KR20110073279A - A method for performance optimisation of publish operation - Google Patents

Abstract

PURPOSE: A performance optimization method of publish operation is provided to identify whether or not a subscriber is waiting for a published message in response to the published message by a publisher. CONSTITUTION: A publisher(205) publishes a message having subject string. A matching engine(215) builds a delivery list of subscribers, which matches the subject string of the published message. A delay device(235) obtains a first subscriber from the delivery list to identify whether or not a subscriber is waiting for the message. The delay device checks the queue status information related to the subscriber to identify whether or not the subscriber is actively waiting for the message.

Description

A METHOD FOR PERFORMANCE OPTIMISATION OF PUBLISH OPERATION}

The present invention relates to a method of optimizing the performance of a publish operation.

Publish and subscribe (pub / sub) is an effective way of disseminating information to multiple users. pub / sub applications can help simplify the task of obtaining business messages and transactions in a timely manner for a dynamically changing and potentially large audience.

In a pub / sub system, publishers aren't interested in where their messages go, and subscribers aren't interested in where their messages come from. Instead, a broker generally guarantees the integrity of the message source and also manages the distribution of messages according to subscriptions registered with the broker.

Message topic strings generally provide a key for the delivery of messages between publishers and subscribers. Referring to the pub / sub system 100 shown in FIG. 1, instead of including a specific destination address in each message, the publisher 105 assigns a topic string to the message. The broker 110 matches the subject string of the published message with a list of subscribers 120, which subscribers 120 subscribe to receive messages published for the subject string. It includes. In response to the match, broker 110 sends the published message to subscriber 120.

In a queued pub / sub environment, publishing one message to multiple subscribers may involve a performance penalty when serializing delivery to the subscribers. For example, if a subscriber is not actively waiting for a message at the time the message is published, the message may need to be queued to disk (eg, by the publisher or by a broker). This entails a high performance cost at the time of publication and when the subscriber gets the message. Furthermore, if the delivery of a message occurs by using one thread, the process of queuing and subsequent delivery of the message may be timed by other subscribers further down in the processing chain. To prevent receiving it.

According to a first aspect of the invention, a method of optimizing performance of a publishing operation is provided for use in a system comprising a publisher for publishing a message and means for matching subscribers with the published message. The method further comprises: in response to the publisher publishing a message, determining whether the matching subscriber is actively waiting for the published message; And in response to determining that the subscriber is not actively waiting, deferring delivery of the published message.

According to a second aspect of the invention, there is provided an apparatus for optimizing performance of a publishing operation, the apparatus comprising a publisher for publishing a message and for use in a system comprising means for matching a subscriber with the published message. And the apparatus further comprises: means for determining, in response to the publisher publishing a message, whether the matching subscriber is actively waiting for the published message; And means for delaying delivery of the published message in response to determining that the subscriber is not actively waiting.

According to a third aspect of the invention, a computer program is provided, which, when executed on a computer, comprises program code means prepared for performing each step of the method.

The invention will now be described with reference to the preferred embodiments shown in the following figures. However, the accompanying drawings and descriptions are for illustration only.
1 is a block diagram of a pub / sub system in which a preferred embodiment is implemented.
2 is a more detailed block diagram of the system of FIG.
3 is a flowchart showing operational steps associated with a process according to a preferred embodiment.

Referring now to the drawings, preferred embodiments will be described.

Referring to the pub / sub system 200 shown in FIG. 2, the publisher 205; A broker 210 including a matching engine 215 and a defer appratus 235 of the present invention; And a plurality of subscribers 220, 225, and 230. Subscribers 220, 225, and 230 specify one or more queues 227, 240, and 245 used to receive published messages.

One or more applications may be associated with queues 237, 240, and 245. One or more applications may share access to the queue, in which case the subscription information further specifies a correlation identifier associated with the messages and the application also specifies a correlation identifier.

The process of the preferred embodiment will now be described with reference to FIGS. 2 and 3.

In the pre-processing phase, publisher 205 issues a message with a subject string. Furthermore, timeout is configured and subscribers' average message processing rates are obtained.

In step 300, the matching engine 215 builds a delivery list of subscribers that match the subject string of the published message.

Preferably, a lock is established to prevent additional subscribers from requesting messages about the published topic.

