By Matt Lord-Oracle on Oct 18, 2016
The development focus has been on improving overall performance and ease of use, combined with improved SQL support and the introduction of read optimized and fully replicated tables, all of which allow for better support of a variety of use cases.
Improved Performance: Improvements in index/range scan performance and internal thread handling will allow for better use of existing hardware and a better overall user experience.
Improved Ease-of-Use: Improved logging and SQL views offer greater insight into the current configuration, state, and general database operations, all of which helps to decrease maintenance and support costs, increasing your ability to resolve issues early enough that they don't become bigger user impacting problems.
Increased Capacity: With support for 128TB+ data set sizes per data node you can better utilize the much bigger commodity machines available today, which increases your ability to scale-up, in addition to scaling out.
Improved Read Scalability: You can mark all or individual tables as being read-optimized, thus effectively doubling your read throughput at the cost of slightly slower writes--a feature that is ideal for read-mostly tables. You can also declare some tables as being "global" or fully replicated--a feature that is ideal for lookup or fact tables which are relatively small, static, and often joined against (e.g. City, State, Product Type, etc.). These two features can greatly improve overall data locality--meaning that the data used for more complex queries (e.g. scans and joins) is local to a given data node, allowing you to avoid costly overhead associated with cross-node/cross-shard network communication and transaction management.
Download the source and binaries today and get started now!