FiloDB

What FiloDB is

FiloDB is a free, open-source distributed database for real-time time-series, event, and operational data, with Prometheus query support. It is designed to ingest and query many millions of time series by sharding data across processes and providing distributed queries. Columnar compression and off-heap memory management help it keep substantial data in memory while reducing garbage collection. PromQL queries are available through an HTTP API and CLI; the README describes its PromQL support as about 60%. Histograms are stored as a first-class column type, which the project says can improve compression and query speeds by up to two orders of magnitude. A typical real-time ingestion setup reads directly from Apache Kafka, with cluster processes coordinated through Akka Cluster. Its Gateway accepts metrics over TCP using Influx Line Protocol, though the only tested setup uses Telegraf with a Prometheus endpoint source. The README describes use as a Grafana data source through the Prometheus-compatible API and identifies Cassandra as its persistence store. FiloDB runs as standalone processes in a cluster and requires Java, Cassandra, Kafka, Rust, a C compiler, and SBT to build. The project labels the Gateway experimental and Spark deprecated.

Who it suits

FiloDB suits teams operating time-series, event, or operational data workloads that can manage a self-hosted cluster. Its documented use cases include operational metrics, distributed tracing storage, application-metric debugging, and real-time event storage and querying.

Limits

PromQL support is described as about 60%. The Gateway is experimental, while Spark is deprecated; the project lists heavily transactional update-oriented workflows and OLAP or analytics as anti-use-cases.

Getting started

1. Visit https://github.com/filodb/FiloDB.
2. Prepare Java, Cassandra, Kafka, Rust, a C compiler, and SBT to build.
3. Deploy FiloDB as standalone processes in a cluster.
4. Configure real-time ingestion from Apache Kafka, or use the Gateway for Influx Line Protocol metrics.
5. Query with PromQL through the HTTP API or CLI.

Verdict

Choose FiloDB if you need a self-managed, distributed time-series database for operational metrics, tracing, or real-time events and can work with its listed prerequisites. Look elsewhere for heavily transactional update-oriented workflows or OLAP and analytics, which the project identifies as anti-use-cases.