We joined client's teams in two areas: first is working on code refactoring, system maintenance and building new features. The other is developing a critical part of the system–infrastructure of agents that drills into the customer's environment, and collects footprints to send them back to the main system, which decides on testing relevance. The Clurgo team works with developers on a client's side on a daily basis, influencing critical improvements in the areas of test optimization, software quality data collection, integration testing, and new features development. Let's look at some examples of our significant input.

The solution of automated CD Agents helped to accelerate and simplify the procedure. System setup is currently limited to the configuration of the applications to be tested. Everything that was previously done in CI/CD processes is automatically discovered by the CD Agent (resolve, discover), and minimal manual configuration is required. This approach provides users with rapid deployment of 90% of the functionality without the need for complex changes to CD processes. The missing 10% are just nice-to-have features, not critical for the entire solution relevance. The solution was designed with reverse compatibility and performance requirements in mind. Before, some of the items performed by CD Agents were executed by running the tested system. We optimized the solution and implementation to ensure high performance of the client’s application.

The system’s architecture consists of agents for Java, .NET and Node.js that communicate with the backend. The client attaches agents on the CI/CD and tested application process, initiates tests and listens to the process execution, then passes information to the backend. The footprints arrive at the backend, where they are analyzed and preliminarily transformed before being stored in AWS S3. This process was suboptimal as agents may generate a high volume of footprints, resulting in high S3 maintenance costs. We designed a solution that optimizes the process and minimizes the number of data sent: instead of multiple records generated by agents, we send a single, consolidated file. Benefits? Substantial cut on the cloud and infrastructure costs.