Streaming Partner
LaserData

Real-time event streams,
flowing into LakeSail.

LaserData delivers ordered, low-latency event streams directly into LakeSail's Rust-native compute engine. Continuous data engineering and AI agent execution with no JVM in the path, no garbage collection pauses, and predictable tail latency end to end.

01stream02compute03consumeLaserDataApache Iggy / io_uringLakeSailLakeSail · Rust · Spark-compatibleJobs · Agents · Outputsfresh results, lower cost 01stream02compute03consumeLaserDataApache Iggy / io_uringLakeSailLakeSail · Rust · Spark-compatibleJobs · Agents · Outputsfresh results, lower cost
The Stack, in two words

Rust to Rust. No JVM in between.

0 ms GC pauses 0 JVM in the path Apache 2.0 at the core
About LaserData

Streaming, reimagined in Rust.

LaserData builds a next-generation streaming platform on Apache Iggy, the Rust-native message broker that uses io_uring for kernel-level async I/O. Ordered events move at predictable, low latency, without the JVM and garbage collection overhead of traditional brokers.

Traditional brokers carry JVM garbage collection through every operation, introducing tail latency spikes and unpredictable throughput at scale. LaserData eliminates that layer entirely, delivering predictable scheduling, tighter performance bounds, and reduced memory usage.

That philosophy lines up with how LakeSail is built. Together they form a stream-to-compute path where ordered events move directly from the broker into the engine, with no serialization detours and no runtime mismatch.

What LaserData brings

Streaming primitives, engineered for predictable performance.

01Foundation

Built on Apache Iggy.

The Rust-native streaming broker. Apache 2.0 license, open specification, fully portable. No proprietary protocols, no vendor lock-in.

2.0
Apache License
Rust
Memory safe
OSS
Open spec & source
02Kernel I/O

io_uring streaming.

Async I/O at the kernel. Predictable tail latency, minimal syscall overhead.

03Guarantee

Ordered event delivery.

Events stream into compute in the order produced. Continuous, no batch windows.

04Performance

No JVM. Ever.

Rust ownership replaces garbage collection. Predictable p99, lower memory, no JIT warmup.

Why LakeSail + LaserData

Stream to compute. End to end.

Two stacks built on the same foundation. Joining them produces a continuous, low-latency path from event to result with no JVM anywhere in the loop.

  1. Real-time input

    LaserData is the live signal feeding LakeSail.

    LakeSail consumes event streams from LaserData and reacts continuously, with no batch windows and no polling overhead between layers.

  2. One stack discipline

    Rust through and through.

    io_uring streaming and LakeSail's Rust engine share the same posture: zero GC pauses, no JVM anywhere in the path, predictable latencies at every layer.

  3. Agentic execution

    AI agents on a continuous compute layer.

    Agents subscribe to LaserData streams and delegate heavy work to LakeSail's elastic Rust runtime, which provisions compute per workload and releases it on completion.

  4. Open at the core

    Inspectable, portable, lock-in free.

    LakeSail is Apache-licensed and developed openly on GitHub. LaserData builds on Apache Iggy. Both stacks are yours to audit and to run anywhere.

Joint Stack

From event to result, in Rust.

LaserData delivers ordered event streams into LakeSail's Rust compute layer. Continuous, low-latency data engineering and agent workloads, with no JVM in the path.

01LaserDataEvent Stream02LaserDataReal-time Delivery03LakeSailDistributed Compute04LakeSailAgent Execution 01LaserDataEvent Stream02LaserDataReal-time Delivery03LakeSailDistributed Compute04LakeSailAgent Execution
Ready

Build the streaming
compute layer.

Run LaserData streams through LakeSail's Rust-native engine in your own cloud account. End-to-end Rust, open source, deployed where your data already lives.