News: BRMS visual rules 3.5 includes new component 'Build Tools'
Version 3.5 of visual rules, a business rules management system (BRMS), has been released. Visual rules is used to graphically model business rules and generate program code as well as documentation of these rules. The focus of version 3.5 is "Quality in Teamwork." With a new component, 'Build Tools,' build and deployment processes (execute rules, tests, test coverage analyses, generate reports and documentation, and generate and deploy code) for the business rules are automated. This guarantees that rules defined or modified in teamwork have been completely tested and are logically correct before they enter into productive use. Rule projects can now be split into multiple smaller projects with full reusability of rules and data to help streamline rule management within teams. With visual rules, how rules are processed is defined in the models. In this way business rules become transparent and can easily be customized to new business requirements. The generated program code is ready to integrate in enterprise applications. No inferences are used. The rule models can also be deployed as EJB on all application servers that conform to J2EE standards, such as SAP Web AS and IBM Websphere. The business rules can be executed ad-hoc within visual rules, without being integrated in any application. The license is free for a 30-day evaluation and for non-commercial use. What do you think of this latest release of visual rules?
- Posted by: Diana Caputo
- Posted on: October 25 2006 12:13 EDT
I'm curious to know if visual rules has any performance numbers that show scalability with respect to rule complexity, rule count and dataset size. Looking at the website, it seems like a procedural approach where graphs are compiled to pure java code. peter
You are right, visual rules® follows a rather procedural approach although visual rules operates directly on the business object model. The graphical business rules models that define the rule flow are automatically translated into executable Java code. This code needs no inference what results in a very high performance of rule execution. It therefore performs much better than business rules engines that use a RETE inference engine. Thus, visual rules® is perfectly suited for mass data processing or for fast online processing. visual rules® typically scales better than linear with increasing length of data input/output vector. It also scales better than linear with increasing amount of rule trees. This good scalability is caused by the fact that within rule trees typically only a small fraction of facts and rule nodes are required per rule tree call. Whereas RETE inference engines scale very poorly with increasing number of facts and rules. Just to give you some figures, we have done performance measurements of the rule execution when using visual rules' Database Connectivity to access data in a database and perform some rule based calculations. We have done this performance test with an Oracle 9.2 and the DB driver ojdbc14.jar on a PC with a single Intel Pentium M 2 GHz /w 1 GB RAM under MS Windows 2000 5.00.2195 /w Service Pack 4. - Insert of 1m data sets in a DB table execution time: 490 sec. = 8.2 min. - Select of 1m data sets, execution of one calculation per data set and Update of the data set execution time: 540 sec. = 9 min. - Select of 1m data sets, execution of more than 100 Rule nodes and Update of the data set execution time: 630 sec. = 10.5 min. Another example: customers are using visual rules® with an Oracle DB for very complex transaction processing (money laundering detection of financial transactions) with the following system key indicators · data quantity: 2.6 m business connections, 6 m persons, 8.5 m addresses · 10 rules projects with up to 15,000 rule nodes · 5 - 35 m transactions/day, e.g. 9.4 m transactions processed in 43 min. Please feel free to address further requests to firstname.lastname@example.org
The latest version of Visual Rules can be found here: