Reseller News

Oracle accelerates MySQL HeatWave queries with machine learning

MySQL Autopilot uses advanced machine learning techniques to accelerate HeatWave query performance and scalability in Oracle Cloud’s MySQL Database Service.

Taking aim at competitors including Amazon Aurora and Snowflake, Oracle has enhanced the MySQL HeatWave in-memory query accelerator in the Oracle Cloud’s MySQL Database Service by leveraging advanced machine learning.

But Oracle insists the improvements do not mean the MySQL Database Service is encroaching on its flagship Oracle Database.

The vendor has rolled out MySQL Autopilot, a component of HeatWave that uses advanced machine learning techniques to accelerate query performance and scalability. MySQL HeatWave works with the MySQL Database Service in Oracle Cloud Infrastructure (OCI) to accelerate performance for analytics and mixed OLTP (online transaction processing) and OLAP (online analytical processing) workloads.

Included with HeatWave at no extra charge, Autopilot automates aspects of achieving high query performance at scale including provisioning, data loading, query execution, and failure handling.

Advanced techniques are used to sample data, collect statistics on data and queries, and build machine learning models using Oracle AutoML to model memory usage, network load, and execution time. Autopilot increases the intelligence of the HeatWave query optimiser as more queries are executed, resulting in continually improving performance, Oracle said.

Key capabilities of MySQL Autopilot include auto-provisioning by predicting the number of HeatWave nodes needed to run a workload using adaptive sampling of data that requires analytics, alongside auto parallel loading to optimise load time and memory usage by predicting optimal parallelism for each table being loaded into HeatWave.

This is in addition to auto data placement through predicting the column on which tables should be partitioned in-memory to help achieve the best query performance, auto query time estimation prior to query execution and auto error recovery to provision new nodes and reload necessary data if one or more nodes is unresponsive.

Despite increasing the capabilities of its MySQL database line, Oracle insists there is no encroachment onto its own Oracle enterprise database services. The Oracle Database is for large-scale enterprise deployments while MySQL is for developers, cloud-native open source applications, and companies who have never had an on-premises environment, company officials insist.

For example, an Oracle Exadata cloud service could accommodate a 2.5-petabyte database, while HeatWave could accommodate 32 TB. A publicly traded, large financial services corporation would be using Oracle Autonomous Database or Exadata cloud service, but not typically MySQL with HeatWave, by virtue of the sheer size of the data set, Oracle said.

Along with introducing MySQL Autopilot, Oracle introduced MySQL Scale-Out Data Management to improve performance of reloading data into HeatWave by as much as 100x. HeatWave now can support a cluster size up to 64 nodes. It previously had been limited to 24 nodes. HeatWave also can process as much as 32 TB of data—up from 12 TB previously.