Performance & Compliance Benchmarks

Evaluated under heavy concurrent OLTP transactions using standard HammerDB TPROC-C profiles. Delivers sub-millisecond p99 commit latency with linear throughput scaling across symmetrical active-active nodes mounting shared NVMe block storage.

Demonstrated scale-out efficiency exceeding 1.70x write throughput and 1.90x read throughput across dual-node active-active configurations.

Executes all 43 ClickBench analytical queries directly over the primary transactional storage engine without ETL pipelines, leveraging Apache DataFusion and Apache Arrow vectorized columnar scanning.

Enables real-time analytics and high-volume reporting concurrent with active transactional ingestion.

Validated against a 5,000+ test enterprise SQL compliance corpus, stored procedural packages, triggers, recursive CTEs, and hierarchical queries.

Full support for standard dynamic cluster performance views and catalog metadata dictionaries.

Tested across 50,000 deterministic cluster chaos seeds with random process kills, split-brain network drops, and storage latency injection with zero lost committed transactions.

Automated dead-node fencing and buffer pool lock recovery execute in under 500 milliseconds without interrupting surviving cluster instances.

Verified compatibility with enterprise Java frameworks (Spring Boot, Hibernate) and connection pools via drop-in JDBC driver shims supporting connection unwrapping and cursor casting.

Compiled C/C++, Pro*C, and Go binaries link directly to the native C-ABI client shared library without recompilation.