MySQL HeatWave: Single system for OLTP and OLAP
Abstract
HeatWave (aka RAPID) is a high performance analytics engine for Oracle's MySQL Database Service. At core, it is a massively parallel and distributed in-memory query processing engine developed by Oracle Labs. The engine has been natively integrated with MySQL database to provide real time analytics on the data stored in MySQL storage engine (InnoDB), thus eliminating the need of expensive data movement between OLTP and OLAP engines. HeatWave provides 100x-1000x acceleration over MySQL and MySQL compatible databases like Aurora. It makes Oracle's MySQL Database Service 2.7x faster at one-third the cost of Amazon Redshift.
Some of the highlights of HeatWave engine:
- Native integration with MySQL database
- Native integration with Oracle Cloud Infrastructure
- Vectorized in-memory columnar processing
- Massively parallel architecture
- State of the art, hardware aware distributed query processing algorithms
- Intelligent scheduler to hide compute time with network communication
- Data driven automation
Short Bio
Nitin Kunal is director of software development at Oracle Labs Switzerland. Currently, he is leading the development effort of MySQL HeatWave engine (aka RAPID). He is one of the early members of Oracle Labs Switzerland; he joined the RAPID team in 2012 and since then he has been contributing to the project. Nitin led the research & development of RAPID query optimizer and helped in designing several query processing algorithms.
Prior to Oracle, he worked at two early stage startups. He led projects to design & implement a practical software transactional memory system and a high throughput storage engine optimized for SSDs for a Swiss startup, Staila Technologies (an ETH spinoff).
Nitin has 14 years of research & advanced development experience. His area of interest revolves around scalable, efficient databases and storage systems. He has filed over 10 patents and holds an engineering degree from National Institute of Technology, India.