Observability in Action: Dashboards, LLMs & Out Run

Three talks and live demos on observability, hosted by Grafana & Friends Manchester at bet365's Manchester office. Mihaela Maior, Director of Engineering at Grafana Labs, on building one dashboard that serves a whole team instead of ten conflicting copies - filters, multi-property variables and variables scoped at instance, folder and dashboard level. Ben Kaimogul (Senior Infrastructure Engineer, Callcare) and Ruth Kegicha put a RAG application on vLLM and NVIDIA T4 GPUs under increasing load with Prometheus, Grafana and k6, and show how to tell a busy system from a saturated one. Simon Prickett of Grafana Labs instrumented the 1986 arcade game Out Run with the OpenTelemetry C++ SDK and pipes the traces into Tempo and Loki. Free. Food from 17:45, talks from 18:15. The host team checks you in downstairs at the Zenith Building. Why go: the GPU load-testing talk puts real numbers on where an LLM service breaks, which almost no meetup does. Best for SREs, platform engineers and anyone running inference in production.