettre le message de requête de service à la deuxième file d'attente de messages,

le serveur transactionnel utilisant la deuxième file d'attente de messages pour communiquer avec un serveur client local (105) dans la première machine transactionnelle, et le serveur transactionnel se mettant à l'écoute à la fois de la première file d'attente de messages et de la deuxième file d'attente de messages simultanément dans des threads différents de sorte que le serveur client soit apte à choisir d'envoyer un message de requête de service au serveur transactionnel au moyen soit de la première file d'attente de messages, soit de la deuxième file d'attente de messages.


 
2. Système selon la revendication 1, dans lequel le serveur client dans la deuxième machine transactionnelle est associé à une quatrième file d'attente de messages (111) dans la deuxième machine transactionnelle, le serveur client dans la deuxième machine transactionnelle fonctionnant pour recevoir des messages au moyen de la quatrième file d'attente de messages dans la deuxième machine transactionnelle.
 
3. Système selon la revendication 2, dans lequel le serveur transactionnel fonctionne pour recevoir le message de requête de service depuis la première file d'attente de messages et envoyer un message de réponse de service directement au client serveur dans la deuxième machine transactionnelle au moyen d'une quatrième file d'attente de messages dans la deuxième machine transactionnelle.
 
4. Système selon l'une quelconque des revendications 1 à 3, dans lequel il n'existe aucun goulot d'étranglement ponctuel lors du transfert de messages entre des machines dans l'environnement de machines logicielles intermédiaires transactionnelles.
 
5. Système selon l'une quelconque des revendications 1 à 4, dans lequel la première file d'attente de messages utilise un protocole RDMA pour le transfert de données entre les première et deuxième machines transactionnelles sur un réseau InfiniBand.
 
6. Procédé permettant d'assurer un transfert direct de messages dans un environnement de machines logicielles intermédiaires transactionnelles comportant une première machine transactionnelle (101) et une deuxième machine transactionnelle (102), comprenant :

la fourniture d'une première file d'attente de messages (110) associée à un serveur transactionnel (103), la première file d'attente de messages et le serveur transactionnel résidant dans la première machine transactionnelle et la première file d'attente de messages consistant en une file d'attente d'accès direct en mémoire à distance, RDMA ;

la publication, par le serveur transactionnel, de la première file d'attente de messages dans une table annoncée dans l'environnement de machines logicielles intermédiaires transactionnelles ;

la localisation, par un serveur client (106), dans la deuxième machine transactionnelle, de la première file d'attente de messages dans la table annoncée et l'établissement, par le serveur client (106), d'une connexion avec le serveur transactionnel dans la première machine transactionnelle par placement d'un message de requête de service directement dans la première file d'attente de messages ; et

l'utilisation, par le serveur transactionnel, d'une deuxième file d'attente de messages (107) dans la première machine transactionnelle et associée au serveur transactionnel pour communiquer avec un serveur client local (105) dans la première machine transactionnelle,

une troisième file d'attente de messages (108) dans la deuxième machine transactionnelle et associée au serveur client,

la deuxième file d'attente de messages et la troisième file d'attente de messages consistant en des files d'attente de communication interprocessus, IPC, la deuxième file d'attente de messages étant associée à un premier processus pont local (114) sur la première machine transactionnelle et la troisième file d'attente de messages étant associée à un deuxième processus pont local (115) sur la deuxième machine transactionnelle,

le serveur client envoyant un message de requête de service à la troisième file d'attente de messages, le deuxième processus pont local retransmettant le message de requête de service au premier processus pont local et le premier service pont local retransmettant le message de requête de service à la deuxième file d'attente de messages,

et le serveur transactionnel se mettant à l'écoute à la fois de la première file d'attente de messages et de la deuxième file d'attente de messages simultanément dans des threads différents de sorte que le serveur client soit apte à choisir d'envoyer un message de requête de service au serveur transactionnel au moyen soit de la première file d'attente de messages, soit de la deuxième file d'attente de messages.


 
7. Procédé selon la revendication 6, comprenant en outre l'association du serveur client dans la deuxième machine transactionnelle à une quatrième file d'attente de messages dans la deuxième machine transactionnelle, le client serveur dans la deuxième machine transactionnelle fonctionnant pour recevoir des messages au moyen de la quatrième file d'attente de messages dans la deuxième machine transactionnelle.
 
8. Procédé selon la revendication 7, comprenant en outre la réception, par l'entremise du serveur transactionnel, du message de requête de service depuis la première file d'attente de messages, et l'envoi, par l'entremise du serveur transactionnel, d'un message de réponse de service directement au serveur client dans la deuxième machine transactionnelle au moyen de la troisième file d'attente de messages dans la deuxième machine transactionnelle.
 
9. Programme d'ordinateur comprenant des instructions de programme destinées à être exécutées par un ou plusieurs systèmes ordinateurs pour réaliser le procédé selon l'une quelconque des revendications 6 à 8.
 
10. Produit-programme d'ordinateur comprenant le programme d'ordinateur selon la revendication 9 fourni sur un support d'enregistrement lisible par ordinateur.
 




Drawing























Cited references

REFERENCES CITED IN THE DESCRIPTION



This list of references cited by the applicant is for the reader's convenience only. It does not form part of the European patent document. Even though great care has been taken in compiling the references, errors or omissions cannot be excluded and the EPO disclaims all liability in this regard.

Non-patent literature cited in the description