Store metrics. Query insights.
Scale without limits.
Arc is a high-performance time-series data warehouse for engineers building observability platforms, IoT systems, and real-time analytics. Built on DuckDB and Parquet with 2.42M records/sec columnar ingestion, SQL analytics, and flexible storage options for unlimited scale.
git clone https://github.com/basekick-labs/arc.git
cd arc && ./start.sh nativeFastest time-series database
Arc outperforms VictoriaLogs (3.3x), QuestDB (6.5x), Timescale Cloud (18.2x), and TimescaleDB (29.7x)
Performance and simplicity through proven components
Real-time analytics requires massive streams of time series. Arc delivers high-throughput ingestion, powerful SQL analytics, and seamless storage integration.
MessagePack Columnar Format
2.33M records/sec with zero-copy passthrough. 2.57x faster than row format, 9.7x faster than Line Protocol.
DuckDB Analytics Engine
Columnar SQL engine. Tested on ClickBench: 36.43s cold run on 99.9M rows. Direct Parquet reads.
Automatic Parquet Compaction
Merges small files into 512MB ZSTD-compressed Parquet. 10-50x faster queries. Snappy for ingestion.
Write-Ahead Log (WAL)
Optional durability: fdatasync, fsync, or async modes. Zero data loss on crashes.
S3-Compatible Storage
MinIO, AWS S3, GCS support. Unlimited scale with REAL separation of compute and storage.
REST API & Multi-Database
InfluxDB 1.x/2.x compatible. Multi-database namespaces. JSON and MessagePack endpoints.
Built for the most demanding workloads
Arc powers observability platforms, IoT systems, and real-time analytics for engineers who need high performance, unlimited scale, and flexible deployment options.
IoT & Industrial Monitoring
Ingest millions of sensor readings per second from MQTT, Telegraf, or HTTP. Store years of telemetry with automatic compaction (80% storage savings) and query historical trends in seconds.
Observability Platforms
Build next-generation monitoring with multi-database architecture, Apache Superset dashboards, and Grafana integration (coming soon). Import existing data from InfluxDB, TimescaleDB, or Kafka topics.
Real-Time Analytics
Query billions of rows in seconds with DuckDB SQL engine. Enable query caching for sub-millisecond responses. Deploy on local NVMe, MinIO, S3, GCS, or any object storage.
High Performance
2.33M records/sec ingestion with MessagePack columnar format. Fastest time-series database with 36.43s cold run on ClickBench. p50: 6.2ms, p95: 37.9ms, p99: 55.4ms latency.
Automatic Compaction
Merge small files into optimized 512MB chunks. Achieve 80% storage savings and 10-50x faster queries automatically.
Zero Data Loss (WAL)
Optional Write-Ahead Log for finance/healthcare compliance. Zero data loss on crashes. Maintains high throughput even with durability guarantees.
InfluxDB Compatible
Drop-in replacement for InfluxDB 1.x/2.x. Telegraf ready. Import existing data from InfluxDB, TimescaleDB, or Kafka topics.
Open Format, No Lock-in
Standard Parquet files on S3/MinIO/GCS/local. Your data remains accessible outside Arc.
Multi-Database Architecture
Organize data by environment, tenant, or application. Query across databases with SQL joins and cross-database analytics.
Query Caching
Configurable result caching with TTL for sub-millisecond repeated queries. Perfect for dashboards and real-time monitoring.
Flexible Storage Options
Deploy on local NVMe, MinIO, S3, GCS, or any S3-compatible storage. Scale from edge devices to petabyte deployments.
Rich Integrations
Visualize with Apache Superset. Connect Telegraf for metrics collection. Grafana integration coming soon.
Build your next observability platform with Arc
Deploy in minutes with Docker or native mode. Ingest millions of metrics per second, query millions of rows in seconds, and scale from edge to cloud.