DataMatch Enterprise 10 Million Record Benchmark Methodology
Data Ladder ran a controlled internal benchmark comparing the DataMatch Enterprise web application (build 1.0.16) with the legacy application (build 3.9.3) on a 10,000,000-record dataset: same source file, same match definition, same machine.
What this benchmark measured
The test measured elapsed time for data import, data profiling and data matching. It was designed to compare the two DataMatch Enterprise application architectures under a consistent configuration.
It did not test matching accuracy or compare DataMatch Enterprise with another vendor's product.
Measured In scope
- Data import elapsed time
- Data profiling elapsed time
- Data matching elapsed time and completion
Not measured Out of scope
- Matching accuracy
- Comparison with other vendors' products
Same data. Same rules. Same machine.
Both applications were tested one at a time under identical conditions.
Test data and software
Hardware and test controls
How the test was run
Use the same 10-million-record source file and match definition for both applications.
Run one application at a time on the same machine.
Clear resources between sessions and begin with fresh imports without cached data.
Capture elapsed time from the Job Execution Centre in each application.
Repeat the controlled test to check whether the completion outcome is consistent.
Execution time, side by side
Elapsed times are rounded and reported as observed in each application's Job Execution Centre.
Data import
Data profiling
Data matching
| Operation | New web app | Legacy app | Observed result |
|---|---|---|---|
| Data import | About 4 min | About 5 min | Comparable Web app finished about one minute sooner |
| Data profiling | About 6 min | About 10 min | ~40% less time for the web app |
| Data matching | About 41 and 43 min | About 120+ mins | Completed Web app completed both runs; legacy app did not complete |
How to read these results
Import performance was comparable in this test. The bigger differences appear in profiling and matching.
Less profiling time
Profiling elapsed time fell from approximately 10 minutes to approximately 6 minutes, a reduction of about 40%.
The matching result should be read as a completion test rather than a direct speed ratio because the legacy application did not complete the workload. Under this configuration, the web application completed both recorded matching runs in less than 45 minutes.
What these results don't tell you
This was an internal Data Ladder benchmark, not an independent third-party test.
The comparison used one dataset, one match definition and one hardware environment.
Actual elapsed time will vary with dataset structure, data quality, matching rules, thresholds, infrastructure and allocated resources.
The test measured elapsed time and completion. It did not measure or compare matching accuracy.
Separate Data Ladder figures for 2-million-record processing, matches per second and matching accuracy are not part of this benchmark and should not be combined with these results.
Validate on your own data
Organizations should validate performance with representative records, matching rules and infrastructure before using these results for capacity planning. A useful proof of concept should measure:
Run your own proof of concept
Test DataMatch Enterprise with your records, your matching rules and your infrastructure.