At step 305, the deferral device 235 gets a first subscriber from the delivery list to determine if the subscriber is waiting for the message (step 310). Preferably, the deferral device 235 checks the state information of the queue associated with the subscriber to determine if the subscriber is actively waiting for a message. The waiting subscriber is the one who issues the command to get a message from its queue.

If the deferral device 235 determines that the subscriber is waiting for the message, the matching engine 215 sends the message to the subscriber (step 320), and in step 325, the deferral device 235 sends the delivery list. Remove the subscriber from the

If the deferral device 235 determines that the subscriber is not waiting for the message, at step 315 the deferral device 235 adds the subscriber to the deferred list. At step 325, the deferral device 235 removes the subscriber from the delivery list.

At step 330, the deferral device 235 checks the delivery list to determine if there are more subscribers. If there are more subscribers, steps 305 to 325 are repeated so that each subscriber is processed and a delivery list and possibly deferred list are obtained.

In response to the lack of more subscribers, at step 335, the deferral device 235 determines whether there are subscribers in the deferred list. If the deferral device 235 determines that there are no subscribers on the deferred list, then the publishing work initiated by the publisher 205 is considered complete (step 340).

If the deferral device 235 determines that there are subscribers in the deferred list, in step 345 the deferral device 235 removes from the deferred list to determine if the subscriber is waiting for the message (in step 350). 1 Get subscribers.

Note that although at step 310 the subscriber may not have been waiting, when it reaches step 350, the subscriber may actually be actively waiting, and thus at step 350 there is a need for further checks.

If the deferral device 235 determines that the subscriber is waiting for the message, the matching engine 215 sends the message directly to the subscriber (step 355), and in step 360 the deferral device 235 defers the deferral. Removes the subscriber from the list. The process proceeds to step 365 as described below.

In step 350, if the deferral device 235 determines that the subscriber is not waiting for the message, in step 370, the deferral device 235 determines whether the set timeout mentioned above has been reached.

If the set timeout is reached, the matching engine 215 sends the message to the subscriber's queue (step 375). In step 360, the deferral device 235 removes the subscriber from the deferred list. As will be described later, the process proceeds to step 365.

If the set timeout is not reached, the deferral device 235 checks the average message processing rate of the subscriber against the set timeout to determine whether the subscriber will be active before reaching the timeout ( Step 380).

If the deferral device 235 determines that the subscriber will be active before reaching the timeout, then at step 385, the deferral device 235 is configured to prevent the expensive queuing operation from being performed by the subscriber in the deferred list. Leaves. As will be described further below, the process proceeds to step 365.

If the deferral device 235 determines that the subscriber is unlikely to become active before reaching the timeout, then at step 375, the matching engine 215 sends the message to the subscriber's queue. That is, no further delay can be accepted in delivering the issued message so that the message is queued to disk. In step 360, the deferral device 235 removes the subscriber from the deferred list, and the process proceeds to step 365, as described later.

Note that the timeout allows a balance between expensive queuing operations and the prevention of unacceptable delays in delivering issued messages.

In step 365, the deferral device 235 checks the deferred list to determine if there are more subscribers.

If there are more subscribers, steps 345 to 385 are repeated. Thus, the subscriber receives the message immediately, or causes the subscriber to remain in the queued or deferred list.

If, at step 365, there are no more subscribers, the publishing work initiated by the publisher 205 is considered complete (step 340).

Preferred embodiments will now be described.

In a first example, the timeout is set to 2 seconds, where subscriber 1 220 and subscriber 3 230 process messages on average, every second, and subscriber 2 225 processes messages every 10 seconds on average. do.

In step 300, the matching engine 215 constructs a delivery list of subscribers that match the subject stream of the published message, the delivery list comprising: subscriber 1 220; Subscriber 2 225; And subscriber 3 230.

At step 305, the deferral device 235 obtains subscriber 1 220 to determine if subscriber 1 220 is waiting for the message (step 310).

The deferral device 235 determines that subscriber 1 220 is waiting for the message, and at step 320, the matching engine 215 sends the message directly to subscriber 1 220. At step 235, the deferral device 235 removes subscriber 1 220 from the delivery list.

At step 330, the deferral device 235 checks the delivery list to determine if there are additional subscribers.

