Oracle today announced general availability of Release 2 of it 11g database with grid provisioning and database consolidation said to easier manage scale and cost. The release also includes in-memory caching for certain data warehouse and decision support applications.
While most customers are running the 10gR2 or 11gR1 version of Oracle’s database, Andy Mendelsohn, SVP of database server technologies at Oracle says new functionality in 11gR2 are important because they represent where most customers will be migrating. 
Grid technology goes back seven years to Oracle’s 9i release, but Mendelsohn says the original and new versions of 11g that include Oracle Real Application Clusters (RAC) allow more flexible plug-and-play scalability and the ability to add or reduce nodes in short order. 
“With 11gR2 we spent a lot of time on provisioning clusters and automating that,” Mendelsohn said at a press conference. A cost benefit of 11g is the ability to better leverage low-cost servers to build out and scale infrastructure, he says. “If you’re running your customer workload today and it has 10,000 transactions per second, you can just provision enough servers in your cluster to deal with that and if the workload grows over time you can just roll in servers and grow your environment.”
New to R2 is RAC One Node, a failover capability that allows smaller databases consolidated into a single server to remain running in the event of a server failure. “A lot of customers are interested in taking smaller databases and consolidating them into one server environment using RAC,” Mendelsohn said. “But in a lot of cases these database servers don’t need the scalability of RAC, they just need the availability.” RAC One Node offers most of the availability benefits of RAC at a lower cost for customers that don’t need scalability. 
“11gR2 focuses on making databases run quicker, have better availability and scale using lower cost hardware,” says Forrester Research analyst Noel Yuhanna. Rac One Node, he says, “is a compelling business case … for those that support smaller or less mission critical apps that need better availability.” 
For data warehouse and other decision support customers, Oracle also released a technology called in-memory parallel query. This product is designed for relatively small tables of under one terabyte of data, and allows caching of aggregate memory capacity across all the servers in a cluster to cache the entire DB in the memory of the servers in the cluster. By bringing the data into main memory and running queries from there, Mendelsohn says customers will see a significant performance “bump” through eliminating the usual I/O processes. For typically larger databases, Mendelsohn reiterated the importance of Oracle’s Exadata and other database machines. 
“With low memory prices, running large and complex queries in-memory makes a very appealing case for business,” Yuhanna. “In-memory parallel execution across nodes in a cluster removes the need for disk I/O for large tables, which typically has been the biggest bottleneck.”