Kolloquiumsvortrag: 2. Dezember 2025, Oliver Kardos (Betreuer: Hielscher)
Impact of down scaling the two most important resources of a network element controller on microservices based software
This thesis investigates microservices software performance under CPU frequency and core count downscaling. It aims to understand behavior on weaker computers, predict performance, and propose countermeasures. Addressing a little-researched field of computer science, the study utilized Nokia’s network element controller, a real-world software solution, and the Opentelemetry demo webshop for validation. The performance is measured with Opentelemetry observability methods across typical use cases of both software solutions.
Key findings reveal that microservices performance degradation is proportional to CPU frequency decrease. For core removal, it specifically depends on the shares of parallel and sequential execution parts of the software. Here, the performance prediction is challenging but more accurate for systems with two or more cores. Additionally, weaker computers were found to be susceptible to timeouts and message queue overflows. These findings make a significant contribution to the work of software architects, as they enable them to appropriately dimension microservices for resource-constrained environments. This includes removing microservices and functionality, correct queue dimensioning, and appropriate timeouts. Future work is suggested to explore performance on also RISC-V architectures and the impact of smaller main memory.
Zeit: 10:15 Uhr
Ort: Raum 04.137, Martensstr. 3, Erlangen
oder
Zoom-Meeting beitreten:
https://fau.zoom-x.de/j/68350702053?pwd=UkF3aXY0QUdjeSsyR0tyRWtLQ0hYUT09
Meeting-ID: 683 5070 2053
Kenncode: 647333