500,000 Decisions per Second

Decision Table Example – Tax

Here is another example of a decision table simplifying and solving complex business logic.

Similar to Decision Table Example – Vacation Days, this is another challenge posed by the DM Community.

“With tax season upon us, take a look at the following decision table created by a business analyst from a tax agency:”

tax form
As is typical, the model contains a number of discrepancies and can be simplified. A full description of how this model was simplified with a discussion of the gaps and overlaps in the specification is shown in my proposed solution.

In summary, I proposed this DMN decision model.
DMN decision table

Execution

I deployed the decision model into a RapidGen Genius DMN execution engine running on an HP EliteDesk 800 G1 SFF with a 4-core Intel Core i5 4570K at 3.2GHz, 16Gb RAM and 64-bit Windows 10, to check the veracity of the rules and test performance.

Against bulk generated data, and with transaction logging disabled, I measured a throughput of 500k decisions per second – very impressive for such modest hardware!

And just to be clear that was only using one of the four cores.

“RapidGen execution engine: 500k decisions per second on one core”

To measure how scalable these results were, I reconfigured RapidGen to run concurrently on all four cores. As shown below there is some variability but throughput for two of the four cores exceeded 500,000 decisions per second.

500,000 decisions per second

Using additional cores scales almost linearly, reaching over 1.9M decisions per second with four cores, as shown below.

4 cores 1.9M decisions per second

This certainly outperforms other Business Rules Management Systems I have used over the years.

About the Author

Jan PurchaseJan Purchase has been working in investment banking for 19 years, the last 13 of which he has focused exclusively on helping clients structure their requirements and automate their financial processing with business decision management, decision modeling, business rules management and business process modeling.

He is a founder of Lux Magi Financial Rules, a company specializing in delivering, training and mentoring all of these concepts to financial organizations. Lux Magi has been applying business decision modeling and rules to the automation of financial processing and regulatory compliance for over 14 years. In December 2016, Mr Purchase published a book Real World Decision Modeling with DMN with James Taylor which addresses many of the topics discussed in more detail.

purchase@luxmagi.com   |   @JanPurchase   |   dmblog.luxmagi.com