David McCoy

A member of the Gartner Blog Network

David W. McCoy
Managing VP
15 years at Gartner
29 years IT industry

David W. McCoy is a managing vice president and Gartner Fellow emeritus. He currently leads the Business Process Management group as team manager and researches business process management (BPM), business rule management (BRM) and cost optimization. Read Full Bio

Coverage Areas:

← →

Are You Building Your Own Business Rule Engine Functionality?

by David McCoy  |  October 30, 2008  |  2 Comments

Simple question.  Are you – vendor or user – writing your own BRE?  I think you are.  In fact, I’m sure you are. Limited standards-based routes, merely-competitive functionality expectations, shrinking independent BRE market, OEM overkill, specialized integration needs, specific vertical flavoring – all these factor in to the decision.

If you are willing to talk, let me know your plans and expectations.  Jim Sinur and I have a research position on this, soon to be published.  We know what we see happening and what we think is going to happen, but it never hurts to get another bit of corroborating evidence, or to be told we are completely wrong.

Either way, we thank you in advance…

UPDATED on 30 October -  When I talk about building your own BRE, I mean, “coding specific, custom logic to process a collection of business rules in any fashion instead of using an existing BRE to process those rules.”  This means that I am not just asking for examples of building a RETE-based (or other algorithm-based) complex business rule engine.  I still want those, but I am also asking about scenarios where you are building any alternative to a traditional BRE, regardless of its power or lack thereof.  Also, if you are extending an open source BRE (e.g., DROOLS) that is of interest as a “build.”  Finally, I am not interested in the basic IF-THEN-ELSE processing that we all do anytime we write software.  Thanks to James Taylor for his post which triggered my decision to update the original query.

2 Comments »

Category: Business Rule Management (BRM)     Tags: , ,

2 responses so far ↓

  • 1 James Taylor   October 30, 2008 at 5:38 pm

    Personally I think you would have to be insane to do this – what would you hope to gain? This is a bit like saying there are too few choices for database management systems so I will build my own!
    If nothing else you should start with the Drools open source engine and go from there…

    JT

    James Taylor
    Author of Smart (Enough) Systems

  • 2 David McCoy   October 30, 2008 at 11:31 pm

    Hi James:

    We live in an insane world and people do insane things every day, so a plea to sanity is not a good way out of this one. The analogy to databases is a good one, but slightly unfair. The DB world has killer good standards (ODBC, JDBC, SQL) and most organizations already have a database or two on-site. No need to build one when (a) you already have access to one and (b) you have a simple standards mechanism to get to it. As for BRE, most organizations do NOT already have one and the standards are not at all hot. This tends to push people to build, from what we are seeing.

    However, I think that I do need to clarify one thing – if I am reading between the lines on your post. Any end user building the inferencing part of the BRE – e.g., implementing the RETE algorithm– is really going overboard and I tend to agree with the insanity charge. When I talk about building your own BRE, I mean, “coding specific, custom logic to process a collection of business rules in any fashion instead of using an existing BRE to process those rules.” We see this when vendors and users really just want a stripped down rule processing capability. Often, they have no interest in a RETE level of processing.

    Starting with DROOLS – yes, totally agree – many of the JSR-94 based open source offerings are a great starting point. But look at all of them out there… it seems that everyone has been building a JSR-94 engine. Even so, if you start with DROOLS and extend it, you are in effect, still building BRE-level functionality.

    Look at the BPM world – how many of those guys have hand-built a BRE (many without a RETE class engine)? Lots. We talk to users who are doing the same. Many of them don’t want a high-end BRMS or even an impressive BRE. They just want some simple rule-based routing capabilities. Again, not so insane – certainly not as wicked tricky as building a full-blown BRE from scratch.

    Remember, I am interested in actual behavior and not ideal behavior. However, given my caveat, we might be closer on this topic than it seemed. Thanks for your insight.

Leave a Comment