Writing
Posts about OpenTelemetry, observability, and building Telflo.
GreptimeDB gives you one database for metrics, logs, and traces. Telflo manages the OpenTelemetry Collectors that decide what lands in it. Here's why the two fit together and what the ready-made GreptimeDB flow does.
Scrape the Supabase Metrics API and catch the OTLP log drain with one collector. Every receiver, processor, and pipeline explained, with a ready flow.
AI agents can read telemetry almost anywhere now; nearly every observability vendor ships an MCP server for it. But managing your OpenTelemetry collectors is still a very manual process today, with no agentic operations. The Telflo MCP server and CLI give an agent that side too: edit collector YAML, validate it on a real binary, test it on sample data, and deploy it to your fleet.
Grafana is the visualization layer most teams already run, and ClickHouse keeps coming up as the telemetry store. Gigapipe connects the two: it stores logs, metrics and traces in ClickHouse and serves the Loki, Prometheus and Tempo APIs, so Grafana reads from it through its built-in datasources. We built a ready-made collector config for getting data in.
Writing the collector config is still the hardest part of adopting OpenTelemetry. Flows is our curated library of prebuilt, validated collector pipelines: about a dozen so far, one per use case, with more on the way.
Netflix, OpenAI, Anthropic, and Uber all store their telemetry in the same database, and so does nearly every observability product launched in the last five years. The reasons are mechanical. Here's how ClickHouse actually works, what it does to the bill, and when it's still the wrong call.
Turn the spans you already emit into RED metrics with the spanmetrics connector, and avoid the one mistake that quietly multiplies your time series into a five-figure metrics bill.
Open Telemetry Fleet Management
The cheapest place to cut your observability bill is the Collector, not the backend. A practical guide to dropping noise with OpenTelemetry, then flipping to full fidelity when an incident hits.
The agent + gateway collector pattern for Kubernetes. Every receiver, processor, and pipeline explained, with two ready templates.
OpenTelemetry deep dives, collector config patterns, and what's new in Telflo. Starting with a clean OTLP-to-multiple-backends setup.
