Overview
Arc is an open-source time-series database designed for high-volume telemetry and long-term, full-resolution retention. It ingests data through the InfluxDB Line Protocol, stores records as open Parquet files, and provides a single SQL interface across recent and historical data.
The database runs as a single Go binary without required external dependencies. Storage can remain on infrastructure controlled by the user rather than inside a proprietary database format.
Key Features
- Queries telemetry with PostgreSQL-compatible SQL through an embedded DuckDB engine, including joins, common table expressions, and window functions.
- Stores portable Parquet data on object storage or local disk, with support described for S3, Azure Blob, MinIO, and other compatible environments.
- Keeps live and archived telemetry accessible through the same query surface, avoiding separate hot and cold data systems.
- Supports Telegraf-compatible ingestion and high-cardinality identifiers used in machine, fleet, mission, and test data.
- Uses Parquet compression and compaction to reduce the storage footprint of raw telemetry.
Deployment
Arc can run locally in a container, as a native package, or in Kubernetes through Helm. Its lightweight, single-binary design supports deployments ranging from remote test stands and edge systems to cloud and on-premises infrastructure.
The open storage format also allows Parquet-aware analytical tools to read the underlying data without requiring a database export.
Use Cases
Arc targets telemetry-heavy workloads such as observability, IoT, product analytics, fleet tracking, aerospace testing, and industrial monitoring. Teams can use the same retained data for live dashboards, anomaly investigation, post-test analysis, historical comparisons, and simulation inputs.
For InfluxDB users, support for its line protocol and standard SQL provides a migration path that preserves access to detailed historical readings.