As there are additional subscribers, the deferral device 235 obtains the next subscriber in the delivery list, subscriber 2 225, to determine if subscriber 2 225 is waiting for the message (step 310).

The deferral device 235 determines that subscriber 2 235 is not waiting for the message, and at step 315, the deferral device 235 adds subscriber 2 225 to the deferred list. In step 325, the delivery device 235 removes subscriber 2 225 from the delivery list.

In step 330, the delivery device 235 checks the delivery list and determines that there are additional subscribers. In response, the deferral device 235 obtains the next subscriber, namely subscriber 3 230, from the delivery list to determine if subscriber 3 230 is waiting for the message (at step 310).

The deferral device 235 determines that subscriber 3 230 is waiting for the message, and at step 320 the matching engine 215 sends the message directly to subscriber 3 230. At step 325, the deferral device 235 removes subscriber 3 230 from the delivery list.

At step 330, the deferral device 235 checks the delivery list to determine if there are additional subscribers.

Since there are no more subscribers, at step 335, the deferral device 235 determines if there are subscribers in the deferred list.

The deferral device 235 determines that subscriber 2 235 is on the deferred list, and at step 345, the deferral device 235 determines whether subscriber 2 225 is waiting for the message (step 350). Subscriber 2 (225) is obtained.

The deferral device 235 determines that subscriber 2 225 is not waiting for the message, and at step 370, the deferral device 235 determines whether the set timeout mentioned above has been reached.

The deferral device 235 determines that the set timeout mentioned above has not been reached, and prepares for the set timeout (ie, 2 seconds) to determine whether Subscriber 2 225 will be active before reaching the timeout. Check the average message processing rate (ie, 10 seconds) of subscriber 2 (225) (step 380).

The deferral device 235 determines that subscriber 2 225 will not be active before reaching the timeout, and at step 375, the matching engine 215 directs the message to queue 2 of subscriber 2 225. Send. In step 360, the deferral device 235 removes subscriber 2 from the deferred list.

In step 365, the deferral device 235 checks the deferred list to determine if there are more subscribers, and since there are no more subscribers, the publishing work initiated by the publisher 205 is considered complete (step 340).

In the second example, the timeout is set to 2 seconds, subscriber 1 220 and subscriber 3 230 process messages every 1 second on average, and subscriber 2 225 processes messages every 10 seconds on average. .

In step 300, the matching engine 215 may include subscriber 1 220; Subscriber 2 225; And subscriber 3 230 to build a delivery list of subscribers.

At step 305, the deferral device 235 obtains subscriber 1 220 and determines that subscriber 1 220 is waiting for the message (step 310). In response, at step 320, the matching engine 215 sends the message directly to subscriber 1 220, and at step 235 the deferral device 235 removes subscriber 1 220 from the delivery list.

In step 330, the deferral device 235 checks the delivery list to determine if there are additional subscribers, and since there are additional subscribers, the deferral device 235 determines whether subscriber 2 225 is waiting for the message. The next subscriber, subscriber 2 225, is obtained from the delivery list.

Delivery device 235 determines that subscriber 2 225 is not waiting for the message, and at step 315, postponed device 235 adds subscriber 2 225 to the deferred list. At step 325, the deferral device 235 removes subscriber 2 225 from the delivery device.

At step 330, the deferral device 235 checks the delivery list and determines that there are additional subscribers. In response, the deferral device 235 obtains the next subscriber, subscriber 3 230, from the delivery list to determine if subscriber 3 230 is waiting for the message (at step 310).

The deferral device 235 determines that subscriber 3 230 is not waiting for the message, and at step 315 the deferral device 235 adds subscriber 3 230 to the deferred list. At step 325, the deferral device 235 removes subscriber 3 230 from the delivery list.

At step 330, the deferral device 235 checks the delivery list to determine if there are additional subscribers.

Since there are no more subscribers, at step 335, the deferral device 235 determines whether there are subscribers in the deferred list.

The deferral device 235 determines that there are subscribers in the deferred list, and in step 345, the deferral device 235 determines whether subscriber 2 225 is waiting for the message (step 350) and subscriber 2 225 Get)

The deferral device 235 determines that subscriber 2 225 is not waiting for the message, and at step 370, the deferral device 235 determines whether the set timeout mentioned above has been reached.

