The coexistence between enhanced Mobile Broad-band (eMBB) and Ultra Reliable Low Latency Communications (URLLC) is challenging in modern communication systems. To support such diversity, Multi-access Edge Computing (MEC) and Network Function Virtualization (NFV) emerge as complementary paradigms that shall offer fine-grained on-demand distributed resources closer to the User Equipment (UE). In this work, we address the combination of MEC, NFV and dynamic virtual resource allocation to overcome the challenge of resource dimensioning in the network edge. A Continuous Time Markov Chain (CTMC) based model was designed to evaluate how requests are managed by the virtualization resources of a single MEC node, with a primary focus on meeting the requirements of both eMBB and URLLC services. Practical factors such as resource failures, service prioritization, and setup (repair) times were integrated into the formulation. Some of our key findings include the idea that higher eMBB arrival rates decrease availability and increase response times, while URLLC availability remains stable, and that the container setup rates and failure rates substantially affect both availability and response times, with higher setup rates enhancing both availability and reducing response times.

Spectrum-based fault-localization tools, such as Tarantula, have been developed to help guide developers towards faulty statements in a system under test. These tools report statements ranked in order of suspiciousness. Unfortunately, the reported statements can often be unrelated to the error. This paper evaluates the impact of several approaches to ignoring such unrelated statements in order to improve the effectiveness of fault-localization tools.