THE PROJECT
A large Swiss retail bank needed a simple search service for bank customers that could be seamlessly integrated into various IT systems, such as the bank’s credit application system. The loan officer needed to be able to search by first name, last name, address and age in a single field. The service should also be able to find the right person even if the name has been misspelled, e.g. Akermann instead of Ackermann.
OUR CONTRIBUTION
abaQon was responsible for the architecture and implementation of the search service. Elastic Search was chosen as the search engine. Elastic search offers a fast and scalable search as well as a find-tuned relevancy, which was essential for our service to return meaningful result, even if no perfect match could be found. The elastic search sink connector provided by Confluent was used to load person data from the bank’s data source system Kafka into the elastic search database. A simple REST service was developed to handle the requests and the communication between the consumer and the elastic search cluster.

REALISED ADDED VALUE
The elastic search sink connector is able to load over a million of person documents within minutes from Kafka into the elastic search cluster. The elastic search cluster itself can search for persons based on various different attributes within milliseconds. The REST service on the solution stack makes it easier to integrate the search service into various customer journeys. The three components are highly performant and can be scaled up independently if the situation should require it.
Project Responsibility
