Parseable

Parseable is an observability platform from Parseable Inc. that stores logs, metrics, traces and events as open Parquet files on object storage. It is written in Rust, ships as a single binary, and is available as open source (AGPLv3), a managed cloud service, and an enterprise edition. The GitHub project lists release v3.2.4 as the latest.

What It Is

Parseable is a column-oriented data lake built specifically for telemetry. Teams send data through logging agents, OpenTelemetry, Kafka, eBPF or other integrations, then query it with SQL or PromQL. Built-in features include alerting, dashboards, anomaly detection, APM, service maps and an error page, so separate tools are not needed for each signal type.

Architecture and Data Ownership

Compute is stateless and sits over object storage, so storage and compute scale independently. Data stays in open Parquet inside the customer's own cloud account. Deployment options include public or private cloud, VPCs and air-gapped networks. The vendor states that columnar compression reduces storage costs, and the GitHub description cites up to 90%.

AI and Agent Features

Parseable supports natural-language querying that produces inspectable SQL, and AI-assisted anomaly investigation and root cause analysis across logs, metrics, traces and events. It can also observe AI agents, with ingestion docs for OpenAI and Anthropic. Companion tools listed on the site include an MCP server for connecting AI agents to live telemetry, a Slack bot, PAI for Kubernetes auto-instrumentation, the PB CLI, and OTex, an AI observability engineer.

Setup Path

The quickstart downloads and runs the binary locally with a curl install script (a PowerShell script for Windows). A POST to the ingest API then sends events to a stream that can be viewed in the console on port 8000. Production guidance is in the installation docs, and Parseable Cloud offers a hosted alternative.