DIVID, die Einwegkunststofffonds-Plattform
The DIVID project is the digital platform of the Federal Environment Agency (UBA) for implementing the Single-Use Plastics Fund Act (EWKFondsG), through which Germany has transposed the extended producer responsibility under the EU Single-Use Plastics Directive into national law. The platform serves as the central point of contact for manufacturers, municipalities, and other eligible parties. Manufacturers of single-use plastic products are thereby involved in the costs of cleaning, collection, and disposal. The DIVID platform is developed iteratively on the basis of a modern technology stack with the guiding principles of modularity, microservices, and API-First, and enables the required digital processing of all processes from manufacturer registration and quantity reporting through to fund management and payment settlement for eligible groups.
- Merging functionally related microservices
- Automated data migration of distributed databases into a common schema
- Adaptation of the API Gateway
- Creating the CI pipeline in Azure DevOps
- Configuring code coverage measurement
- Rolling out SonarQube in an IaC pipeline
Re-Submitting Binary Dead Letter Queue Messages for Azure Service Bus
There are two interfaces for the Azure Service Bus. One is a Windows application called ServiceBusExplorer, originally developed for the Windows Service Bus. The other is the Azure ServiceBusExplorer, a web interface in Azure DevOps modelled on the original ServiceBusExplorer. Both interfaces allow individual messages in a dead letter queue to be re-sent. If the message is binary-encoded (e.g. via Protobuf), it is interpreted as text and thus rendered unusable. The task of the project was to extend both interfaces to enable the re-sending of individual messages.
- Reproduction of the problem in a simple example
- Creation of the feature request for the Azure ServiceBusExplorer
- Creating and discussing the issue for ServiceBusExplorer
- Implementing a solution for ServiceBusExplorer
Self-Service Payment Portal
Lowell had already implemented an online payment option via the consumer portal. However, since the registration process for the portal was complex and involved sending a letter, management wanted straightforward solutions for paying claims directly online. Within the project, several options were developed and implemented.
- Designing solutions together with the managing director
- Implementation of three concepts (Self-Service Payment Portal with postal code, payment link sent in the welcome letter, ad-hoc payment link for callers)
- Monitoring of collected payments
- Privacy-compliant design of the Self-Service Payment Portal
cLean – Debt Collection Software
With the cLean project, Lowell pursued the development of a unified debt collection software for all group companies. All claims of the DACH organization were to be processed as automatically as possible in a highly available system. A key focus was determining the possible scalability of the system at certain times when large volumes of claims had to be imported. C# was the primary programming language, but individual services were implemented in Kotlin and Go.
- Architecture of the microservices system
- Defining service boundaries along business domain lines (DDD)
- Partitioning data into aggregates
- Concept and design of a case reference number readable by both humans and machines
- Concept for an immutable accounting system using Event Sourcing
- Concept and implementation of the accounting component in compliance with business rules
- Research into settlement logic in Germany and Austria
- Implementation of business logic
- Realization of search queries across distributed systems (separate service using CQRS)
- Cross-team coaching, training, and support
API-First
Lowell follows an API-First strategy. The goal was to make the API available to external processes as well. To this end, a central API was provided fully automatically and used by all services. The API is subject to strict guidelines regarding design and backwards and/or forwards compatibility depending on the use case.
- Concept and creation of a central API-First repository based on Protobuf
- Automated generation of release and pre-release versions for multiple programming languages to optimize development speed
- Concept for API versioning
- Creating APIs
- Documentation of the API
- Reviewing APIs with a primary focus on clarity, documentation, and breaking changes
Migration of an On-Premise Kubernetes Cluster to Azure Cloud
Initially, services were operated in an on-premise Kubernetes cluster. The goal of the project was to migrate all services to Azure Cloud. Additionally, technologies and cloud services were evaluated, utility classes and templates were created for future development, and a CI/CD pipeline was established.
- Selection of cloud services to be used in coordination with the architect
- Agreements on future autonomy of development teams
- Review of existing code and resulting best practices
- Documentation of the Transaction Outbox Pattern, idempotency, aggregates, and other useful microservice patterns
- Utility classes for sending and receiving Protobuf business events
- Creation of the CI part of the CI/CD pipeline
- Evaluation and implementation of Azure Serverless
- Setting up infrastructure using Infrastructure as Code (IaC)
- Liveness check / readiness check for services in Kubernetes
- Creation of cron jobs in Kubernetes
- Creation of C# templates for building new services
Caller Verification
Consumers can call Lowell. At peak times, call volumes reach one call every two seconds. In collaboration with a provider of IVR systems (Interactive Voice Response), a system was created in which callers automatically verify themselves using their case reference number and personal data. After successful verification, they are transferred to a case worker, whose system automatically opens the correct file. The project focus was on the highest possible availability, as outages would have caused significant additional workload for case workers.
- Onboarding new Lowell subsidiaries
- Adaptations and extensions of the interfaces
- Definition of the API with a focus on compatibility
- Communication of the API to third-party providers
- Testing the interface locally, in the CI build, and in production (partially semi-automated)
- Rolling out the API with zero downtime in a Kubernetes cluster
- Analysis and resolution of production errors based on log entries
- Analysis of connection issues in the Akka.NET code
- Monitoring of incoming calls
Claim Data for the Consumer Portal
Lowell's consumer portal is a central digital service platform for individuals with outstanding claims at the debt collection agency. Lowell consists of multiple companies that are separated both organizationally and technically. Several different debt collection software systems had to be mapped onto a single API.
- Onboarding new Lowell subsidiaries
- Adaptations and extensions of the interfaces
- Definition of the API with a focus on compatibility
- Testing the interface locally, in the CI build, and in production (partially semi-automated)
- Rolling out the API with zero downtime in a Kubernetes cluster
- Analysis and resolution of production errors
- Reduction of the number of services, as the microservices had been cut too small
- Fixing security vulnerabilities and hardening API interfaces after penetration tests
- Guiding the development team to sustainably avoid identified weaknesses
- Defending the system architecture against external security auditors