Observability Lessons NVIDIA and CoreWeave
NVIDIA now ships VictoriaMetrics as the default metrics backed in their self-managed Cloud Functions, and CoreWeave uses VictoriaMetrics for infrastructure monitoring and AI capabilities.
The VictoriaMetrics Observability Blog
Author
- Adam Yates
- Aditya Kulkarni
- Adrian Bridgwater
- Agustín Gallego
- Alex Woodie
- Alexander Marshalov
- Aliaksandr Valialkin
- Aman Agarwal
- Antony Savvas
- Artem Navoiev
- Bill Tanner
- Cer6erus
- Claudio Masolo
- David Marshall
- Denys Holius
- Derek Foster
- Diana Todea
- Dima Lazerka
- Dmytro Kozlov
- Emily Foster
- Fintech Herald
- Fred Navruzov
- Gary Flood
- Gonzalo García Labat
- Ivan Yatskevich
- Jaime Hampton
- Jan Sokol
- Jason Bloomberg
- Jason English
- Jean-Jerome Schmidt-Soisson
- Jesús Espino
- Joab Jackson
- John Seekins
- Jose Gomez-Selles
- Julien Menan
- Karan Virdi
- Laveesh Kocher
- Leigh McGowran
- Lindsay Clark
- Marc Sherwood
- Mark Baker
- Martijn Van Best
- Martin Veitch
- Mathias Palmersheim
- Michelle Sebek
- Nick Gibson
- Nikolay Khramchikhin
- Pablo Fernandez
- Phuong Le
- Rafal Szypulka
- Renato Losio
- Richard Speed
- Roman Khavronenko
- Vadim Alekseev
- Vadim Rutkovsky
- Yurii Kravets
- Zakhar Bessarab
- Zhu Jiekun
No matching authors found.
Category
- Benchmark
- Community
- Company News
- Customer Stories
- Developer Experience
- Distributed Tracing
- Events
- Go
- Go @ VictoriaMetrics
- High Cardinality
- Kubernetes
- Monitoring
- Observability
- Open Source Tech
- OpenTelemetry
- OTLP
- Performance
- PostgreSQL
- Product News
- Tech Talk
- Time Series Database
- VictoriaLogs
- VictoriaMetrics
- VictoriaTraces
No matching categories found.
From the article
Blog posts
Observability Lessons NVIDIA and CoreWeave
NVIDIA now ships VictoriaMetrics as the default metrics backed in their self-managed Cloud Functions, and CoreWeave uses VictoriaMetrics for infrastructure monitoring and AI capabilities.
What Is Observability? It’s More Than Metrics, Logs, and Traces
Observability is a practice, not a product. Learn how metrics, logs, traces, and OpenTelemetry help engineers investigate problems and understand their systems.
Go, Logs and the Engineering Behind VictoriaLogs Performance
Tech Talk #15: Jesús Espino explains how architecture, data locality, Bloom filters, and Go help VictoriaLogs deliver efficient log storage and fast queries.
VictoriaLogs Has Now a Long Term Support (LTS) Release
VictoriaLogs v1.52.1 is our first Long Term Support release. Enterprise users get 12 months of guaranteed security and bug fix backports, with no new features or breaking changes.
VictoriaMetrics September 2026 Ecosystem Updates
September release roundup brings new releases across the VictoriaMetrics ecosystem, plus the launch of VictoriaTraces in Cloud, completing our managed observability offering of metrics, logs, and traces stack.
The Data Race That Wasn't a Bug (and the One That Was)
You turn on Go’s race detector and get a data race inside net/http that makes no sense at first. This post digs into how Go’s HTTP client really sends a request body, why it can keep reading your buffer after Client.Do has returned, why we decided that race is harmless in vmagent, and why the same behavior could corrupt data in vmauth, so we fixed it there.
The Year-2 Price Cliff: What Your Observability Stack Really Costs Over 3 Years
Why aggressive year-1 discounts in observability often lead to 2–3x renewal bills, plus a simple 3-year TCO framework and five questions to ask before signing.
VictoriaTraces in VictoriaMetrics Cloud: Complete Observability is Here
Announcing VictoriaTraces in VictoriaMetrics Cloud: managed distributed tracing with native OpenTelemetry support, LogsQL for powerful trace analysis, and integrations with Grafana and Jaeger for complete observability, offering predictable pricing and operational simplicity without the overhead of managing a tracing backend yourself.
How Linux readahead works, and the two ways to change it
The kernel reads your files before you ask for them, guessing from where your reads land. Here is how that guessing actually works (the window, the marker, the background fetch) and the two ways to influence it: read_ahead_kb on the device, and posix_fadvise from inside the program. Neither does quite what its name suggests.