Data processing: database and file management or data structures – Database design – Data structure types
Reexamination Certificate
2000-03-17
2003-09-02
Mizrahi, Diane D. (Department: 2175)
Data processing: database and file management or data structures
Database design
Data structure types
Reexamination Certificate
active
06615215
ABSTRACT:
FIELD OF THE INVENTION
The invention relates generally to the field of message processing, and more specifically, to a system and method for adjusting the processing of messages according to changes in queue depth.
BACKGROUND OF THE INVENTION
MQSeries™, which is a product of the International Business Machines Corp. (IBM), performs message queuing and transport within and among a large variety of computer system types and networks. This product enables a computer program to place (enqueue) a data message into a named queue, the data message generally being picked up by another computer program. It is also possible that the queue designated as the target of the enqueue actually resides in a different physical computer. In this case, MQSeries accepts the enqueue request into the local system and then asynchronously transmits the message via a network to the remote system in which the queue is actually found. The message may then be dequeued by a program on that system. As described above, MQSeries enables different computer programs to exchange information via messages on queues and to maintain ignorance of where the recipient of a sent message is or the type of network implemented between the host machines.
In addition to this primary functionality, there are several ancillary features that aid in the use and management of MQSeries. Among these ancillary features are the performance event message feature and the trigger feature.
The performance event message feature is designed to report a “significant” change in the depth of a queue, where the depth of the queue corresponds to the number of messages in the queue. What constitutes a significant change is determined by a combination of settings, described below, which may be specified by a system administrator in MQSeries. The settings are attributes of all “local” queues, which are queues that actually hold or can hold messages.
A first of these attributes of a queue is a maximum depth attribute. This attribute sets the maximum number of messages that a queue can hold. After the maximum is reached, no more enqueues, i.e. placement of messages, will be accepted into the queue. Another attribute of each queue is a queue depth high percentage attribute. This attribute sets a percentage of the maximum depth value that is recognized as a high threshold. If the depth of the queue equals or exceeds the number of messages represented by this value, the queue is a candidate for generation of a performance event-high message.
The value of a queue depth high event status attribute determines whether to check if the high threshold has been reached. This attribute is a flag value that can be set to enabled or disabled. If the flag is enabled, then a performance event-high message will be generated whenever the number of messages in the queue reaches or exceeds the high threshold, which corresponds to the percentage of the maximum depth set by the queue depth high percentage. If the status is set to disabled, however, the performance event-high message will be suppressed. Anytime a performance event-high message is generated, the queue depth high event status attribute is automatically set to disabled, and a queue depth low event status attribute, described below, is set to enabled.
Like the high threshold attributes, MQSeries also maintains attributes for a low threshold. Among the low threshold attributes is a queue depth low percentage attribute, which sets a low threshold as a percentage of the maximum depth value described above. If the depth of a queue is equal to or less than the low threshold, the queue is a candidate for generation of a performance event-low message.
Each queue also has a queue depth low event status attribute. This attribute represents a flag value that can be set to enabled or disabled. If the flag is enabled, then a performance event-low will be generated whenever the number of messages in the queue is less than or equal to the low threshold, which corresponds to the percentage of the maximum depth set by the queue depth low percentage. If the status is set to disabled, the event will be suppressed. Anytime a performance event-low message is generated, the queue depth low event status attribute is automatically set to disabled, and the queue depth high event status attribute is set to enabled.
In addition to these local queue attributes, MQSeries implements special facilities for the delivery of events to a monitoring computer program. Part of this implementation is a performance event queue. The performance event queue is a specially named local queue into which MQSeries places performance event messages, such as the performance event-low and performance event-high messages. There is only one performance event queue in each executing instance of MQSeries. If a program wishes to receive the messages placed in the performance event queue, the program opens the queue and waits for the messages to appear using the same application programming interfaces (APIs) provided by MQSeries for use in dequeuing messages from any of its local queues. The performance event messages are enqueued synchronously to the performance event queue with the messaging event, such as an enqueue or dequeue on some queue that caused the performance event message to be generated.
A typical use of the above features can be exemplified with a queue defined with the following attributes: maximum depth set to one thousand; queue depth high percentage set to ten; and queue depth high event status set to enabled. With maximum depth set to one thousand and queue depth high percentage set to 10, the high threshold is 100. In addition, queue depth low percentage is set to one, and queue depth low event status is set to disabled. With queue depth low percentage set to 1, the low threshold is 10.
As messages first begin to arrive in the queue, no performance event is recognized. Although the queue is a candidate for a performance event-low each time an enqueue occurs and the resulting queue depth is ten or less, the event is suppressed because the queue depth low event status is disabled. When the queue's depth reaches one hundred, the queue becomes a candidate for generation of a performance event-high message. This performance event message will be generated because the queue depth high event status is enabled. At the same time the performance event message is generated, the queue depth high event status will be set to disabled, and the queue depth low event status will be set to enabled. If over time the depth of the queue drops back to ten or less, then the performance event-low message will be generated, the queue depth low event status will be disabled, and the queue depth high event status will be re-enabled.
As demonstrated by the above example, the queue depth high event status and the queue depth low event status are designed to operate in a flip-flop manner, such that when one is enabled, the other is disabled and vise versa. This flip-flop manner supports a model often used in many computer system management products, in which various managed resources are seen as being either in an alertable state or a normal state. If the resource was included as an item in a management display panel, the visual representation of the resource may, for example, change its color to red when in an alertable state, and then back to green when the state returns to normal. When the alertable state is displayed, computer operations procedures may be invoked as a response to the state with the intent of returning the resource to a normal state.
While this behavior is well suited to a systems management model for resource monitoring, it is not well suited for a task dispatching monitoring model. MQSeries provides a trigger feature for task dispatching. There is no coordination, however, between the performance event message feature and the trigger feature in the MQSeries product.
With respect to the trigger feature, MQSeries employs a set of facilities for starting computer programs when queues become active, such as when messages have arrived in a queue, wit
CommerceQuest Inc.
Mizrahi Diane D.
Mofiz Apu M
LandOfFree
Method for graduated load sensitive task dispatching in... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for graduated load sensitive task dispatching in..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for graduated load sensitive task dispatching in... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3024748