All the features Shannonbase have.
ShannonBase is a HTAP database provided by Shannon Data AI, which is an infra for big data & AI.
ShannonBase: The Next-Gen Database for AI—an infrastructure designed for big data and AI. As the MySQL of the AI era, ShannonBase extends MySQL with native embedding support, machine learning capabilities, a JavaScript engine, and a columnar storage engine. These enhancements empower ShannonBase to serve as a powerful data processing and Generative AI infrastructure.
Firstly, ShannonBase incorporates a columnar store, IMCS (In-Memory Column Store), named Rapid, to transform it into a MySQL HTAP (Hybrid Transactional/Analytical Processing) database. Transactional and analytical workloads are intelligently offloaded to either InnoDB or Rapid using a combination of cost-based and ML-based algorithms. Additionally, version linking is introduced in IMCS to support MVCC (Multi-Version Concurrency Control). Changes in InnoDB are automatically and synchronously propagated to Rapid by applying Redo logs.
Secondly, ShannonBase supports multimodal data types, including structured, semi-structured, and unstructured data, such as GIS, JSON, and Vector.
Thirdly, ShannonBase natively supports LightGBM or XGBoost (TBD), allowing users to perform training and prediction directly via stored procedures, such as ml_train, ml_predict_row, ml_model_import, etc.—eliminating the need for ETL (exporting data and importing trained ML models). Alternatively, pre-trained models can be imported into ShannonBase to save training time. Classification, Regression, Recommendation, Abnormal detection, etc. supported.
Fourthly, By leveraging embedding algorithms and vector data type, ShannonBase becomes a powerful ML/RAG tool for ML/AI data scientists. With Zero Data Movement, Native Performance Optimization, and Seamless SQL Integration, ShannonBase is easy to use, making it an essential hands-on tool for data scientists and ML/AI developers.
At last, ShannonBase Multilingual Engine Component. ShannonBase includes a lightweight JavaScript engine, JerryScript, allowing users to write stored procedures in either SQL or JavaScript.