Browse Prior Art Database

A method and mechanism to manage business processes on distributed servers

IP.com Disclosure Number: IPCOM000235927D
Publication Date: 2014-Mar-31
Document File: 8 page(s) / 154K

Publishing Venue

The IP.com Prior Art Database

Abstract

The disclosure solves below three problems in business process management centralized model: 1. Each node(management node, compute node) is not in equality, 2. Complex to scale in /scale out, 3. The database bottleneck may be hit along with servers number increasing via four actions as below: 1. Store process instance location info in process id, or wrapper process instance location info with process id, 2. Route process instance handling according process instance location info, 3. Query process instances in parallel, and do claim 1 for the query result, 4. Decouple process instance location and the physical target, and do fail over via this

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 55% of the total text.

Page 01 of 8

A method and mechanism to manage business processes on distributed servers

We keep each virtual machine is in equality. Each virtual machine contains:

A business process management server running,


1.

A database running for the process server,


2.

A volume is attached, which is used to


3.

store the database's data,


1.

store process instance location name, here we name it server name.

2.

A process agent application running in the process server,


4.

      Figure 4. Process agent, database and attached volume Please notice here:

A volume,


1.

comparing with virtual machine disk, it can be reattached even if the virtual machine is destroyed, on the contrary, the virtual

1



Page 02 of 8


2.


3.


4.

machine disk can not be reattached.

The map

, here server is the server name, which can be used to represent the process instance location, and host can be

the physical target, for example, it's host name or ip address.

Process agent can run in process server as a normal application or system application, it can also work as a component of process

server

A process agent:

Maintains a process instance location and target location map, which can be used to do broadcast or route,

Contains logic to handle query in parallel, remote process routing, and local process handling

Query processes in parallel scenario:

End user query all processes under certain condition,


1.

Load balancer dispatches the query to any one of the server under it, say it dispatches the query to server 2,


2.

Process agent on server 2 works as leader for this query, except do the query in local, it broadcast the query to others also in


3.

parallel with response end point provided,

Each process agent do query local and send result to the response end point,


4.

When the result is ready, process agent on server 2 return the result to load balancer,


5.

Please notice the result format: 1@server1, here 1 is the process id, and server1 is the process instance location (on server1). See figure 6 for details.

Load balancer response result to end user.


6.


1.

2.

2



Page 03 of 8

Figure 5. Query in parallel

Figure 6. Pro...