With DRAM sizes growing into the range of Terabytes, it is becoming less important to structure database algorithms around storage IO. In-Memory technology can provide orders of magnitude performance improvements for Databases, and In-Memory is therefore becoming an essential component of an enterprise data management strategy.
In this session we will first introduce Oracle's broad in-memory portfolio, which includes Oracle TimesTen, a dedicated In-Memory Database for microsecond OLTP and Oracle Database In-Memory, intended for processing billions of rows per second for for real-time analytics.
We will then provide a brief overview of the "dual-format" in memory architecture underlying the Database In-Memory Option which was released in 2014 along with Oracle Database 184.108.40.206, featuring an in memory column store maintained seamlessly along the existing row-oriented buffer cache. This architecture enables both orders of magnitude faster Analytics as well as significantly faster Enterprise hybrid OLTP workloads. We will also include some customer use cases to show how this technology is used in practice.
We will then describe the key enhancements to Database In-Memory in Oracle Database 12.2 - to further enhance performance, capacity, and ease of use. We will provide "under the hood" illustrations of the key Database In-Memory 12.2 features such as In-Memory Expressions, In-Memory Join Groups, In-Memory on Active Data Guard, In-Memory columnar on Exadata Flash, and Automatic In-Memory policies.
Having journeyed through the past and the present incarnation of the technology we will then discuss the future for In-Memory in Oracle including upcoming enhancements that will allow for even greater performance and ease of use of this essential new technology.