Grid telemetry, PMU data, sensor networks.
Arc is the columnar analytical database for utilities, grid operators, and energy companies who run on high-frequency sensor data and need long-horizon history for operations, planning, and compliance. Open Parquet. Sovereign storage. Standard SQL.
Single Go binary. Open Parquet. Standard SQL.
The shape of the problem
The modern grid generates more telemetry than the systems built to hold it. Phasor measurement units at 30 to 60 hertz across thousands of points. SCADA at every substation. Smart meters at every endpoint. Renewable assets reporting weather, inverter, and output state continuously. Each system was usually purchased separately, with its own historian, its own format, its own query interface.
Cross-system questions are slow. Compliance questions are slower. NERC CIP and regional regulatory windows require long retention with chain-of-custody, and the legacy historian licenses by tag, so retaining more data quietly costs more every year.
Sustainability and renewables introduce another data layer entirely. Yet another database, yet another silo.
Arc is the unified data layer. Ingest from every system, every protocol, every site. Hold years of high-frequency history as compressed Parquet on storage you own. Query across all of it with standard SQL.
Built for the workloads utilities actually run
Grid and substation telemetry
PMU data at 30 to 60 Hz, SCADA at every substation, transformer monitoring, line sensors. Arc ingests at sustained rates up to 19.9M records/sec on a single instance, partitioned by time, queryable about 100 milliseconds after it lands. Cross-substation queries that take hours on legacy historians return in seconds.
Smart metering at fleet scale
AMI deployments produce continuous endpoint data across millions of meters. Arc handles the volume natively and treats meter ID, region, tariff, and circuit as ordinary column data, not as an exploding tag index. Fleet-wide queries like load by feeder over the last year, or anomaly detection across a region, are routine.
Renewables and distributed energy resources
Solar, wind, battery storage, and DER all generate their own time-series streams. Arc unifies them with grid telemetry, so the same SQL query joins generation, load, weather, and storage state across the full fleet.
Compliance, retention, and chain-of-custody
NERC CIP, regional regulatory frameworks, and customer audits all require long retention of operational and event data, often with strict access controls. Arc holds years of full-resolution history as open Parquet on storage you own, with auth and audit logging on every query. The data, the access record, and the retention policy are all in one place.
Long-horizon planning and forecasting
Capacity planning, demand forecasting, and renewable integration modeling all run on multi-year historical data. Arc keeps the archive on cheap object storage and makes it queryable through the same SQL surface used for live operations. Data scientists and the operations team use the same store.
What Arc gives a utility
Sovereignty and sovereign cloud support
Utilities operate under regulatory regimes that often require data to stay within national borders, within specified clouds, or fully on-premise. Arc runs as a single binary on any infrastructure you choose, with storage on any object store or filesystem you control. No SaaS dependency, no required cloud account, no hosted service.
Years of history at object-storage prices
Arc stores data as ZSTD-compressed Parquet, 5-7x smaller than raw. A decade of PMU and SCADA data costs object-storage prices, not premium-historian-per-tag prices. The cost pressure to downsample or shorten retention disappears.
One data layer, every protocol
Arc ingests via MQTT, OPC UA bridges, Telegraf, line protocol, and standard HTTP APIs. Whatever your existing SCADA, EMS, historian, or renewable monitoring system can export, Arc can ingest. One place to query, instead of one place per vendor.
Standard SQL, not a vendor DSL
Arc speaks standard SQL through DuckDB. Operations engineers, data analysts, and planning teams who know SQL are immediately productive. No proprietary tag-query language, no specialist required to extract a regulatory report.
Edge to control center
One Go binary at a substation, the same binary at a regional operations center, and the same binary in the central data plane. Same database, same SQL, same Parquet.
No tag licensing
Adding a new sensor, a new substation, a new renewable site does not add a per-tag fee. The cost of expanding observability does not scale with the count of points monitored.
Why utilities choose Arc over the alternatives
Legacy historians (PI System and equivalents) are deeply embedded and increasingly expensive. Per-tag licensing, proprietary formats, and forklift upgrades to add new programs make scaling painful. Arc is one binary, one license, and stores everything as open Parquet you own. Operational data leaves the closed system.
Cloud data warehouses were built for BI on settled data, not for streaming PMU telemetry. The cost model breaks at utility data rates. Arc keeps the data on infrastructure you already control and queries it without egress fees or per-query meters.
Tag-indexed time-series databases hit a cardinality wall once you tag with meter ID, feeder, substation, and tariff. Arc treats high-cardinality data as ordinary column data.
Build-it-yourself DuckDB + Parquet is exactly what Arc is. DuckDB plus the streaming ingest from every utility protocol, autonomous compaction, retention, replication, governance, and operational tooling. In one binary.
What this looks like in production
Arc is running in production utility telemetry workloads today. High-frequency sensor data, multi-year archives, sovereign on-premise and sovereign cloud deployments. SQL across the full operational and historical data plane.
AGPL-3.0 with a commercial license for organizations that need it.
Get started
See it run on telemetry workloads
Our live demos show real Arc instances on sensor and asset data.
Get Arc
Arc is open source and ships as a precompiled single binary. Pull it from the download page as Docker, Helm, or a .deb, and point your existing MQTT, OPC UA, or Telegraf pipeline at it. Source on GitHub.
Talk to us about your historian
For utilities looking to modernize or unify operational data across multiple systems, we run a discovery to scope ingestion, sovereignty, retention, and compliance.