The deferral device 235 determines whether subscriber 3 230 is currently waiting for the message, and at step 355, the matching engine 215 sends the message directly to subscriber 3 230. In step 360, the deferral device 235 removes subscriber 3 230 from the deferred list.

In step 365, the deferral device 235 checks the deferred list to determine if there are more subscribers, and since there are no more subscribers, the publishing work initiated by the publisher 205 is considered complete (step 340). .

In a third example, referring to the second example above, in step 350, the deferral device 235 determines that subscriber 3 230 is currently waiting for the message, and in step 370, the deferred device 235 Determines whether the set timeout mentioned above has been reached.

The deferral device 235 determines that the set timeout mentioned above has not been reached, and at step 380, the deferred device 235 is configured to determine whether subscriber 3 230 will be active before reaching the timeout. Check the average message processing speed (ie, 1 second) of subscriber 3 230 against the timeout (ie, 2 seconds).

The deferral device 235 determines that subscriber 3 230 will be active before reaching the timeout, and at step 385, the deferral device 235 leaves subscriber 3 230 in the deferred list.

In step 365, the deferral device 235 checks the deferred list to determine if there are more subscribers. Since subscriber 3 230 is on the deferred list, at step 345, the deferral device 235 obtains subscriber 3 230 (at step 350) to determine if subscriber 3 230 is waiting for the message. do.

The deferral device 235 determines that subscriber 3 230 is currently waiting for the message, and at step 355 the matching engine 215 sends the message directly to subscriber 3 230. In step 360, the deferral device 235 removes subscriber 3 230 from the deferred list.

In step 365, the deferral device 235 checks the deferred list to determine if there are more subscribers.

Since there are no more subscribers, the publishing work initiated by the publisher 205 is considered complete (step 340).

Alternatively, if, in step 370, the deferral device 235 reaches the set timeout mentioned above, in step 375, the matching engine 215 sends the message to queue 3 of subscriber 3230. . In step 360, the deferral device 235 removes subscriber 3 230 from the deferred list.

In step 365, the deferral device 235 checks the deferred list to determine if there are more subscribers.

Since there are no more subscribers, the publishing work initiated by the publisher 205 is considered complete (step 340).

Preferably, the preferred embodiment employs a "two-pass" approach to delivering published messages, based on the latency associated with the subscriber activity. By deferring delivery for inactive subscriptions, it provides a mix of active and inactive subscribers.

The deferred list is repeated over and over after each active subscriber receives their copy of the message. The deferred list is cross referenced against a subset of the list of waiting subscribers that became active after the publishing job began.

If the subscriber is not active within the set timeout, the decision is made to queue the message. Thus, other subscribers further down in the processing chain have the opportunity to receive a copy of the message within an acceptable threshold.

Advantageously, delivery to infrequently active subscribers can still be queued. However, queued delivery to these subscribers is delayed until the latest possible opportunity.

Preferably, the timeout is configurable (e.g. with one or more average time intervals for previous publishing jobs).

Preferably, the option as to whether the deferral device is called is associated with the number of subscribers to achieve the best overall performance (eg, the deferral device may be more effective if there are a relatively large number of subscribers).

Advantageously, the performance of the deferral device is increased by offloading the transfer of the deferred list processing to secondary threads.

Preferably, note that new subscribers are not excluded from subscription to the subject of the published message once the iteration of the deferred list begins. Rather, the secondary list of new active subscribers is maintained for the time that the published message is being delivered. If the publishing task is deemed complete, a lock established to prevent further subscribers from requesting messages about the published topic is released, and the secondary list is then published. It is added to the delivery list configured for the message.

Some or all of the methods of the preferred embodiments of the present invention may be suitably and usefully implemented in a plurality of logic devices, or logic devices, including logic elements configured to perform the steps of the method, and may also be such logic elements. It will be apparent to those skilled in the art that these may include hardware components, firmware components, or a combination thereof.

Likewise, some or all of the logic arrangements in accordance with preferred embodiments of the present invention may be suitably implemented in a logic device including logic elements to perform the steps of the method, such logic elements being, for example, It will be apparent to those skilled in the art that it may include components such as logic gates in a programmable logic array or application specific integrated circuit. This logic configuration further advances the logic structures in such an array or circuit using, for example, a virtual hardware descriptor language, which can be stored and delivered using a fixed or transportable carrier medium. It may be further implemented in enabling the elements to establish, either permanently or permanently.

