Engineering 2 Sep 2026
How we ingest 1.2M metric points per second on commodity hardware
Inside the TraceLoop storage engine: columnar compression, pre-aggregation and the write path we rebuilt three times.
read post →Resources
Documentation engineers actually finish reading.
Quickstarts measured in minutes, a query reference with 120+ runnable examples, an alerting handbook written by people who have been paged at 3am — and the engineering blog where we explain how TraceLoop itself works.
Documentation
Start here
Quickstart
Ship your first trace in under five minutes with the TraceLoop agent or OpenTelemetry SDK.
Guide · 5 min read →
Agent Reference
Configuration, sampling strategies, tail-based sampling and resource attributes for every supported runtime.
Reference · updated weekly →
Query Language (TQL)
The TraceLoop Query Language: filter logs, aggregate metrics and stitch traces with one composable syntax.
Reference · 120+ examples →
Alerting Handbook
SLO-based alert design, burn-rate windows, escalation policies and on-call rotation patterns.
Handbook · 42 pages →
Guides & Best Practice
Field guides from the on-call trenches
Migrating from legacy APM stacks
A pragmatic path off per-host licensing: dual-write with OTel, validate parity, then cut over.
Migration guide →
Designing SLOs engineers trust
How to pick SLIs, set error budgets and wire them to alerts that never cry wolf.
Best practice →
Tail-based sampling at scale
Keep every interesting trace, drop the noise — a worked example at 900k spans/second.
Deep dive →
Incident retros that stick
Blameless post-mortem templates and the TraceLoop timeline export that writes half the report for you.
Templates →
Engineering Blog
How we build TraceLoop, in public
Architecture 19 Aug 2026
The case for logs, metrics and traces sharing one clock
Why correlation IDs are only half the story, and how monotonic timeline alignment changes incident review.
read post → Product 4 Aug 2026
Anomaly detection without the PhD
Our seasonal-baseline detector explained in plain English, with the false-positive budget we hold it to.
read post → Open source 21 Jul 2026
OpenTelemetry is the winning bet — and what we add on top
We are all-in on OTel wire formats. Here is the honest list of where our platform does more than raw collectors.
read post →Platform Status
TraceLoop's own observability, publicly
We run our production estate on TraceLoop and publish the same status view our customers see. 99.99% availability, measured and posted honestly.
platform status
all components · updated 12s ago
-
Telemetry ingest (eu-west-2)
Operational -
Telemetry ingest (eu-west-1)
Operational -
Query & search
Operational -
Alerting & routing
Operational -
Workspace UI & API
Operational -
Integration delivery
Degraded — filelog agents (legacy hosts)
uptime · 90 days
monthly availability · SLO 99.95%
July 202699.995%
August 202699.991%
September 202699.998%
Subscribe to status alerts from any incident feed.
From docs to production in one afternoon
Read the quickstart, wire up a collector, and ask our engineers anything that the docs don't answer.