Velostream
42by velostreamai
Velostream is a high-performance streaming SQL engine built in Rust, designed for real-time analytics on financial and time-series data. It supports pluggable formats (JSON, Avro, Protobuf) and emphasizes low-latency processing, precision, and seamless integration with modern data pipelines
Getting Started
README
Velostream
High-performance streaming SQL engine written in Rust. Velostream provides real-time data processing with pluggable serialization formats (JSON, Avro, Protobuf), optimized for financial analytics use cases.
Features
- Streaming SQL — Tumbling, sliding, and session windows with continuous and emit-mode aggregations
- Stream-to-stream joins — Windowed joins across multiple Kafka topics
- Materialized tables — CTAS (CREATE TABLE AS SELECT) with point lookups and stream-table joins
- Pluggable serialization — JSON, Avro, and Protobuf with schema registry support
- Financial precision — ScaledInteger arithmetic (42x faster than f64, exact decimal results)
- Observability — Prometheus metrics, OpenTelemetry traces, distributed trace propagation
- Multi-job server — Run multiple SQL pipelines in a single process via
velo-sql
Quick Start
# Build all binaries
cargo build --release
# Run a SQL pipeline
echo "CREATE STREAM trades_agg AS
SELECT symbol, SUM(quantity) as total_qty, AVG(price) as avg_price
FROM trades
GROUP BY symbol
WINDOW TUMBLING(INTERVAL '1' MINUTE)
EMIT FINAL;" | ./target/release/velo-sql --file -
Binaries
| Binary | Description |
|---|---|
velo-sql |
Multi-job SQL server — deploy streaming pipelines from .sql files |
velo-cli |
Interactive CLI for managing running jobs |
velo-sql-batch |
Batch SQL execution for testing and CI |
Related Projects
| Project | Description | License |
|---|---|---|
| velo-test | SQL application test harness with Kafka testcontainers | Proprietary |
| velo-studio | AI-powered SQL Studio web application | Proprietary |
Documentation
- SQL Grammar — Complete EBNF grammar and AST reference
- SQL Examples — Working examples for all query types
- SQL Annotations —
@job_mode,@metric,@observabilityreference - Kafka Schema Config — Avro/Protobuf schema configuration
- Benchmarks — Performance testing methodology and scripts
Development
# Run unit tests
cargo test --lib --no-default-features
# Run comprehensive tests (excludes integration/performance)
cargo test --tests --no-default-features -- --skip integration:: --skip performance::
# Pre-commit checks
./run-commit.sh
See CLAUDE.md for the full development guide.
License
Licensed under the Business Source License 1.1. Production use is permitted except for competitive hosted/embedded offerings. Converts to Apache 2.0 four years after each version's publication.