Benchmarking Ollama and vLLM for Concurrent LLM Serving: A Multi-Scenario Evaluation of Performance and Scalability | Synapse