bespinian is Supporting CI Tech Sensors in Building a Central Banknote Data Management Platform


"bespinian provides a convincing mix of hands-on engineering, coaching, and consulting. This makes them the ideal partner and companion on our journey to the world of containers, microservices, and DevOps."


Lukas Burger, Head of Software Engineering



Customer

CI Tech Sensors AG is a development, manufacturing and service provider for banknote sensors, which are used in ATMs, automatic deposit machines as well as cash counting and sorting machines. CI Tech Sensors is a joint subsidiary of Diebold-Nixdorf and Giesecke+Devirent. The company has approximately 100 employees working at their location in Burgdorf, Switzerland.


Background

Large amounts of banknote data are required when developing banknote sensors and recognition software. Among this data, images play a central role in configuring and adapting banknote recognition algorithms and in the subsequent verification of these algorithms with respect to the required recognition rates. Beyond image data, other types of metadata are also required. This includes banknote serial and emission numbers, quality and usage information, as well as technical parameters regarding the device with which a certain banknote data item was recorded.


Project Goal

The goal of this project was to design and establish a central platform for all banknote data, to transfer all banknote data into the central platform and to make the data available for all tools and processes within CI Tech Sensors. Newly recorded banknote data was to be checked for validity before being stored with minimal redundancy. Each banknote record was to be made uniquely addressable. Furthermore, all banknote data was to be stored in an immutable way to guarantee strict reproducibility of all process steps involved in developing and validating banknote sensors.


bespinian's Role

bespinian played a key role in the following areas.


Microservice Architecture

In a series of regular architecture workshops, CI Tech Sensors and bespinian designed a microservice architecture for managing all banknote data. Using a step-by-step approach for introducing clearly scoped microservices, the central banknote management platform was introduced with minimal impact on the daily business of CI Tech Sensors. In doing so, bespinian took on conceptual responsibilities in developing the architecture as well as a hands-on role in its implementation.


Kubernetes Infrastructure

All of CI Tech Sensors’ microservices for banknote data management are deployed on a Kubernetes-based infrastructure. bespinian was in the lead for the configuration of all Kubernetes environments and for configuring the automated deployment of each service. Using a mixture of training, code-review and pair-engineering, bespinian enabled the CI Tech Sensors team to build up the necessary Kubernetes know-how and continue the operation and extension of the platform.


Monitoring and Alerting

The microservices for management of banknote data inherently play a critical role in CI Tech Sensors’ core business. Accordingly, these services needed to be closely monitored from an operational point of view. Furthermore, business metrics like, for example, the quantity of invalid banknote data also play a central role. Together with the CI Tech Sensors team, bespinian developed a monitoring and alerting architecture based on Prometheus and Grafana and took the lead in implementing the monitoring and alerting infrastructure and dashboards. While doing so, bespinian enabled the CI Tech Sensors team in the relevant technologies in order to ensure that monitoring and alerting could evolve together with the platform.


Transitioning to DevOps

With the introduction of Kubernetes-based infrastructure and a highly-available microservice architecture, the CI Tech Sensors team undertook a natural step towards a DevOps-way of working. bespinian guided CI Tech Sensors through this transition by coaching, and supporting the teams involved, as well as by establishing the necessary tooling and processes.


Technologies

  • Python / Flask
  • PostgreSQL
  • Redis
  • MongoDB
  • Kubernetes
  • Prometheus / Grafana / Alertmanager
  • Elasticsearch / Kibana