Browse Prior Art Database

A Method and System to dispatch BPM tasks on cloud

IP.com Disclosure Number: IPCOM000248698D
Publication Date: 2016-Dec-28
Document File: 9 page(s) / 217K

Publishing Venue

The IP.com Prior Art Database

Abstract

In Business Process Management(BPM) cloud, there may be thousands of BPM engine nodes and thousands of BPM tasks. This disclosure presents a method and system to first split the BPM process into task groups based on transaction boundary. Then task groups will be dispatched to different BPM nodes according to task resource consumption and node resource availability. If there are several nodes available, adhesiveness and location will be taken into consideration. This will help to achieve high efficience, avoid crash and provide differen SLA(service level achievement) in BPM cloud.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 28% of the total text.

1

A Method and System to dispatch BPM tasks on cloud

In Business Process Management(BPM) cloud, there will be thousands of BPM engine nodes and thousands of BPM tasks. A typical BPM cloud infrastructure involves a task dispatch component that matches a BPM task to a BPM engine node based on available runtime resources of the node. Current workloads of individual BPM node may be monitored and newly coming task may be assigned to nodes that are less busy. But the dispatching rule does not consider the following conditions:

The nodes may have different capacity/resources/hardware allocated when it is created. For example, some BPM nodes may have more powerful  CPU while other BPM nodes may have quicker I/O. Different BPM tasks consume different resources. Some BPM tasks requires more CPU while other BPM tasks may requires more memory. Same BPM tasks of different BPM instances require different resource. Different BPM instances may have different business object associated with  them, which makes some BPM instances requires more resources than others of the same template. BPM tasks in one transaction may be dispatched to different nodes, which requires higher communication among different nodes during commit and  rollback.

So sometimes, as an example, a I/O oriented task may be dispatched to a node with slow I/O, which affect the task execution efficiency. As another example, a memory oriented task may be dispatched to a node without enough free memory which causes the node crash.

Accordingly, there is a need to dispatch BPM tasks to different BPM nodes not just based on current node resource availability. This disclosure presents a method and system to dispatch tasks to different BPM nodes considering task resource consumption and node resource availability. This disclosure presents a method and system to dispatch process task groups on BPM cloud according to task group resources assumption and node

resource availability.

Key Modules: 1. Task Bundler Task Bundler parses the process model and splits processes into task groups based on transaction boundary after process application is deployed 2. Node Monitor Node Monitor in every BPM node provides real-time statistics of that cloud node's resource, to dynamically classify it to an I/O or CPU or Memory node 3. Task Group Resource Analyzer Task Group Resource Analyzer analyze the task group resource consumption based on task type and Business Object (BO) size 4. Task Group Dispatcher Task Group Dispatcher distributes newly coming task groups to different BPM Engine images based on dispatch rules

Advantages:

1. Optimize BPM tasks execution

2. Avoid image crash due to resources use-up

2

3. More efficient use of cloud computing resources 4. Generate different execution plan based on customer requirements: economical, efficient, etc.

System Architecture

(*) BPMDB means Business Process Management Database

BPM Engine1. BPM Engine which runs business process instance . In BPM cloud, each node is a BP...