During Quest Forum Digital Event: Innovation Week, William Hardie, vice president of Oracle Database product management, outlined the newest features and updates to Oracle Database. His presentation included a reminder of recent updates—particularly 19c—and an explanation of the features in the preview release of Oracle Database 20c. Each of the latest features fits into Oracle’s new mission statement: To help people see data in new ways, discover insights, unlock endless possibilities.
Strategy Behind Oracle Database
The new strategy, based on the mission statement, is to provide customers with a converged or multi-model database for all different types of data, models, and use cases. One example is that Oracle Database enables customers to use the same database for a myriad of use cases. It is a converged database, as opposed to multiple single use databases, that enables a multitude of uses of SQL tools and application program interfaces across all types of data. From an administrative perspective, it is much easier to manage a converged database than multiple single-use databases. Even more significantly, he said, this format eliminates the need for data movement and integration across different databases. By applying common governance models across a single converged database, customers can simplify implementation of security and compliance governance and financial governance.
Oracle Database 19c, along with Exadata and machine learning workload optimizations for autonomous data warehouses and transaction processing services, is a core pillar of the Oracle Autonomous Database. Many compelling features were made available with 19c that can help with performance, scalability, reliability, security, and more. The key capabilities of 19c are a multitenant architecture, database in-memory, and sharding.
Licensing and Support for Oracle Database
In addition to the enhanced functionality of 19c, there are support and licensing updates that should also be taken into consideration. Oracle Database 19c represents the long-term support for 12c R2, 18c, and 19c. This offers customers a long support window with premier support to March 2023 and extended support to March 2026. The multitenant license change applies to all on-premise editions. Customers only require a separate multitenant license when they wish to consolidate four or more pluggable databases to a container database. Moreover, separate licenses are no longer required for customers wishing to use the spatial, graph, and machine learning capabilities of Oracle Database.
Oracle Database Ranked by Gartner
Gartner has identified four operational database management system (DBMS) use cases. The first use case is traditional transactions. DBMSs manage centralized transactions with fixed, stable schema, high-speed, high-volume data insert/update, and ACID properties. Hybrid/cloud deployment is increasingly important. The second use case is management of distributed variable data, which requires automated management of the data, often with multiple datatypes, structures, and schemas that vary. Consistency models have a high weight in this area. Third is event processing and data in motion-use-cases that involve events and observations captured at the edge. This may include processing at the edge and transmission of results to other stages of a business process. And the last use case is augmented transactions that can occur with analytics and affect processing through multiple states within the transaction scope in a single database, while maintaining low latency.
Oracle Database ranked first in all four use cases for the second consecutive year. This stands as testimony for the capabilities of converged architecture of 19c to handle multiple use cases in real life implementations.
Multitenant Container Architecture for Oracle Database 19c
Another factor in planning upgrades to 19c is in regard to adoption of multitenant container architecture. From Oracle Database 20c onward, non-CDB (container database) architecture will no longer be supported. However, with the recent multitenant licensing change permitting up to three pluggable databases, customers planning to update to 19c have the choice to update to container, non-container, or a combination of both architectures.
New Features in Oracle Database 20c
A preview release of 20c is currently available on Oracle Cloud Infrastructure Database Cloud Service. As expected with any release, 20c introduces several new features and enhancements. These include the following:
- AutoML in OML4Py (Oracle Machine Learning for Python)
- JSON Speed and Flexibility
- Native Blockchain Tables
- Graph Performance
- Automatic Zone Maps
- Persistent Memory Support
- In-Memory Improvements
- Sharding Enhancements
A preview release is available to anyone with Oracle Cloud credits for familiarization. The preview release is not designed for production deployment.
Oracle natively includes machine learning algorithms. It brings algorithms to the data, not data to algorithms. Moreover, the machine learning available includes deep learning, such as parallel performance of machine learning directly in a data warehouse or OLTP database, and fast model building and real-time scoring of new data. It extends in-database algorithms with Python and R, and easily deploys models via REST APIs.
Oracle will soon be introducing auto-machine learning for Python. Non-experts will be able to leverage machine learning. Some of the benefits include the following:
- Auto-Model Selection
- Identify the in-database algorithm that achieves highest model quality
- Find the best model faster than with exhaustive search
- Auto-Feature Selection
- Reduce number of features by identifying most predictive
- Improve performance and accuracy
- Auto-Tune Hyperparameters
- Significantly improve model accuracy
- Avoid manual or exhaustive search techniques
A common use case is a mix of analytical and operation workloads, achieved with Oracle’s unique two format in-memory row store and column store. Several enhancements deliver even faster performance. Spatial analytics are 10x faster, and text analytics are 3x faster than previous speeds. Vector joins are 5x-10x faster using SIMD (single instruction, multiple data) vector instructions. Hybrid scans are 10x faster, and self-management is auto in-memory populate and evict.
Another enhancement is the introduction of high-performance binary JSON datatypes in SQL and PL/SQL. These can now be created with a native JSON datatype. Additionally, SQL statements can be used for set operations against underlying JSON data.
A new use case for 20c is the extension of Oracle’s converge capabilities with native blockchain tables. This is enabled by a specialized table that allows normal SQL inserts and queries. It creates blockchain table ledger_of_trades, insert-only, and rows are cryptographically chained, so the chain is verifiable by participants. The main advantage is that customers are still able to use standard and SQL to access and join that information with existing relational tables.
Oracle Database 20c is bringing about a new generation of Exadata—X8M. This is 100Gb/sec RDMA over Converged Ethernet (RoCE), resulting in 2.5x faster networking. Database use of RDMA instead of I/O to access PMEM (persistant memory) bypasses the network and I/O software stacking, resulting in 10x lower latency. Customers will have the choice of deploying RoCE-based Exadata or continuing to use InfiniBand-based Exadata for compatibility.Customers have a few options for deploying Exadata:
- Private Cloud: Customer data center; purchased
- Cloud at Customer: Customer data center; pay-as-you-go or subscription
- Public Cloud: Oracle Cloud; pay-as-you-go or subscription
Exadata has also become a common component in many hybrid cloud implementations.
For developers, analysts, and data scientists, the Oracle Free Autonomous Database (ADB) service is available. The always-free micro instances offer up to two Always Free ADB instances per OCI tenancy, 20GB database storage per instance, and 10CPU per instance. This runs on the same Autonomous Database infrastructure with full functionality, full experience, and including APEX, REST Interface (ORDS), SQL Developer Web, and Machine Learning notebooks.
Another important update is Oracle Data Safe, which is free with all Oracle Cloud Databases. While autonomous security covers a lot, it cannot cover everything. However, once data is lowered into the database, there will be a unified database security control center. This will house a security configuration assessment, user risk assessment, user activity auditing, sensitive data discovery, and data masking.
In summary, Oracle Database 19c represents the long-term support for the 12c, 18c, and 19c family of releases. Customers can have up to three PDBs (pluggable databases) per CDB without requiring separate multitenant licenses. Spatial, graph, and machine learning is included with all editions.
As for Oracle Database 20c, a preview is available for the purpose of familiarization with OCI DBCS (Oracle Cloud Infrastructure Database Cloud Service). Lastly, Oracle Autonomous Database is free forever in “micro” instances of ADW (Autonomous Data Warehouse) and/or ATP (Autonomous Transaction Processing).