WHITE PAPERS
White Papers From Bassirah Innovation & R&D.
In-depth technical papers on data platform design. Series 1 examines how modern storage engines are built and how to choose an engine class; Series 2 examines how to cache data fast without serving it wrong.
Free to read and download · No registration required
SERIES 1
Data Storage Engines
Six papers, from a survey of the whole field to a deep dive on each class of engine. All six can be read online or downloaded as PDF.
PAPER 1 · PUBLISHED
State-of-the-Art Data Storage Engines
A survey of five engine classes (write-optimized, analytical, hardware-native, vector and lakehouse) and the trade-offs that separate them.
PAPER 2 · PUBLISHED
LSM-Tree Engines in Production
Compaction, write amplification and tuning in RocksDB-family engines, with the open challenges and research directions.
PAPER 3 · PUBLISHED
Columnar and Vectorized Analytical Engines
How columnar layout, compression and vectorized execution use CPU caches and SIMD, with the open challenges and research directions.
PAPER 4 · PUBLISHED
Vector Storage Engines
ANN indexes, quantization, filtering and updates in vector search, with the open challenges and research directions.
PAPER 5 · PUBLISHED
Open Table Formats as Storage Layers
Apache Iceberg, Delta Lake and Apache Hudi: transactions, metadata, maintenance and interoperability, with the open challenges and research directions.
PAPER 6 · PUBLISHED
Distributed SQL Storage
Consensus replication, sharding, distributed transactions and PostgreSQL compatibility, with the open challenges and research directions.
SERIES 2
Database Caching
Six papers: a foundation on caching strategies, architectures, eviction and pitfalls, then five research frontiers in depth. All six can be read online or downloaded as PDF.
PAPER 1 · PUBLISHED
Database Caching Fundamentals
Read and write strategies, cache architecture types, eviction and invalidation policies, and the common pitfalls with their mitigations.
PAPER 2 · PUBLISHED
Learned Cache Replacement
Machine learning for eviction and admission decisions, from approximating the optimal policy to production-ready designs.
PAPER 3 · PUBLISHED
Semantic Caching for LLM Applications and Vector Databases
Reusing answers and computation by meaning, with the correctness, access control and evaluation questions it raises.
PAPER 4 · PUBLISHED
Non-Volatile Memory, NVMe SSDs and Hybrid Tiered Caching
Caches that span DRAM, CXL-attached memory and flash, and the placement and endurance trade-offs between tiers.
PAPER 5 · PUBLISHED
Consistency and Invalidation: Dependency-Graph and Automated Approaches
Tracking what cached data depends on, so that changes invalidate exactly what they should.
PAPER 6 · PUBLISHED
Multi-Level Caching for Geo-Distributed, Edge and IoT Data
Cache hierarchies from device to edge to region to core, and the consistency and placement choices between them.
EDITORIAL STANDARDS
How We Write White Papers
- Every paper names its author, its date and the scope it covers.
- Claims about systems cite published research or public documentation.
- No invented benchmarks, customer names or results.
- Product names identify technologies; they do not imply a business relationship unless we name a Strategic Technology Partner.
Want a topic covered?
Tell us which storage or caching decision your team is facing.
Related Pages
Turn the Survey Into a Decision
A short architecture workshop maps your workloads to engine classes and caching strategies, and plans how to validate the choice.