🔬 What's Novel
- WebAssembly as compilation target for database query operators with a stable ABI
- Portable query execution across server (Wasmtime), browser (native Wasm), and edge (CDN workers)
- Query shipping framework based on Wasm-compiled query plans sent to data locations
- Performance analysis of Wasm operators for data-intensive operations (within 2x of native)
🔧 Technical Approach
Phase 1 — Operator Library
Core operators (scan, filter, project, join, aggregate) implemented in Rust and compiled to WebAssembly modules. A stable ABI for data exchange between operators and the runtime.
Phase 2 — Query Compilation
Plan-to-Wasm compilation: operator selection from the library, operator fusion for pipeline efficiency, and serialization of compiled plans for network transmission.
Phase 3 — Runtime Environments
Three deployment targets: server-side execution via Wasmtime, browser-based execution via native Wasm support, and edge node execution on CDN worker runtimes.
Phase 4 — Optimization
SIMD utilization within Wasm, memory layout optimized for cache efficiency, and streaming evaluation patterns to handle result sets larger than available memory.
🧪 Hypotheses
Wasm query operators achieve performance within 2x of native execution while providing safety guarantees and portability.
Wasm operators can execute in browsers, enabling client-side query evaluation over locally cached data.
Wasm compilation enables query shipping to edge nodes, reducing data transfer compared to result shipping.
🔗 SkeinDB Integration
📚 Key References
- Neumann — "Efficiently Compiling Efficient Query Plans for Modern Hardware" (2011)
- Haas et al. — "Bringing the Web up to Speed with WebAssembly" (2017)