Business Rule Engine Benchmarks
Since the inclusion of these benchmarks in a recent Infoworld article, we have received a great deal of feedback on the website and about the various rule engines. We will be updating these benchmarks based on this feedback in the coming months. In the meantime, all comments or questions are welcome. Also, a special thanks to James Owen and Mark Proctor for their comprehensive and useful feedback.
As rule engines are being used to solve increasingly complex problems, the importance of performance and efficiency becomes more important. Whilst there is no substitute for good rule architecture, these benchmarks objectively compare some of the most common rule engines and their performance.
The 'classic' benchmarks, Miss Manners & Waltz have become a bit dated, but are still useful comparisons. WaltzDB is the benchmark we now consider the standard for comparison, in addition to a few specialist tests (currently under development) that specifically stress things like alpha-node discrimination. These specialist tests are useful for companies that need the absolute maximum in performance.
Finally, it's important to remember that these benchmarks only compare rule engines that implement true inferencing, also known as predicate logic (see a Wikipedia definition), as opposed to those that implement only propositional logic (Wikipedia definition).
Rule engines that implement only propositional logic can solve useful problems, predicate logic (the type implemented by Rete) is required to solve certain classes of problem, such as actuary. A full explanation of the differences can't be covered here but there are many references available that can explain this.