The method and configuration described above may also be appropriately performed, in whole or in part, on software executing one or more processors (not shown in the figures), which software may be adapted to a suitable data-carrier (such as a magnetic or optical disk). It will be appreciated that this may also be provided in the form of one or more computer program elements executed on the drawings (not shown). Channels for the transmission of data can likewise include a storage medium of all descriptions as well as a signal-carrie medium such as a wired or wireless signal-carrie medium.

The invention may further be suitably further implemented as a computer program product for use in a computer. Such an implementation may be immobilized on a physical medium, such as a computer readable medium (eg, a diskette, CD-ROM, ROM, or hard disk), or on a physical medium (eg, as a physical medium, for example, optical or Similar communication lines and the like, but are not limited to such examples, or may be substantive using wireless technologies (e.g., including, but not limited to, microwave, infrared, or other transmission technologies). As such, it may include a series of computer readable instructions transmittable to a computer system via a modem or other interface device. The series of computer readable instructions implements some or all of the functionality previously described herein.

Those skilled in the art can write these computer readable instructions in a number of programming languages for use in many computer architectures or operating systems. Furthermore, these instructions may be stored using current or future memory technology (eg, including but not limited to semiconductor, magnetic, or optical technology), and may be stored in current or future communication technology ( For example, but not limited to, optical, infrared, or microwave. Such a computer program product may be distributed, for example, on a system ROM or fixed disk as a removable medium with printing or electronic documentation (eg, shrink-wrapped software, computer Pre-loaded in the system), or from an electronic bulletin board or server over a network such as, for example, the Internet or the World Wide Web.

Alternatively, the preferred embodiment of the present invention may be realized in the form of a computer-implemented method. The computer-implemented method employs a service that includes, when employed in and executed on a computer infrastructure, computer program code operable to cause the computer system to perform all the steps of the method described above. It is a way.

It will be apparent to those skilled in the art that many modifications and variations can be made to the embodiment described above without departing from the scope of the invention.

Claims (13)

A method of performance optimization of a publish operation for use in a system comprising a publisher for publishing a message and a means for matching a subscriber with the published message. ,
In response to the publisher issuing a message, determining if the matching subscriber is actively waiting for the published message; And
In response to determining that the subscriber is not actively waiting, deferring delivery of the published message.
Way. The method of claim 1, wherein the method
Removing the subscriber from a delivery list; And
Adding the subscriber to the deferred delivery list.
Way. The method according to claim 1 or 2, wherein the method
Determining whether a configurable timeout has been reached;
Way. The method of claim 3, wherein the method
In response to determining that the settable timeout has been reached, recording the issued message to disk.
Way. The method of claim 4, wherein the method
Removing the subscriber from the deferred delivery list.
Way. The method of claim 3, wherein the method
In response to determining that the settable timeout has not been reached, determining whether the subscriber is actively waiting before reaching the settable timeout.
Way. The method of claim 6, wherein the method
In response to determining that the subscriber is not actively waiting, continuing with the deferred delivery of the published message.
Way. The method of claim 6, wherein the method
In response to determining that the subscriber is not actively waiting, writing the issued message to disk.
Way. The method of claim 6, wherein determining whether the subscriber is actively waiting further comprises analyzing a stored metric associated with the subscriber.
Way. 10. The method of claim 9, wherein the stored metric comprises an average message processing rate.
Way. The method according to claim 1 or 2, wherein the method
In response to deferring delivery of the published message, re-determine whether the subscriber is actively waiting; And
In response to determining that the subscriber is actively waiting, delivering the published message to the subscriber;
Way. An apparatus for performance optimization of a publish operation, for use in a system comprising a publisher for publishing a message and means for matching subscribers with the published message.
Means for determining, in response to the publisher publishing a message, whether the matching subscriber is actively waiting for the published message; And
Means for delaying delivery of the published message in response to determining that the subscriber is actively waiting;
Device. A computer program, wherein the computer program is prepared to perform each step of the method of any one of claims 1 to 11 when executed on a computer.
Computer programs.

Images