Deep Engine Brings Enterprise-Level Performance to MySQL Databases

By Frank Ohlhorst

Deep Engine is much more than a simple storage engine. The technology interacts at the kernel level of the database to empower its algorithms to use machine learning techniques, thus creating an environment that is adaptive and predictive for record retrieval.

Indexed Insertion Benchmark (iiBench) testing shows that pedestrian database functions, such as row insertion and reindexing, experience a 64-fold increase in performance when compared with innoDB, the native storage engine for MySQL.

Deep Engine brings performance improvements to standardized tasks, such as data ingestion, which is a common occurrence for building big data sets. Deep’s integrated heuristics and machine learning capabilities, along with support for multiple threads, increase parallel data importation by more than 20 times.

Deep Engine brings as much as a twofold performance increase to MySQL queries. Several technologies combine to make that possible, including multiple threads, faster decompression and predictive integrated analytics, which over time learn the shortest path to the data.

Deep Engine increases compression/decompression events by leveraging a multi-threaded kernel, as well as new compression algorithms that can increase both the level of compression and the speed at which compression and decompression events occur.

In the real world, database performance ultimately comes down to transaction processing speed. Here, Deep Engine’s use of machine learning heuristics brings self-tuning capabilities to MySQL, which when combined with improved compression and multiple threads along with advanced caching can bring significant transactional performance improvements to very large MySQL databases.

The Internet of things (IoT) will potentially bring billions of devices and associated logs into data analytics scenarios. Deep Engine enables MySQL to create rows faster and retrieve information more quickly than InnoDB, bringing the ability to deal with trillions of rows of data to support very large IoT analytical projects, such as worldwide sensor placements or scientific and meteorological research chores.

MySQL deployed on infrastructure-as-a-service (IaaS) platforms is traditionally limited by single thread processing. With Deep Engine, multiple threads allow MySQL to outperform enterprise-grade hosted solutions, such as Amazon RDS (Relational Database Service).

Deep Engine enables MySQL to leverage IBM and Zenith IaaS platforms using bare metal server implementations. The combination of hosted platforms along with bare metal servers allows Deep Engine to bring enterprise-level performance to MySQL, while still outperforming fully hosted Amazon RDS implementations.

Performance testing shows that Deep Engine brings increased value to MySQL databases by helping to reduce storage needs while increasing performance and offering higher scale. Deep Engine accomplishes all of that with an easy-to-install plug-in that can be used in conjunction with existing storage engines or used as a replacement for a storage engine.