At Criteo, we’ve relied on automatic aggregations for years. “Automatic aggregation” is the name we give to a system of recording rules that matches most metrics and removes certain dimensions, such as the instance emitting the metric, to reduce the cardinality (i.e., the number of metrics) and thus makes queries faster. What started as a workaround has become a key part of how we ensure backend stability and reliability at scale, with hundreds of millions of active metrics, all without requiring users to write a single recording rule. It also significantly reduces the cost of metrics storage. Internally, we call this approach zero-effort Observability, as most teams don’t have to write/maintain recording rules. In this talk, Raphael will explain how our approach to automatic aggregations has evolved over time and how we’ve adapted it to fit naturally into our Prometheus-based stack. He will share the different implementations we’ve tried, the lessons we’ve learned, and how our latest version takes advantage of recent improvements in Prometheus (new type label).

Prometheus has become the go-to standard for metrics-based monitoring, but as environments grow in complexity and scale, teams often find themselves hitting its operational limits, especially around cardinality and long-term storage. This talk explores how VictoriaMetrics builds on Prometheus fundamentals to offer a more scalable and efficient alternative for teams managing high-ingestion workloads and demanding retention needs, without abandoning the familiar Prometheus ecosystem. I’ll dive into how VictoriaMetrics supports Prometheus-compatible scrape configurations and exporters, allowing seamless integration with existing workflows. The session will showcase practical strategies for setting up and tuning scrape jobs, managing cardinality through label analysis and relabeling, and using VictoriaMetrics’ UI and tools to gain insight into metric usage patterns. This talk is tailored for advanced users eager to push the boundaries of Prometheus-based observability, demonstrating how the core philosophy of Prometheus can be extended and elevated through the integration of high-performance systems like VictoriaMetrics.

A single-node version of VictoriaLogs can handle hundreds of terabytes of logs. What if this isn't enough for you? Then the cluster version of VictoriaLogs comes to the rescue! It can scale to tens of petabytes of logs. This talk dives into the architectural details of the VictoriaLogs cluster, which explains how it achieves linear horizontal scalability for both data ingestion and querying paths. There is no magic - the cluster architecture is clear and quite simple. The talk also covers typical use cases for the VictoriaLogs cluster when a single-node version isn't enough.