Ab Initio includes extensive simulation and testing capabilities that make it easy for business experts to test their rules.
Non-developers having the ability to author their rules is not enough — simulation and testing are also required to unlock the real self-service benefits. Simulation and testing are built into the core of Ab Initio’s rules environment. A user can author rules using a familiar spreadsheet-like model, and then, using real or simulated data, they can see exactly how every expression reacted to the data provided, with full traceability. The user can easily understand why a rule was successful or not for a given set of data, and then step graphically through the data one record at a time.
Ab Initio’s high-performance processing allows users to run a large set of data through the rules — for example, to simulate yesterday’s production run for a risk or fraud application. Ab Initio tracks what happened to every expression during that run and compares the result to a provided baseline result set — that is, it automatically performs a full regression test. A user can quickly and easily perform “what-if” simulations (or back-tests) to understand the impact of a rule change. Although these tests can be performed on real data, the rule changes are kept within the simulation environment, so there is no risk to actual production initiatives. The experts can ensure the quality of their rules and promote them with confidence to production on the same day.
A global retailer developed their software at multiple locations around the world. Unfortunately, creating valid, anonymized datasets for testing was becoming a major headache.
A global retailer’s large software development projects involved development teams from around the world. Software testing requires valid data, and they had to test against the content of their databases — that’s the data that they’ll be running against in production. However, that data could contain personally identifiable information (PII). Transmitting customer PII around the world would quickly run afoul of many countries’ privacy regulations — this problem can really complicate the testing process.
This is where Ab Initio came in.
The standard solution to handling PII is to anonymize the data used for testing, which isn’t as simple as it sounds. Sure, it’s possible to simply overwrite customer names with “X” or insert random numbers in place of an ID number, but the end result is no longer valid data. Without valid data, it’s not possible to fully test the new software. Even replacing customer names with fake names is tricky: if “John Smith” maps to “Fred Block” in one location and “Ivan Tadeov” in another, joins on customer names can’t be tested. Even the fact that one obfuscation maintains the same number of letters and one does not can mess up the testing process.
The retailer was already using Ab Initio software to integrate different systems and platforms, including RDBMS from multiple vendors, software applications from third-party providers, and different real-time messaging technologies. It was a small step to take advantage of Ab Initio’s anonymization and test data management capabilities. Ab Initio software allowed the data to be anonymized yet remain valid for testing. No matter which team was doing the work, “John Smith” would always be “Fred Block.”
With Ab Initio software, the company anonymized their entire multi-terabyte data warehouse in a stunningly short time. They then created meaningful small (250-GB) subsets that could be easily and quickly transmitted to development teams worldwide.
Thanks to Ab Initio, the retailer could develop, test, and run their software anywhere in the world — and effortlessly maintain full customer confidentiality and anonymity of personal information.