Oracle Real Application Clusters (RAC) provides scalable performance, rich functionality, and high availability. It is the database chosen by thousands of organizations around the world.
Oracle TimesTen In-Memory Database (TimesTen) is an industry-leading in-memory relational database that provides breakthrough performance, low latency, and high availability to the most demanding applications.
In general, combining RAC with TimesTen—deployed as an application-tier database cache (TimesTen cache)—gives customers the best of both worlds. This especially applies in scenarios where long running queries, or a heavy Data Manipulation Language (DML) workload, running on a subset of Oracle data, depletes many of the Oracle RAC resources. Heavy DML workloads can lead to a lack of Oracle resources for other database operations and can have an impact on the overall database response times.
In this solution, we deploy a TimesTen cache in front of a RAC database to offload processing (both queries and DML) from the RAC database. We measure how the deployment impacts the overall database performance and the resource utilization of the RAC database. The study includes end-to-end Dell EMC enterprise IT infrastructure—PowerEdge servers, Dell EMC Networking switches, and Dell EMC Enterprise shared-data storage arrays.
The study also includes the use of Intel Optane persistent memory (PMem) to increase the capacity of the TimesTen cache, enabling far greater data volumes to be cached within a single server than in a DRAM-only configuration. We measured the performance impact of running TimesTen on PMem in Memory Mode (PMem-MM) when configured as a database cache in front of a RAC database.