Slimmetry

Self-hosted event collection and observability for Ruby applications. Ship traces, business events, and custom metrics to a single queryable store you own — no SaaS contract, no per-event billing, no data leaving your infrastructure.

What you get

• A pipeline query language — filter, group, aggregate, and bucket time-series data with a Splunk/Kusto-style syntax that reads top-to-bottom.
• Dashboards with tiles and parameters — build shareable views across multiple streams, mix line/area/bar charts, and parameterize with dropdowns backed by live queries.
• Built-in trace support — every event can carry a trace_id and parent_trace_id, so you can stitch together request flows without a separate APM tool.
• Bearer-token API — ingest from anywhere, query from anywhere, with per-stream ingestion tokens and account-level read/write tokens.

Built for serious throughput

Per-stream concurrency keys keep one noisy producer from starving the rest. On a single container Slimmetry sustains 3,000 events/sec at sub-150 ms p95 and bursts to 9,400 events/sec under load. A million events drain in under two minutes, and the staging backlog returns to zero between flushes.

Designed to be small

Slimmetry runs on a single node. Events stage to SQLite for fast writes, then flush to immutable Parquet files queried in-memory by DuckDB. There is no Kafka, no Elasticsearch cluster, no Postgres — just Docker and a disk. Configurable rolling retention per stream (default 30 days).

Pairs with the Slimmetry Ruby gem

Drop the slimmetry gem into your Rails app to auto-trace methods, controllers, and jobs. Events stream to your Slimmetry server in batches, with negligible overhead.

Who it is for

Rails developers and small teams who want real observability — traces, dashboards, ad-hoc queries — without the SaaS bill or the operational weight of a full ELK/Prometheus stack. If you can run Docker, you can run Slimmetry.

One container. One disk. Nine thousand events per second. Your data, your queries, forever.