VectorWise

VectorWise
Actian corporation · CC BY-SA 4.0 · source
Name	VectorWise
Title	VectorWise
Developer	C-Tree International / Actian / VectorWise B.V.
Released	2009
Latest release	2015 (as Vectorwise 3; later integrated into Actian X)
Programming language	C++
Operating system	Linux, Solaris
Genre	Analytical database, columnar store, vectorized execution
License	Proprietary, commercial

VectorWise VectorWise is a high-performance columnar analytical database engine originally developed by researchers from the Centrum Wiskunde & Informatica and commercialized by VectorWise B.V., later acquired and integrated into products from Actian and other companies. It emphasizes vectorized execution, cache-conscious algorithms, and compressed columnar storage to accelerate complex analytical queries across large datasets. The project draws on research from academic systems and targets enterprise analytical workloads in data warehousing, business intelligence, and scientific analysis.

History

VectorWise originated from academic work at the Centrum Wiskunde & Informatica and spin-out research groups that included contributors affiliated with CWI and Dutch universities. The initial technology emerged in the mid-2000s, building on concepts from projects such as MonetDB and influenced by research at Vrije Universiteit Amsterdam and Eindhoven University of Technology. VectorWise B.V. was formed to commercialize the engine; early funding and partnerships included ties to Oracle Corporation researchers and European technology transfer offices. Over time the company partnered with and was acquired by firms like Actian Corporation and integrated into wider analytics offerings used by enterprises across Europe and North America. Major milestones include product releases around 2009–2015, public performance publications at conferences such as the International Conference on Very Large Data Bases and technology demonstrations at venues including Strata Conference and SIGMOD.

Architecture and Technology

The engine implements a columnar storage layout and a vectorized execution model inspired by research from projects like MonetDB and processor-aware designs discussed at VLDB workshops. It uses compressed columns with lightweight encodings to reduce I/O and exploit modern CPU caches; these techniques relate to studies presented at ICDE and EuroSys. The execution engine processes data in tight loops operating on ranges of values (vectors) to improve instruction-level parallelism, an approach comparable to work from HyPer and discussions in the SIGMOD Record. Query optimization leverages cost models informed by research from institutions such as ETH Zurich and the University of Waterloo. Storage and indexing features incorporate ideas similar to those from PostgreSQL extensions and column-store research at Brandeis University. Parallelism is achieved through multi-threaded operators coordinated by a runtime influenced by designs from Tanenbaum-style systems research groups. The stack integrates with standard interfaces and connectors used by enterprise tools from MicroStrategy, Tableau, and SAP BusinessObjects.

Performance and Benchmarking

VectorWise published benchmark results targeting analytical workloads, comparing favorably to disk-based and some in-memory systems in tests derived from the TPC-H and workload profiles discussed at SIGMOD and VLDB evaluations. Performance claims emphasized CPU efficiency, reduced memory bandwidth usage, and lower I/O compared with row-oriented engines like IBM Db2 and competitors such as Teradata and HP Vertica. Independent academic comparisons at institutions like MIT and University of California, Berkeley examined vectorized execution and column compression, situating VectorWise among designs including HyPer and MonetDB/X100. Real-world case studies published by enterprise partners illustrated improvements in query latency for OLAP tasks used by organizations including financial firms and telecom operators, with deployments showcased at industry events such as Gartner summits.

Use Cases and Applications

The engine was targeted at data warehousing, business intelligence, and ad hoc analytics for sectors including finance, telecommunications, and online services. Typical applications included interactive dashboarding for vendors like Tableau and Qlik, complex reporting for firms using SAP ecosystems, and analytics pipelines integrating with data ingestion tools from Informatica and Apache Kafka. Scientific and research groups adapted the engine for processing large observational datasets and time-series workloads, similar to deployments seen with Hadoop-adjacent analytics stacks and integration patterns common to Apache Spark ecosystems. Customers often paired the engine with OLAP front ends from Microsoft Power BI and ETL orchestration from vendors like Talend.

Licensing and Commercialization

VectorWise was commercialized under proprietary licensing terms by VectorWise B.V. before acquisition and further commercialization by Actian Corporation. Licensing models included per-core and enterprise subscriptions, professional services, and support agreements aligned with enterprise software practices used by firms such as IBM and Oracle Corporation. After integration into Actian products, the technology became part of bundled offerings and supported through commercial maintenance channels used by large enterprises and managed-service providers including system integrators and cloud partners. Academic and research editions were occasionally available under restricted terms for evaluation and collaboration with universities such as Delft University of Technology and University of Amsterdam.

Category:Database engines