Electrical computers and digital processing systems: multicomput – Multicomputer data transferring via shared memory
Reexamination Certificate
1999-04-05
2002-08-13
Coulter, Kenneth R. (Department: 2154)
Electrical computers and digital processing systems: multicomput
Multicomputer data transferring via shared memory
C709S223000, C709S224000, C709S206000, C714S043000, C370S428000
Reexamination Certificate
active
06434605
ABSTRACT:
TECHNICAL FIELD
The present invention relates generally to data processing systems and more particularly to automatic detection and recovery for problems arising on a channel that interconnects queue managers.
BACKGROUND OF THE INVENTION
The MQSeries software from International Business Machines Corporation is a type of message oriented middleware. In particular, MQSeries is messaging software that allows application programs to exchange information across different software platforms. The MQSeries software resolves problems of interconnectivity. Hence, an application that runs on a first type of platform may communicate with another application that runs on a different type of platform. The differences in platforms may be due, for example, to differences in operating systems.
The MQSeries software utilizes queues to facilitate the exchange of information between applications. As shown in
FIG. 1
, a first application
10
may send a message
12
to a queue
14
. Subsequently, second application
16
may receive the message
12
from the queue
14
. The transfer of the message
12
may be performed asynchronously such that the application
10
need not wait on the application
16
and application
16
need not wait on the application
10
.
With the MQSeries software, a message is a string of bytes that has meaning to the applications that use the message. Each message includes application data and a message descriptor. The message descriptor identifies the message and contains control information such as type information and the priority assigned to the message by the sending application. The application data may take many forms. The content of the application data is controlled by the application that constructs the message.
A queue is a data structure that stores messages. Each queue exists independently of the applications that use the queue. As shown in
FIG. 2
, queues are managed by queue managers. The queue manager insures that messages are put on the proper queue and routed to the appropriate destinations. Each queue manager is manifest as a separate object of the queue manager or object type. As shown in
FIG. 2
, the application
10
sends a message
12
to the queue manager
18
, which stores the message
12
on the queue
14
. The queue manager
18
retrieves the message
12
from the queue
14
and forwards the message to the application
16
. As will be described in more detail below, the application programs are written using a common program interface, known as message queue interface (MQI). The MQI allows application programs that are running on different computers and/or different operating systems at different locations to communicate with queue managers. The applications communicate with the queue manager by issuing MQI calls.
Communication between applications need not be in one direction alone. As shown in
FIG. 3
, applications
10
and
16
may both receive messages and transmit messages. In the example depicted in
FIG. 3
, message
12
is sent from application
10
through queue
14
onto application
16
. Conversely, message
20
is sent through queue
22
from application
16
to application
10
.
The MQSeries software exploits the concept of a channel. A channel is a logical communication link between two queue managers.
FIG. 4
depicts an instance wherein two channels
38
and
40
interconnect queue managers
32
and
36
. Channels are available in two varieties: message channels and MQI channels.
FIG. 4
depicts some examples of message channels. Message channels are unidirectional and serve to transfer messages from one queue manager to another. Message channels connect the queue managers via message channel agents (MCA's). In the example depicted in
FIG. 4
, queue manager
32
manages queue
30
and queue manager
36
manages queue
34
. Message channel
38
links queue manager
32
and
36
so that messages may be sent over the message channel from queue manager
32
to queue manager
36
. Message channel
40
carries messages from queue manager
36
to queue manager
32
.
An MQI channel connects a client to a queue manager on a server machine. A client is a section of code that is installed on a machine to accept MQI calls from applications and pass them on to an MQI server machine. An MQI server is a queue manager that provides queuing services to one or more clients. In the example shown in
FIG. 5
, client
40
is interconnected by MQI channel
48
with queue manager
44
. The queue manager
44
manages access to queue
42
. The MQI channels, unlike message channels, are bi-directional so that both calls and responses may be sent over a single channel.
As mentioned above, message channels interconnect MCA's. An MCA is a program that controls the sending and receiving of messages. An MCA is positioned at each end of a channel. One MCA takes messages from a transmission queue (defined below) and puts the messages on the communication link. The other MCA receives
35
messages and delivers the messages to a remote queue manager. The MCA that initiates a communication is called a caller MCA, whereas the MCA that responds to a communication is a responder MCA. As shown in
FIG. 6
, MCA's
62
and
64
are at respective ends of channel
60
. Channel
60
interconnect queue manager
50
and queue manager
52
.
Queue manager
50
includes a transmission queue
54
. A transmission queue is a special type of local queue that is used to store messages temporarily before the messages are transmitted by an MCA to a remote queue manager. In the example shown in
FIG. 6
, MCA
62
transmits messages from the transmission queue
54
to the remote queue manager
52
.
The queue manager
50
includes an initiation queue
56
. The initiation queue
56
holds messages that are used to trigger a channel initiator
66
to start the MCA
62
. Hence, before a message is sent, the message is first stored in the transmission queue
54
. If the message satisfies triggering criteria for the queue, a separate message is sent to the initiation queue
56
, causing the triggering of the channel initiator
66
. The channel initiator
66
acts as a trigger monitor for sending MCA's (such as sender MCA
62
). The channel initiator
66
causes the sender MCA
62
to grab the message from the transmission queue
54
and send the message over the channel
60
to the remote queue manager
52
.
The remote queue manager
52
includes a channel listener
68
. A channel listener
68
is a program that monitors the channel to determine when to start the responder MCA
64
. The responder MCA
64
is started in response to a start-up request from sender MCA
62
. The channel listener
68
detects incoming network requests and starts the associated channel.
One difficulty with systems developed using the MQSeries software is that problems may arise relating to a channel. For example, a channel may fail. The MQSeries provides no built-in mechanism for automatically detecting such a failure and responding by attempting to restart the channel. Similarly, channel initiators and listeners may fail. Once again, there is no built-in mechanism within the MQSeries software for automatically restarting or reviving channel initiators or listeners.
The MQSeries software allows messages to be sent in predefined sequences over channels. Sequence numbers may be attached as tags to messages so that messages are sent and received in the same order. Unfortunately, there is no mechanism provided within the MQSeries software for automatically detecting and recovering from sequence errors when the messages get out of sequence.
SUMMARY OF THE INVENTION
The present invention addresses the limitations of conventional systems by providing a mechanism for automatically detecting and recovering from problems that arise relating to a channel in a distributed computing system. Specifically, in one embodiment of the present invention, failure of the channel is automatically detected. Upon detection of failure of the channel, the embodiment of the present invention attempts to automatically r
Faulkner Paul
Horton Jeffrey S.
Coulter Kenneth R.
WorldCom, Inc.
LandOfFree
Automatic detection and recovery for problems arising with... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Automatic detection and recovery for problems arising with..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Automatic detection and recovery for problems arising with... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2927332