The new features make it easier to use technologies like blockchain to prevent fraud, provide flexibility in working with JSON documents, and make it easier to train and evaluate machine learning algorithms by integrating them into databases. Effective use of data increases the competitiveness of companies. Next-generation data-driven applications are needed to get the most value from enterprise data. A unified database management system simplifies the creation of such applications by allowing you to choose the most appropriate data model, processing type, and development paradigm based on business requirements.

Here are some new features that enhance the convergence capabilities of Oracle Database 20c:

Oracle Machine Learning for Python (OML4Py): Oracle Machine Learning (OML) algorithms in Oracle Database accelerate predictive analytics problems solving. For this, advanced ML algorithms are used that can be applied directly to the data. Since the ML algorithms are placed with the data, there is no need to move the data from the database to somewhere else. Data scientists can also use Python to extend machine learning algorithms in a database. OML4Py AutoML: Machine learning mechanisms can be used even by non-experts. AutoML will recommend the most appropriate algorithms, automate feature selection, and perform hyperparameter tuning to greatly improve model accuracy. Native Persistent Memory Store: Database and redo log data can now be stored in local persistent memory (PMEM). Using simplified I/O algorithms, SQL can work directly with data stored in PMEM. This reduces the need for high-capacity buffer caches and enables faster data access for low-latency workloads, including high-frequency trading and mobile communications.

Automatic In-Memory Management: Oracle Database In-Memory option speeds up processing analytical and mixed online transactions, providing them with high performance, while allowing you to support real-time analytics and reporting. Automatic management of data placement in RAM (In-Memory) greatly simplifies the implementation of In-Memory devices. By automatically evaluating data usage algorithms, this feature determines which tables benefit most from being cached in the In-Memory Column Store and automatically caches them without any human intervention. Native Blockchain Tables: Oracle's native blockchain tables make it easy to use blockchain technology to detect and prevent fraud (anti-fraud). Oracle blockchain tables look like standard tables. They allow SQL inserts and the added strings are cryptographically chained. Optionally, strings can be signed to protect against fraud by means of an electronic signature. Oracle blockchain tables are easily integrated into applications. They can be used in transactions and queries along with other tables. In addition, insertion is very fast compared to a decentralized blockchain, since committing changes does not require consensus.

JSON Binary Data Type: For JSON documents stored in binary format in an Oracle database, the update is performed four times, and the scan – up to 10 times faster.

For more information about Oracle's new solutions, call +38 (044) 492-2929 or e-mail oracle@muk.ua