Thoughts on replacing relational databases with matrix state tables in workflow platforms

 

Currently, all workflow systems require the support of relational databases. A workflow system is required, especially when the workflow engine is started during running, you need to use a database to run the process instance and record the status value of the process. It can be said that the workflow engine has a very close relationship with the database. without the support of the database, the workflow cannot run, however, if we extend our thinking and think about it like this, if the running of the workflow system must depend on relational databases, the application scope of the workflow will be limited. For example, in some embedded platforms, the disk and memory are relatively small, and the operating system does not support databases. If we want to start the workflow engine without the support of relational databases, will the workflow be unable to run?
Of course not.

I am trying to build an embedded workflow engine that does not rely on the database system. The principle is to replace the relational database with a custom topology matrix table, I imagine adding a matrix model in the jwfd platform. The previously designed flowchart can only be parsed into the database, and a method will be added later to resolve it to the matrix status table, this matrix table is actually a two-dimensional array with a state value relationship. It can save the topology relationship of the flowchart or store the state value of the process node running process, then the process engine can be run in a relational database or in a matrix table. If we are on a machine without any database system installed, we only use the jwfd platform, the process from the design flow chart to the fully automated operation flow can be realized .... It's really interesting to do this .... (For more information about the Matrix workflow model, see my blog.
Http://comsci.iteye.com/blog/1188482 system learning documents with reference value "using a structured matrix to model SBA Workflow" on the top of taoke 88)

Jwfd has always been implemented using the swing interface. one advantage is that it can be used across platforms and run in Linux .... A small embedded workflow system does not rely on databases .. Really interesting... In addition, the automatic data processing mechanism... Very interesting .... The advantage of mastering the core technology is to do what you want... Haha