Tag
Los cuadros de mando con IA llevan un par de años prometiendo detección de anomalías mágica y causa raíz automática. La realidad es más modesta pero también más útil, si se sabe separar el ruido del valor real. Repaso honesto de qué funciona y qué no.
After a decade of Prometheus, three years of consolidation around OpenTelemetry, and the open stack now mature with Grafana, Loki, and Tempo, concrete recommendations for teams starting or reviewing their observability layer: what fits, what is excess, and what to avoid.
The combination of Parca for continuous profiling, Beyla for eBPF auto-instrumentation, and Grafana as the visualisation layer delivers deep observability without touching code. A look at how the three pieces fit together and where the limits still show.
Grafana Beyla is an eBPF agent that automatically instruments existing applications without touching their code: it observes kernel syscalls and generates OpenTelemetry traces and RED metrics for services written in Go, Java, Python, Node, and Rust. It gives broad, immediate coverage, but it does not replace the manual SDK for business metrics and internal logic.
Loki indexes only labels, not log content, which cuts storage costs dramatically compared to Elasticsearch. The main production risk is cardinality explosion each unique label-value combination generates a stream that inflates the index and slows queries. Separating read and write paths ensures a heavy query cannot saturate ingestion.
The Grafana stack combines three open source projects: Loki for logs, Tempo for traces, and Mimir for metrics. All three keep data in object storage (S3/GCS) with a minimal index instead of indexing everything like Elasticsearch, which cuts cost sharply at high volume and lets you correlate metric, log, and trace from a single Grafana panel.
